functional literacy essay

MINDFUL TECHNICS

Ruminations on our relationship with technology

What is functional literacy, and why does our high-tech society need it?

Literacy vs. functional literacy.

Literacy is usually defined as the ability to read, write, and do math. But what do we mean by reading? People may be able to read the words in a document, but can they grasp its arguments or analogies? How about writing? Students may be able to write or type sentences, but can they express complete, cogent thoughts? Then there’s math. Workers may be able to count, add, and subtract; but do they know how to analyze graphs, interpret statistics, or plan budgets? These are skills needed to make decisions and function well in daily life.

Clearly, reading or writing words and numbers is not enough to assess literacy. People need to be able to understand and use those words and numbers for practical purposes, such as deliberating ideas and solving problems. This broader definition of literacy is called functional literacy.

What is functional literacy?

Functional literacy refers to the practical skill set needed to read, write, and do math for real-life purposes, so people can function effectively in their community. This definition comes from the United National Educational, Scientific and Cultural Organization, or UNESCO , an international agency that researches education, science, and communication.

In the U.S., functional literacy is assessed by the National Assessment of Adult Literacy, or NAAL . NAAL is put together every few years by the National Center for Education Statistics, a research division of the Institute of Education Sciences in the Department of Education. This research classifies literacy according to four levels :

1) Below Basic Literacy : reading and writing words and numbers in very simple documents.

Examples : locating easily identifiable information on a chart; signing a form; adding a dollar amount to a deposit slip.

2) Basic Literacy : performing simple skills to understand short texts.

Examples : reading a pamphlet; using a TV guide; comparing ticket prices.

3) Intermediate Literacy : performing challenging skills to understand long texts.

Examples : looking up information in a reference book; summarizing a long article; placing an order and calculating the cost.

4) Proficient Literacy : performing creative and critical thinking skills to understand dense or complex texts.

Examples : comparing viewpoints in editorials; interpreting statistical graphs; measuring and calculating the costs of food items per ounce.

Notice that as we advance from Below Basic to Basic, Intermediate, and Proficient, we go from simple tasks to more and more complex skills.

Functional literacy = functional society

So, how functionally literate is the U.S.? According to NAAL’s results , 87% of U.S adults can’t perform at the Proficient level; and 14% of the population—1 out of every 7 people—are functionally illiterate, or perform Below Basic level.

Many leaders, of course, find those figures alarming. For instance, an undeniable point made by high-profile businesspeople is that the 21st-century economy demands highly skilled workers with knowledge in STEM (Science, Technology, Engineering, Math)—or, better yet, STEAM (same subjects plus Arts).

Functional literacy, however, is about much more than getting a job. It’s also about preserving skills that our society needs to function. If we don’t have individuals with mathematical knowledge, reading and writing skills, or analytical abilities, we won’t have functioning businesses, governments, or communities. In other words, we need functional literacy to have a functional society .

Consequently, functional illiteracy will mean living in a dysfunctional society, as satirized by the movie Idiocracy .

Hence, with these definitions and stats in mind, let’s look at some examples of functional literacy that prevent society from becoming dysfunctional.

Kinds of functional literacy

Media Literacy is the ability to access, evaluate, and create messages through different types of media. Its purpose is to turn people from mindless consumers into thoughtful citizens who aren’t susceptible to propaganda or misleading ads.

Religious Literacy is the ability to interpret religious scriptures and communicate with different faiths. Religious literacy is important for combating fundamentalism (e.g., religious fanaticism) and prejudice (e.g., xenophobia).

Financial Literacy is the ability to manage finances and make decisions about money. Whether you’re a consumer, a business owner, or a voter, understanding financial budgets, interest rates, and savings is an essential life skill.

Computer Literacy is the ability to use computers. This skill set can range from basic competency (i.e., using applications like email and Microsoft Office) to advanced knowledge (e.g., programming and computer science).

Legal Literacy is the ability to comprehend laws so you are able to follow policies and legal procedures.

Scientific Literacy does not necessarily mean memorizing facts. Rather, it’s knowing how to conduct experiments and identify evidence that supports or contradicts preconceived beliefs or hypotheses.

Health Literacy is the ability to understand healthcare information, particularly for making medical decisions or lifestyle choices about nutrition, exercise, sleep, and other factors that affect physical and mental well-being.

Civic Literacy (a.k.a. Civics) is awareness of how government works as well as your rights and responsibilities as a citizen and voter.

Functional literacy and technology

Now, because our society is so high-tech, it’s important to emphasize two key points about functional literacy—namely, by connecting it to technology.

First, functional literacy is primarily about skills or applied knowledge .

It’s only secondarily about facts or subject-matter knowledge. For example, scientific literacy doesn’t mean you’ve memorized the intricacies of how quantum computers work. Instead, it means you know how to ask questions and apply methods of verification or falsification that make such technologies possible.

Second, functional literacy keeps our high-tech society functioning .

The kinds of functional literacy mentioned are relative to today’s highly complex, technological society, but that doesn’t make them any less valid. For instance, the ancient Greeks got by fine without media literacy or computer literacy, but that’s because they didn’t have digital networks. If these examples of functional literacy are unique to our high-tech society, we evidently need them.

Know of any other kinds of functional literacy I forgot to mention?  Leave a comment below, or check out more Educational Topics  on this site.

Share this:

• Click to share on Twitter (Opens in new window)
• Click to share on Facebook (Opens in new window)
• Click to share on LinkedIn (Opens in new window)
• Click to email a link to a friend (Opens in new window)

Leave a Comment Cancel reply

Save my name, email, and website in this browser for the next time I comment.

Notify me of follow-up comments by email.

Notify me of new posts by email.

Functional Literacy

Cite this chapter.

• Ludo Verhoeven 3  

Part of the book series: Encyclopedia of Language and Education ((LANG,volume 2))

554 Accesses

3 Citations

Concern about levels of literacy has been growing in many parts of the world in recent decades and is reflected in the greater general awareness of the numbers of illiterates and the consequences of illiteracy for the individual. Increased interest in this area has served as a catalyst for research from which ‘functional literacy’ has emerged as an important theme. This model of literacy recognizes personal and social needs and has played a key role in furthering our understanding of the distribution, consequences and causes of illiteracy.

• Literacy Skill
• John Benjamin
• Literacy Practice
• Communicative Competence
• Educational Implication

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

• Available as PDF
• Read on any device
• Instant download
• Own it forever
• Compact, lightweight edition
• Dispatched in 3 to 5 business days
• Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Institutional subscriptions

Unable to display preview.  Download preview PDF.

Auerbach, E.: 1992, ‘Literacy and ideology’, Annual Review of Applied Linguistics 20(12), 71–85.

Google Scholar  

Bachman, L.F.: 1990, Fundamental Considerations in Language Testing , Oxford University Press, New York NY.

Barton, D. & Ivanič, R. (eds.): 1991, Writing in the Community , Sage, London.

Canale, M. & Swain, M.: 1980, ‘Theoretical bases of communicative approaches to second language testing and teaching’, Applied Linguistics 1, 1–47.

Article   Google Scholar  

Coulmas, F.: 1991, The Writing Systems of the World , Blackwell, Oxford.

Cummins, J.: 1984, ‘Wanted: A theoretical framework for relating language proficiency to academic achievement among bilingual students’, in C. Rivera (ed.), Language Proficiency and Academic Achievement , Multilingual Matters, Clevedon, 1984, 2–19.

Fingeret, A.: 1983, ‘Social network: A new perspective of independence and illiterate adults’, Adult Education Quarterly 33, 133–146.

Fishman, J.A.: 1980, ‘Ethnocultural dimensions in the acquisition and retention of biliteracy’, Journal of Basic Writing 3, 48–61.

Gray, W.S.: 1956, The Teaching of Reading and Writing , Scott Foreman, Chicago.

Guthrie, J. & Kirsch, I.: 1983, ‘What is literacy in the United States? Reading competencies and practices’, Technical Report 5, International Reading Association, Newark.

Hamilton, M.: 1987, Literacy, Numeracy and Adults: Evidence from the National Child Development Study , ALB SU, London.

Harris, R.: 1986, The Origin of Writing , Duckworth, London.

Hymes, D.: 1971, On Communicative Competence , Philadelphia University Press, Philadelphia.

Kirsch, I.S. & Jungebluth, A.: 1986, Literacy: Profiles of America’s Young Adults , NAEP, Princeton, NJ.

Levine, K.: 1986, The Social Context of Literacy , Routledge & Kegan Paul, London.

Levine, K.: 1994, ‘Functional literacy in a changing world’, in L. Verhoeven (ed.), Functional Literacy. Theoretical issues and educational implications , John Benjamins, Amsterdam/Philadelphia, 1994, 113–132.

Lind, A.: 1990, Mobilizing Women for Literacy , Unesco Bureau of Education, Paris.

Luke, A.: 1988, Literacy Textbooks and Ideology , Falmer Press, London.

Olson, D.R.: 1977, ‘Oral and written language and the cognitive processes of children’, Journal of Communication , 10–26.

Olson, D.R. & Torrance, N.: 1991, Orality and Literacy , Cambridge University Press, Cambridge.

Rockhill, K.: 1987, ‘Gender, language and the politics of literacy’, British Journal of Sociology of Education 2, 153–167.

Scinto, L.F.M.: 1986, Written Language and Psychological Development , Academic Press, London.

Street, B.V. (ed.): 1984, Literacy in Theory and Practice , Cambridge University Press, Cambridge.

Street, B.V.: 1994, ‘Cross-cultural approaches to literacy’, in L. Verhoeven (ed.), Functional Literacy. Theoretical Issues and Educational Implications , John Benjamins, Amsterdam/Philadelphia, 1994, 95–112.

Stubbs, M.: 1980, Language and Literacy: The Sociolinguistics of Reading and Writing , Routledge & Kegan Paul, London.

Stuckey, E.:1991, The Violence of Literacy , Portsmouth, NH: Boynton/Cook, Heinemann.

Tannen, D.: 1982, ‘Oral and written strategies in spoken and written narratives’, Language 58, 1–21.

Unesco: 1953, World Illiteracy at Mid-Century , Unesco, Paris.

Venezky, R.L.: 1990, ‘Definitions of literacy’, in R.L. Venezky, D.A. Wagner & B.S. Ciliberti (eds.), Toward Defining Literacy , International Reading Association, Newark De, 1990, 2–16.

Verhoeven, L.: 1994a, ‘Modeling and promoting functional literacy’, in L. Verhoeven (ed.), Functional Literacy. Theoretical Issues and Educational Implications , John Benjamins, Amsterdam/Philadelphia, 1994, 3–34.

Verhoeven, L.: 1994b, ‘Linguistic diversity and the acquisition of literacy’, in L. Verhoeven (ed.), Functional Literacy. Theoretical Issues and Educational Implications , John Benjamins, Amsterdam/Philadelphia, 1994, 199–220.

Download references

Author information

Authors and affiliations.

University of Nijmegen, the Netherlands

Ludo Verhoeven

You can also search for this author in PubMed   Google Scholar

Editor information

Editors and affiliations.

University of Reading, England

Viv Edwards

The Ontario Institute for Studies in Education, University of Toronto, Canada

David Corson

Rights and permissions

Reprints and permissions

Copyright information

© 1997 Springer Science+Business Media Dordrecht

About this chapter

Verhoeven, L. (1997). Functional Literacy. In: Edwards, V., Corson, D. (eds) Encyclopedia of Language and Education. Encyclopedia of Language and Education, vol 2. Springer, Dordrecht. https://doi.org/10.1007/978-94-011-4540-4_14

Download citation

DOI : https://doi.org/10.1007/978-94-011-4540-4_14

Publisher Name : Springer, Dordrecht

Print ISBN : 978-0-7923-4929-7

Online ISBN : 978-94-011-4540-4

eBook Packages : Springer Book Archive

Share this chapter

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Publish with us

Policies and ethics

• Find a journal
• Track your research

• EN / EL / ES / PT / ZH

Chapter 6: Functional Literacy Pedagogy

Video mini-lectures.

6.1 Functional Literacy Pedagogy: An Overview

6.2 The Content Focus of Functional Literacy Pedagogy

6.3 Critical Views of Functional Literacy

Supporting Material

Dimension 1: The Contents of Literacy Knowledge—Learning the Genres of School Success and Social Power

Lock on Functional Grammar

Macken et. al on Reports

Grammar: Making Meaning in Writing

Learning to Mean

Macken et al. on Genre Theory

van Haren et al., Using Field, Mode and Tenor to Understand Hamlet

Snyder, The stories that divide us: Media (mis)representations of literacy education

Dimension 2: The Organisation of Literacy Curriculum—Reading Genre Models and Writing within Generic Frameworks

Martin and Rothery on the Grammar of Reports

Schleppegrell, The role of metalanguage in supporting academic language development

Callaghan, Knapp and Noble on The Genre Curriculum Cycle or ‘Wheel’

Dimension 3: Learners Doing Literacy—Presentation of Genre Scaffolds and Independent Construction in these Scaffolds

Macken et al. on Story Genres

Dimension 4: The Social Relationships of Literacy Learning—Learning Powerful Text Forms for Educational Success and Social Access

Christie on Learning to Mean in Writing

Rossbridge and Rushton, The critical conversation about text: Joint construction

Teaching learning cycle – Ross on Youtube

Humphrey, Creating effective persuasive texts within and beyond schooling

Kalantzis and Cope, Debating Functional Literacy

• The Open University
• Guest user / Sign out
• Study with The Open University

My OpenLearn Profile

Personalise your OpenLearn profile, save your favourite content and get recognition for your learning

About this free course

Become an ou student, download this course, share this free course.

Start this free course now. Just create an account and sign in. Enrol and complete the course for a free statement of participation or digital badge if available.

6 How is critical literacy different from functional literacy?

In the previous section we gave some explanations about the specific ways of understanding functional literacy. In this section we are going to ask you to engage as learners in the process of clarifying and developing your own understanding of critical literacy. As well as asking you to actively analyse the concept and relevant texts, we will be asking you to reflect on texts related to critical literacy.

1. Using a search engine designed to find academic papers, such as Google Scholar [ Tip: hold Ctrl and click a link to open it in a new tab. ( Hide tip ) ] , what information do you find if you search for ‘critical literacy’ as a keyword? Think about the following questions:

• How does critical literacy differ from functional literacy?
• How is the concept of critical literacy different from ‘being critical’ and concepts such as ‘critical thinking’?
• How is critical literacy related to critical theory?

2. Choose two key readings that you would argue are related to critical literacy. We would like you to be clear about why you have chosen these particular readings as representative of critical literacy. Do they interrogate social issues such as poverty, equity, inequality, disability etc.? Do they critique power relations in institutions (such as the school), adult education, and the family?

3. Using a search engine designed to find academic papers, such as Google Scholar , search for both articles listed below.

• Luke, A. (2012) ‘Critical literacy: foundational notes’, Theory Into Practice , vol. 51, no. 1, pp. 4–11.
• Perry, K. (2012) ‘What is literacy? – a critical overview of sociocultural perspectives’, Journal of Language and Literacy Education , vol. 8, no. 1, pp. 50–71.

As you read, consider the following questions:

• What are your understandings of ‘critical literacy’ and ‘functional literacy’?
• Can you see when you might use either approach in your own practice?
• Can you can think of ways to critique both of these concepts in relation to your own engagement with literacy and/or your own experiences of working with learners?

Through engaging with Activity 5 as well as the readings in Activity 6, you may have noted that the concept of functional literacy positions literacy as a set of tangible skills. Functional literacy also usually implies an emphasis on the cognitive, individually based, basic skills of reading or decoding print. Taking a functional literacy perspective towards literacy acquisition would, therefore, emphasise the individual and their basic literacy skills and rather than highlighting the ways in which individuals can use literacy to engage in complex and ever-changing social practices.

Critical literacy, however, has a different emphasis. As Luke (2012) points out, critical literacy has its origins in critical theory. It is thus underpinned by a notion that literacy evolves from entering into dialogue and engagement with cultural and social problems. From this viewpoint, functional literacy can be seen as a form of ‘banking education’ (Freire, 1970), with a passive learner, rather than stressing the need for a learner to be actively engaged in learner–teacher dialogue. Where functional literacy focuses on being able to read the words, critical literacy implies a need to interrogate the text for implicit meanings and agendas.

From this perspective, those who take a critical literacy stance often argue that functional literacy treats reading and writing text as a neutral technique, where nuances of meaning are not stressed. From a functional literacy perspective, surrounding agendas, cultural viewpoints and the influence of different agendas and groups would have very little, if any, influence on students’ abilities to interpret and respond to printed text.

Luke (2012) refers to Paulo Freire and his literacy programmes as an example of a critical approach to literacy. He also notes that this perspective arises from critical literacy and its links to philosophy, sociology and Marxist theory in particular. This contrasts with approaches that have been influenced by psychology, particularly behaviourism and psychometrics. Two specific examples arising from these two very different approaches to literacy are Freirian literacy programmes and dyslexia assessment and related interventions.

Perry (2012) explores ways in which critical and social literacy practices, like Freire’s programmes, attempt to focus on responding to the wider structural inequalities. You may have noted that this contrasts with the way in which functional literacy tends de-emphasise a relationship between literacy acquisition and wider power relationships or social inequalities. Programmes designed to develop functional literacy place the emphasis on the individual decoding text rather than on the wider social and cultural environment.

In her paper, Perry critiques the functional literacy approach. She specifically highlights how it excludes many ways in which people engage with print in everyday contexts and the contextual nature of communication. You may want to reflect on how you view this issue, and on the way in which you think that viewing learners as being deficient in technical skills may or may not be problematic for inclusive practice. For example, would you agree with Perry that literacy development needs to be seen within a wider context, of which formal education is just one aspect? Or would you support a counterargument that literacy needs to address barriers within the individual in order to produce proficient decoders and writers who can gain a functional level in literacy? It is often argued that gaining basic functional literacy skills enables learners to have greater opportunities and access to resources in society. The approach you would develop for learners who struggle with literacy could depend on your answer to these questions. The stance you take in relation to these questions would also influence what you would consider to be inclusive practice for these learners.

This site belongs to UNESCO's International Institute for Educational Planning

IIEP Learning Portal

Search form

Functional literacy.

A person is functionally literate/illiterate who can/cannot engage in all those activities in which literacy is required for effective functioning of his or her group and community and also for enabling him or her to continue to use reading, writing and calculation for his or her own and the community’s development. (Definition originally approved in 1978 at UNESCO’s General Conference, and still in use today.)

UNESCO. Education for all: literacy for life; EFA global monitoring report, 2006 . Paris: UNESCO, 2006.

Although the Work Oriented Adult Literacy Programme did not achieve the success expected – it was, after all, an experimental and pioneering effort – it did help to establish more firmly the principle of linking literacy instruction with uses directly relevant to the adult learner. Indeed, the phrase ‘functional literacy’ remains very much in use. The question for the educational planner – whether a government official or a member of a voluntary body – is what uses or functions will be relevant to which groups of adult learners? (Oxenham, 2008: 62).

Oxenham, John. Effective literacy programmes: options for policy-makers . Fundamentals of Educational Planning 91. Paris: UNESCO-IIEP, 2008.

• Frontiers in Psychology
• Cognitive Science
• Research Topics

Adult Functional (Il)Literacy: A Psychological Perspective

Total Downloads

Total Views and Downloads

About this Research Topic

In today’s knowledge societies, literacy skills are key to promote success in many aspects of people’s lives, including employment and personal development. Yet, many adults are functionally illiterate. This Research Topic aims to bring together experts in the field of adult literacy to share their findings ...

Keywords : reading, writing, functional illiteracy, functional literacy, reading assessment

Important Note : All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.

Topic Editors

Topic coordinators, recent articles, submission deadlines, participating journals.

Manuscripts can be submitted to this Research Topic via the following journals:

total views

• Demographics

No records found

total views article views downloads topic views

Top countries

Top referring sites, about frontiers research topics.

With their unique mixes of varied contributions from Original Research to Review Articles, Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author.

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• Front Psychol

A Review about Functional Illiteracy: Definition, Cognitive, Linguistic, and Numerical Aspects

Réka vágvölgyi.

1 LEAD Graduate School & Research Network, University of Tuebingen, Tuebingen, Germany

Andra Coldea

2 School of Psychology, University of Glasgow, Glasgow, Scotland

Thomas Dresler

3 Department of Psychiatry and Psychotherapy, University of Tuebingen, Tuebingen, Germany

Josef Schrader

4 German Institute for Adult Education – Leibniz Centre for Lifelong Learning, Bonn, Germany

Hans-Christoph Nuerk

5 Department of Psychology, University of Tuebingen, Tuebingen, Germany

6 Knowledge Media Research Center – Leibniz Institut für Wissensmedien, Tuebingen, Germany

Associated Data

Formally, availability of education for children has increased around the world over the last decades. However, despite having a successful formal education career, adults can become functional illiterates. Functional illiteracy means that a person cannot use reading, writing, and calculation skills for his/her own and the community’s development. Functional illiteracy has considerable negative effects not only on personal development, but also in economic and social terms. Although functional illiteracy has been highly publicized in mass media in the recent years, there is limited scientific knowledge about the people termed functional illiterates; definition, assessment, and differential diagnoses with respect to related numerical and linguistic impairments are rarely studied and controversial. The first goal of our review is to give a comprehensive overview of the research on functional illiteracy by describing gaps in knowledge within the field and to outline and address the basic questions concerning who can be considered as functional illiterates: (1) Do they possess basic skills? (2) In which abilities do they have the largest deficits? (3) Are numerical and linguistic deficits related? (4) What is the fundamental reason for their difficulties? (5) Are there main differences between functional illiterates, illiterates, and dyslexics? We will see that despite partial evidence, there is still much research needed to answer these questions. Secondly, we emphasize the timeliness for a new and more precise definition that results in uniform sampling, better diagnosis, conclusion, and intervention. We propose the following working definition as the result of the review: functional illiteracy is the incapability to understand complex texts despite adequate schooling, age, language skills, elementary reading skills, and IQ. These inabilities must also not be fully explained by sensory, domain-general cognitive, neurological or mental disorders. In sum, we suggest that functional illiteracy must be more thoroughly understood and assessed from a theoretical, empirical, and diagnostic perspective.

On the Importance of Literacy

About literacy.

According to the recent literacy rate, 85% of the adult population in the world is literate, and therefore worldwide about 757 million people are illiterate ( UNESCO, 2015 ). Large-scale assessments measuring literacy skills indicate that in developing countries, illiteracy is more prevalent, while in developed countries, functional illiteracy is more prevalent ( Bhola, 1995 , p. 18). According to the Organization for Economic Co-operation and Development (OECD), literacy is defined as follows:

“ Literacy is defined as the ability to understand, evaluate, use, and engage with written texts to participate in society, achieve one’s goals, and develop one’s knowledge and potential ( OECD, 2013 , p. 59).” More detailed, find other institutions, e.g., UNESCO.

Literacy and basic knowledge cannot be clearly separated from each other. Even though the term “literacy” is a part of basic knowledge, it is a precondition as well as an outcome of basic knowledge. Literacy may refer to the ability to read and write, but also to application-oriented basic knowledge that develops during the whole lifetime, not only during school years ( Nickel, 2007 ).

Formal literacy has increased over the last decades. For instance, while in sub-Saharan Africa there are still 29.8 million children who do not have access to education, this number represents a one-quarter decrease from 2000. In contrast, in Europe “only” 0.7 million of children had never attended school in 2011 ( UNESCO, 2013 ). However, despite improvements in formal literacy, many people still have problems understanding formal texts. On the one hand, this is a problem because in today’s society, functioning literacy plays a significant role. It appears in every aspect of daily life, e.g., opening bank accounts, reading ingredients of food products, understanding medication or technical instructions, signing contracts, etc. ( Cree et al., 2012 ). On the other hand, this leads to fewer educational and employment opportunities and hinders living a successful life.

Possessing literacy has many benefits for individuals, families, communities, and nations. The improvement in literacy levels has beneficial effects on individual (e.g., self-esteem), political (e.g., democratic values), cultural (e.g., cultural openness), social (e.g., children’s health), and economic (e.g., individual income) levels ( UNESCO, 2006 ). On the other hand, functioning in a society without literacy becomes more difficult: those who cannot acquire basic literacy skills have fewer opportunities in every area of life ( Cree et al., 2012 ).

About (Functional) Illiteracy

So far, we have talked about literacy. However, many people do not achieve literacy because of inadequate schooling or even despite adequate schooling. On 1949, the United Nations Educational, Scientific, and Cultural Organization (UNESCO) set the generalized functionality of literacy. The acquisition of reading and writing was regarded as basic rights: people should be enabled to become functionally literate in their own culture ( Bhola, 1995 ). A need for a standard and a workable definition materialized to differentiate between literates and non-literates (illiterates) and also to distinguish various levels in between. The result of the demand was realized at the General Conference of the UNESCO in 1978:

“A person is literate who can with understanding both read and write a short simple statement on his everyday life. A person is illiterate who cannot with understanding both read and write a short simple statement on his everyday life. A person is functionally literate who can engage in all those activities in which literacy is required for effective functioning of his group and community and also for enabling him to continue to use reading, writing, and calculation for his own and the community’s development. A person is functionally illiterate who cannot engage in all those activities in which literacy is required for effective functioning of his group and community and also for enabling him to continue to use reading, writing, and calculation for his own and the community’s development ( UNESCO, 1978 , p.183).”

The difference between literate and illiterate people is explicit here: illiterates had never attended school and are unable to read or write even single words while literates can ( Reis and Castro-Caldas, 1997 ).

In contrast with literacy and illiteracy, the difference between functional illiteracy, literacy and illiteracy is not obvious enough. Functionality, which is the essence of the difference between these terms, was never operationally defined. Recently, the number of functional illiterates in Europe was estimated to be about 80 million, their proportion is lowest in Sweden with 8% and highest in Portugal with 40% (e.g., in Eme, 2011 ; Grotlüschen and Riekmann, 2011a ). However, the frequently referred original International Adult Literacy Survey (IALS) report does not imply functional illiteracy ( OECD and Statistics Canada, 2000 ). Different definitions and different diagnostic assessment standards can lead to fundamentally different epidemiological estimations, so any estimations of functional illiteracy rates may be unreliable.

Diagnostics of Functional Illiteracy: Different Approaches

As there is no explicit assessment for functional illiteracy, researchers had to find other techniques to assess the number of functional illiterates or to identify functional illiterates for experimental studies.

The UNESCO, the OECD and the IEA (International Association for the Evaluation of Educational Achievement) measure literacy and other key knowledge skills of children, young adults, and adults a large-scale, international assessment about strengths and weaknesses in different countries. Research such as the IALS and the Adult Literacy and Life Skills Survey (ALL) build on each other ( Thorn, 2009 ; UNESCO, 2009 ). These kinds of international tests generally measure literacy and numeracy skills in various ways, including mapping the whole literacy spectrum and grouping the performance and the abilities into discrete levels. The international, supranational and national political actors are first interested in large-scale assessments, not in individual diagnostics. Against this background, it is understandable (but nevertheless at least unfortunate) that the diagnostic materials lack test criteria (reliability, construct validity, criterion validity), which are demanded in standard individual diagnostic tests.

The IALS, the ALL, and the PIAAC (Survey of Adult Skills) all contain prose and document literacy tasks that purport to understand and use information from different text formats. The quantitative literacy and numeracy tasks measure arithmetic abilities in all three assessments, but problem solving tasks are only included in the ALL and in the PIAAC study ( Table ​ Table1 1 ). However, these studies usually analyze literacy in a theoretical way and give no practical diagnostic advice regarding the assessment of functional illiteracy. It can be only a conclusion from the result of the lowest achievement level.

Summary of international assessments.

A common way of diagnosing functional illiterates is based on the years of schooling . However, the standard seems to vary among cultures. In the USA, 12 years of schooling marks the limit of functional literacy ( Bhola, 1995 ), while in Latin America, only 7 years of effective schooling is sufficient to exceed the level of functional illiteracy ( Infante, 2000 In. Martinez and Fernandez, 2010 ). In the European Union, the compulsory education is between 9 and 13 years, so children can leave school between age 14 and 18 ( European Commission, 2014/2015 ). Therefore, we cannot consider compulsory education as the only diagnostic attribute of functional illiteracy.

Another common diagnostic practice is using grade-equivalent scores and reading-level match designs . This concept is concrete, easy to understand, and it does not require a new specific test because the researchers use general standardized assessments. This method is mostly used when low literate adults are assessed and compared with primary school children ( Greenberg et al., 1997 ; Thompkins and Binder, 2003 ; Greenberg, 2007 ; Rüsseler et al., 2011 ; Grosche, 2012 ; Eme et al., 2014 ). Comparing children who have already acquired basic reading, writing, and mathematical skills with low literate (functional illiterate) adults could answer a few questions. The developmental differences between children and adults can cause problems in interpreting the results of such studies.

The German Federal Ministry of Education and Research (Bundesministerium für Bildung und Forschung) organized a national strategy to reduce the number of people who do not acquire basic literacy skills. To explore the problem, Grotlüschen and Riekmann constructed a representative household survey, the Level One Study (leo.). They specified five alpha-levels a priori in the lea. (Literalitätsentwicklung von Arbeitskräften; Grotlüschen et al., 2011 ), however, the validity of these five alpha-levels (even their eventual borders) has not yet been – to the best of our knowledge – never systematically evaluated in a diagnostic manner. Nevertheless, these five levels were applied to the leo. The lea. was constructed to measure employees’ different competence domains, including literacy and aimed to support individual teaching and development instead of comparing a person to a social norm ( Grotlüschen et al., 2011 ). The leo. aimed to assess people on the lower end of the literacy spectrum. The authors identified functional illiterates as those who perform in the first, second, or third level in the leo. According to their results, 14.5% of the working-age population (about 7.5 million people) in Germany is functionally illiterate ( Grotlüschen and Riekmann, 2011a ). It is important to note that 3.1 million adults (41%) of the estimated functional illiterate population were not native German speakers ( Grotlüschen et al., 2014 ). This is a point which we view as critical, because despite general reading and writing skills, we are all functional illiterates in most foreign languages. In our view (outlined below), language production and comprehension do not need to be that of a native speaker, but should at least be mastered without major problems before a specific deficit in functional illiteracy can be diagnosed. Otherwise, what seems to be a fundamental reading problem is simply a problem of not mastering sufficiently a foreign language. Finally, and unfortunately, the test lacks multivariate analyses of construct validity and only descriptive statistics are available. Consequently, results and conclusions have to be interpreted with caution.

The authors suggest the individual differences resulting from various social roles make it impossible to create a general functional illiteracy test. They argue that different skills are required, for example for a highly qualified IT expert or a motor mechanic ( Grotlüschen and Riekmann, 2011b ).

For specific professions this is a valid argument, but it also raises the question of whether a general construct of literacy exists. To return to the example, in everyday life, IT experts and motor mechanics have to operate machines (e.g., laundry machine), have to read their bank statements, have to take medicine (and read package inserts), have to compare prices in the supermarket, etc. Therefore, we assume that some basic functional literacy skills should exist.

While the leo. is not considered a universal instrument for functional illiteracy by its authors, the Tests of Adult Basic Education (TABE) is a universal instrument to assess the mastery of basic skills and skills-growth measurement. The test includes practical, life-skills stimuli in an adult-relevant context (life-skills, work, and education) and contains tasks from the very low literacy level (e.g., recognizing letters, signs) to the advanced level ( CTB/McGraw-Hill, 2008 ). The comprehensibility of the measured skills and the universality of the tasks suggest that it is possible to create an assessment to measure functional illiteracy, despite the fact that the main aim of the assessment is different.

It is important to note that functional illiterates (as low literate adults) would show floor effects in standard adult literacy (AL) and text comprehension tasks. This would make appropriate identification and within-group distinctions impossible. Therefore, it is worth considering the application of standardized tests for children to measure functional illiteracy. On the one hand, Egloff et al. (2011) argue for a competence-based approach to identify functional illiterates instead of a norm-oriented view . They suggest that it would be better to take different social expectations into account and handle the category of functional illiteracy as a less static phenomena. But on the other hand, they accept to use reading and spelling tasks (with child norms) with well-defined cut-off values to classify functional illiterates ( Egloff et al., 2011 ).

To sum up, many methods have been used to identify functional illiterates, but none of these methods are yet standardized and systematically diagnostically evaluated in a representative sample of functional illiterates and adults. Therefore, they cannot be considered adequate for measuring and identifying functional illiterates on the basis of the current data.

Who is Defined as Functional Illiterate?

Functional illiteracy is assumed to originate from cognitive or linguistic disorders and/or be associated with a sociocultural disadvantage ( Eme, 2011 ; Boltzmann and Rüsseler, 2013 ). The diagnostic assessments and therefore the definition of the sample in different studies is not consistent and sometimes not even explicit.

For a rough categorization, we can divide sample definition of functional illiteracy in scientific publications into three groups:

• simple (1) Some studies call their sample “functional illiterates,” but do not give any reason/explanation/diagnostic justification ( Van Linden and Cremers, 2008 ; Kosmidis et al., 2011 ). From an educational-psychological perspective, it is not acceptable to categorize a subgroup without any empirical reason for doing so.
• simple (2a) Some studies conduct experiments on adults taking part in basic courses [AL or adult basic education (ABE) classes] and call them functional illiterates ( Thompkins and Binder, 2003 ). The similarity between functional illiterates and AL or ABE students is appropriate but has its shortcomings. In particular, it is not evident why people take these courses. Did they have sufficient schooling and nevertheless did not learn to read and write? Did they have insufficient schooling for whatever reason without the chance to become literate? Do they have profound reading/writing problems or are they taking these courses for other reasons (e.g., because the job center recommends doing them)? In short, the problem is that we have no assessment of how severe their functional illiteracy problem really is and whether we are encountering functional illiteracy or real illiteracy due to insufficient schooling.
• simple (2b) It should be noted that there is also another group of studies concerning those who conduct experiments on AL or ABE students but do not call them functional illiterates ( Greenberg et al., 1997 ; MacArthur et al., 2010 ). Despite that, theoretical backgrounds and reviews (e.g., Eme, 2011 ) frequently use these articles, which point out one main limitation of the field.
• simple (3) Only a German and a French research group made explicit how they determine functional illiteracy in their studies. From the German side, Grosche (2012) used reading-level match design is his dissertation and labeled those ABE students as functional illiterates, who performed in two standardized reading tests in the level of first–fourth grade children ( Grosche, 2012 ). While Rüsseler et al. (2013) used German diagnostic reading and spelling tests and involved only those adults to their intervention study who performed worse than average fourth grade level ( Boltzmann and Rüsseler, 2013 ; Boltzmann et al., 2013 ; Rüsseler et al., 2013 ) 1 . The French group measured five components: phonological processing, orthographic processing, sentence comprehension, reading speed, and reading comprehension. Those ABE students who performed below the third grade level were then classified as functional illiterates ( Eme et al., 2010 ). Three problems stick:
• simple (i) The deficits of adult groups are defined as (severe) developmental delays. This cannot be taken for granted; for many adult deficits, and even for dyslexia, different patterns of deficits and developmental delays have been observed.
• simple (ii) Even if one accepts that functional illiteracy is merely developmental delay, there is an inconsistency as regards the severity of the delay. While Rüsseler et al. (2013) suggest lower performance than (average) fourth grade level, Eme et al. (2010) suggest a more severe performance deficit even below third grade level.
• simple (iii) The components for defining functional illiteracy differ between studies: while Rüsseler and colleagues use reading and spelling tests ( Boltzmann and Rüsseler, 2013 ; Boltzmann et al., 2013 ; Rüsseler et al., 2013 ), Eme et al. (2010) use a much broader range of test components. It is still unknown which approach is more valid. In most definitions, functional illiteracy is mainly about impaired understanding of texts. We suggest that diagnostic tests should operationalize this definition and focus on impaired understanding of texts, until other test components prove important for diagnostic assessment of functional illiteracy.

In sum, there is inconsistency in definition and assessment of functional illiterates in the scientific literature. There are only a few studies that include well-established methods in the fundamental sampling question. As the literature lacks a clear definition and clear assessment criteria, we use the term “functional illiterate” to refer to all the participants from the three groups of scientific papers.

Factors Contributing to Functional Illiteracy – The Scientific Aspect 2

Unfortunately, few studies 2 investigated differential diagnostic properties of functional illiteracy. Although there are related deficits that may or may not be part of functional illiteracy depending on the definition and the assessment tool. Here, we focus on three of these related deficits: language-related deficits, general cognitive deficits, and deficits related to numerical abilities (Supplementary Tables S1–S3 ).

Language-Related Deficits

The few articles that assess the basic skills of their specific sample separately have shown that functional illiterates have phonological processing deficits. Their profile is more similar to children with developmental dyslexia than to typical elementary school children. Adults performed much worse in phonological tasks than children matched for reading-level ( Greenberg et al., 1997 ; Thompkins and Binder, 2003 ; Grosche, 2012 ; Eme et al., 2014 ).

Functional illiterates’ spelling skills are also weak ( Greenberg et al., 1997 ; Thompkins and Binder, 2003 ; Eme et al., 2014 ): They rely more on orthographic processes ( Greenberg et al., 1997 ), although they may also have orthographic processing difficulties ( Greenberg et al., 1997 ; Thompkins and Binder, 2003 ; Eme, 2006 ). A comparison with reading-level matched children showed that their vocabulary size is also smaller ( Greenberg et al., 1997 ; Eme et al., 2014 ) and they are slower in naming tasks ( Grosche, 2012 ). Although functional illiterates seem to be a heterogeneous group, on the whole they performed poorer in phonology than in morphosyntax and semantics, with their low performance in oral language tasks being reflected in their written abilities ( Eme et al., 2014 ).

This issue is further complicated by the fact that functional illiterates may not be a homogeneous sample. Eme et al. (2010) suggested that functional illiterates can be divided into five subtypes according to their oral narrative abilities ( Eme et al., 2010 ). However, when the same research group examined the relationship between reading, spelling, and oral language abilities in a later study, the cluster analysis showed four profiles ( Eme et al., 2014 ). So, the subtyping problem is not resolved yet.

Other papers ( Eme, 2006 ; Grotlüschen and Riekmann, 2011a ; Rüsseler et al., 2011 ; Eme et al., 2014 ) mention that functional illiterates have problems in text understanding but only one study examined whether more fundamental factors cause this difficulty. The paper that compared matched normal readers with functional illiterates and children with reading and writing disabilities found that the perceptual skills of functional illiterates are weak but have no impact on reading abilities ( Rüsseler et al., 2011 ).

In sum, functional illiterates seem to have linguistic deficits in several domains, including phonological, orthographic and lexical processing, oral and reading comprehension, and verbal fluency. However, these deficits may not be homogeneous. It is important to note that correlated or co-morbid deficits are not necessarily functionally causal. What is more, they do not necessarily add unique variance to the diagnostic assessment. Finally, we do not know whether the linguistic inabilities described above are their main difficulties or whether these are due to or influenced by other more general cognitive factors (Supplementary Table S1 ).

Cognitive Deficits

Cognitive deficits of functional illiterates have also been reported. Van Linden and Cremers (2008) showed that functional illiterates performed significantly worse than literates not only in language processing, but also in all cognitive tasks such as in copying and recalling the Rey Complex Figure, visual organizational, and visual memory, mental spatial orientation as well sustained or split attention tasks ( Van Linden and Cremers, 2008 ).

Functional illiterates seem to have working memory difficulties: they performed worse than reading-level matched children ( Eme, 2006 ; Grosche, 2012 ) and than normal adult readers ( Grosche, 2012 ) in the verbal tasks. Comparing functional illiterates with children matched for reading-level, adults performed better on a backward, while they did not differ in a forward digit span task ( Thompkins and Binder, 2003 ). However, the studies only used digit or letter span tasks ( Thompkins and Binder, 2003 ; Eme, 2006 ; Grosche, 2012 ).

As regards perceptual skills, functional illiterates perform similar to children with reading and writing disabilities and differ from regular adult readers. This supports a developmental delay view on functional illiteracy ( Rüsseler et al., 2011 ). The authors suggest that perceptual training could develop functional illiterates, as it improved the reading and spelling performance of children with reading and writing disabilities ( Rüsseler et al., 2011 ).

In sum, it is clear that functional illiterates deviate from adults; their performance seems to be more similar to children. However, basic control variables (e.g., intelligence) are often missing, when the cognitive abilities of functional illiterates are assessed. Moreover, again participant selection could drive the results and the subsequent interpretations of deficits. Nevertheless, the available data point to the view that functional illiterates seem to show various cognitive deficits. However, the question about whether these deficits are (partially) causal for the functional illiteracy or just co-morbid impairments remains unanswered so far (Supplementary Table S2 ).

Deficits Related to Numerical Abilities and Dyscalculia

Although numerical abilities are measured as one of the basic skills and are considered as part of functional illiteracy (e.g., in IALS as quantitative literacy, Thorn, 2009 ; in ALL and in PIAAC as numeracy, Statistics Canada and OECD, 2005 ; OECD, 2013 ), research on numerical deficits in functional illiteracy has largely been neglected (Supplementary Table S3 ). Therefore, further experimental studies are needed to answer the question whether functional illiterates have numerical difficulties or not.

(Functional) Illiteracy Programs – The Practical Aspect

In order to eradicate illiteracy, governments, NGOs (non-governmental organization) and supranational agencies such as UNESCO fund numerous programs worldwide ( Abadzi, 2003 ), but the programs are assessed with great skepticism in the literature ( Shi and Tsang, 2008 ). It is important to note that the ABE programs are rarely targeted explicitly at functional illiterates, as they generally aim to increase the participants’ literacy skills 3 .

In Western societies, adult literacy programs are often offered to vulnerable or hard-to-reach learners. Some programs rely extensively on the use of technology and distance learning platforms (e.g., AlphaRoute in Canada), others are tailored to each participant’s needs, both in workshops and individual help (e.g., Fight Against Illiteracy in France). According to their main interest, we can differentiate from general literacy courses the work- (e.g., El Trabajo En Red Como Proyecto Educativo in Spain) and family-oriented (e.g., Family Literacy Project in Germany) programs ( Aker et al., 2010 ). Former supports the (re)integration to labor market ( Bhola, 1995 ), while latter’s key-strategy called the “Teach the parents – reach the children” approach in which parents and their children are working both separately and together. It aims at a long-term effect in the education of next generation ( Nickel, 2007 ). Furthermore, supplementing literacy and numeracy classes with technology, even mobile phones, is restricted by its reduced availability ( Aker et al., 2010 ).

Adult basic education classes are still struggling to overcome high drop-out rates, failure to pass literacy tests, and a fast deterioration of literacy skills. High drop-out rates are associated with younger age, worse blending, slower naming, and comprehension skills, as well as increased avoidance of reading difficult materials. Furthermore, current/past enrollment in ABE classes increased the probability of midpoint completion ( Greenberg et al., 2012 ). Therefore, the programs should pay more attention to the participants that fall within these categories. In Germany, Rüsseler et al. (2012) created and investigated the effects of a special training program called Alpha Plus. While the regular literacy courses offer reading and writing classes once a week, the intensive Alpha Plus training does not only improve reading and writing skills. But it builds also on the progress of other basic, daily and work-related abilities (e.g., perceptual and social skills). The program is clearly more effective than the regular classes offered to functional illiterates by the adult education schools in Germany. The efficiency of Alpha Plus was confirmed by behavioral, ERP, and fMRI studies ( Rüsseler et al., 2012 ; Boltzmann and Rüsseler, 2013 ; Boltzmann et al., 2013 ; Rüsseler et al., 2013 ). The success of the program is evident but the authors stress the large variability between the participants. The achievement would be larger if it could better handle individual differences (e.g., with more groups with smaller sizes; Rüsseler et al., 2013 ) and follow a more personalized adaptive learning approach.

To sum up, solving the problem of illiteracy and functional illiteracy is relevant to governments and various organizations and their efficiency show up in statistics ( UNESCO, 2015 ). But the development of programs based on scientific research (e.g., Alpha Plus: Rüsseler et al., 2012 ) could improve the efficacy of the programs and the persistence of the students.

Dissociating Functional Illiteracy From Illiteracy and Dyslexia

For establishing a solid picture about the construct of functional illiteracy, it is necessary to distinguish it from related constructs such as illiteracy and developmental dyslexia, and to define non-overlapping characteristics. Without such dissociation, functional illiteracy is just a new name for a deficit that is already part of other constructs.

Functional Illiteracy and Illiteracy: What Does Functionality Mean?

Illiteracy is a well-defined phenomenon and the diagnostic criteria for this group are clear-cut. It has been investigated since the 1970s and researchers have investigated many characteristics of illiteracy ( Huettig and Mishra, 2014 ). According to the original notion, the difference between functional illiterates and illiterates is that illiterates are unable to read, write, and understand short sentences. In contrast functional illiterates are unable to use their acquired literacy skills in daily life ( UNESCO, 1978 ), e.g., to read and understand a medicine label or a bank statement, fill out a job application, compare the cost of two items and choose the item that offers the best value ( Cree et al., 2012 ).

When we outline these studies, we focus on the same three related groups of deficits we distinguished for functional illiterates (Supplementary Tables S1–S3 ).

Language-Related Deficits in Illiterates

As the illiterates have never attended school and did not acquire basic language skills, they differ in most language-related abilities. It is known that phonemic awareness is not attained spontaneously, since associations of phonemes with graphemes emerge with reading acquisition ( Morais et al., 1979 ). Indeed, performances on phoneme addition, discrimination, deletion, and pseudoword repetition tasks (e.g., Greenberg et al., 1997 ; Thompkins and Binder, 2003 ) clearly demonstrated that illiterates have phonological processing deficits ( Morais et al., 1979 ; Rosselli et al., 1990 ; Reis and Castro-Caldas, 1997 ; Castro-Caldas et al., 1998 ).

Decreased performance was shown also in orthographic ( Petersson et al., 2000 ) and in lexical processing ( Kosmidis et al., 2006 ) when low literate and literate adults were compared.

In addition, researchers observed impairments in naming ability ( Rosselli et al., 1990 ; Ostrosky-Solis et al., 1999 ; Reis et al., 2006 ), in oral comprehension ( Rosselli et al., 1990 ; Ostrosky-Solis et al., 1999 ) and in verbal fluency skills ( Rosselli et al., 1990 ; Reis and Castro-Caldas, 1997 ; Ostrosky-Solis et al., 1999 ; Kosmidis et al., 2004 ) as well. Yet, it is important to mention that when using ecologically more valid categories in the verbal fluency task (e.g., supermarket), the difference can disappear ( Reis et al., 2003 ).

In sum, illiterates can be characterized by impairments in the whole spectrum of language-related skills (Supplementary Table S1 ), which are less variable than those of functional illiterates.

Cognitive Deficits in Illiterates

As lack of reading and writing acquisition affects language skills, could it be assumed that basic cognitive functions also depend on it? The need for assessing the cognitive abilities of illiterates materialized many years ago.

Illiterates performed significantly worse than the three other assessed educated groups (1–4; 5–9; 10–24 years of education) in abilities as orientation, verbal fluency, attention, perception, and motor functions ( Ostrosky-Solis et al., 1999 ; Dansilio and Charamelo, 2005 ; Landgraf et al., 2011 ). The latter was confirmed in visuo-motor integration tasks as well: while literates used a systematic visual scanning strategy, illiterates were less systematic and slower in a computerized visual-motor task ( Bramão et al., 2007 ).

Oral cultures have better long-term memory abilities, as they can preserve their traditional songs by rote learning ( Huettig and Mishra, 2014 ). Conversely, illiterates did not succeed in standardized working memory tasks ( Ardila et al., 1989 ; Reis et al., 2003 ; Kosmidis et al., 2011 ; Silva et al., 2012 ). In addition, Kosmidis et al. (2011) revealed that literacy per se and not formal schooling affected working memory skills.

In sum, illiterates perform worse in various cognitive skills than literates. The deficits seem more universal than in studies with functional illiterates. Lack of education and basic skill acquisition have been brought forward as the reason for the weakness of cognitive skills in illiterates ( Ardila et al., 1989 ; Rosselli et al., 1990 ) (Supplementary Table S2 ).

Deficits Related to Numerical Abilities in Illiterates

Although illiterates never attended school and never acquired number reading and writing, the majority of the tests that examine mental calculation or basic arithmetical abilities were administered to illiterates in written form. It is not surprising that these studies solidly verified that illiterates have poor mental calculation or basic arithmetical abilities ( Ostrosky-Solis et al., 1999 ; Reis et al., 2003 ; Landgraf et al., 2011 ; Silva et al., 2012 ). Only one experiment gave calculations orally where the illiterates achieved low score as well ( Rosselli et al., 1990 ). However, it is also possible that the deficits extend to basic number sense. Halberda and Feigenson (2008) have shown that early processing of non-symbolic information long before formal schooling influences arithmetic performance at a later age ( Halberda and Feigenson, 2008 ). Whether the so-called approximate number system (ANS) – measured by non-symbolic magnitude comparison – really contributes to symbolic and arithmetic performance when other symbolic factors are controlled is a matter of intense discussion ( De Smedt et al., 2013 ; Lyons et al., 2015 ). The answer to this question is not easy as performance in ANS tasks and their correlations with arithmetic seem to depend on the particular method involved ( Dietrich et al., 2015 ). Nevertheless, it would be helpful to assess more basic numerical abilities like the ANS or spatial-numerical capabilities ( Siegler and Opfer, 2003 ; Moeller et al., 2009 ) or indices of multi-digit integration ( Moeller et al., 2011 ; Nuerk et al., 2015 for a review) to identify basic numerical deficits in functional illiterates that might lead to deficits in later more complex arithmetic tasks.

In sum, illiterates performed less accurately not only in language-related tasks, but also in cognitive and mathematical tasks. But it remains unclear whether the lack of reading acquisition, the absence of formal education, or even basic perceptual and cognitive deficits underlying more than one skill drive their functional illiteracy (Supplementary Table S3 ).

Functional Illiteracy and Dyslexia: Different Constructs for the Same Sample?

Is it possible that functional illiterates are dyslexics with a new name?

We have outlined above that various language deficits are part of functional illiteracy. Some authors even claim that functional illiterates can somehow count as untreated developmental dyslexics ( Greenberg et al., 1997 ; Grosche, 2012 , but see diagnostic problematic outlined above). Therefore, it is unclear whether the terms “functional illiterate” and “dyslexic” reflect different terminology used to refer to the same group of people due to preference and history of the field, rather than due to actual differences between the two groups. It is surprising that we have not found any experimental research that has investigated this thesis. Therefore, we will outline developmental dyslexia in more detail, again with the same three subsections, language-related deficits, general cognitive, and numerical deficits (Supplementary Tables S1–S3 ).

Language-Related Deficits in Dyslexia

Developmental dyslexia is associated with abnormalities in a variety of brain regions, and has a strong genetic basis ( Lyon et al., 2003 ; Fletcher, 2009 ; Habib and Giraud, 2013 ). However, it is not clear whether the neurobiological changes are a cause or consequence of reading difficulties.

Dyslexic children have problems in at least three domains: decoding single words, reading fluency, and comprehension ( Fletcher, 2009 ). Leading theories suggest that the main problem in dyslexia is the phonological processing deficit. It can appear even at a single word level, independently of intelligence and is adequate for a dyslexia diagnosis ( Ramus et al., 2003 ). Such a deficit in phonological awareness was confirmed in children by many studies ( Joanisse et al., 2000 ; Casalis et al., 2004 ; White et al., 2006 ; Everatt et al., 2008 ; Landerl et al., 2009 ; Varvara et al., 2014 ; Zoubrinetzky et al., 2014 ). The most common tasks were phonological fluency ( Landerl et al., 2009 ; Varvara et al., 2014 ) and manipulation with phonemes as phoneme deletion ( Joanisse et al., 2000 ; Landerl et al., 2009 ; Chung et al., 2010 ; Zoubrinetzky et al., 2014 ) and spoonerism tasks ( White et al., 2006 ; Varvara et al., 2014 ).

The results suggest that the phonological symptoms associated with dyslexia persist into adulthood ( Hatcher et al., 2002 ; Ramus et al., 2003 ; Beidas et al., 2013 ; Bogdanowicz et al., 2014 ; Law et al., 2015 ). A study that compared adults with and without learning difficulties demonstrated that even high-achieving dyslexic adults are slower in phonological, semantic, and syntactic judgment tasks ( Rüsseler et al., 2007 ).

In spelling, the tendency remains similar: both dyslexic children ( White et al., 2006 ; Everatt et al., 2008 ; Chung et al., 2010 ) and adults ( Hatcher et al., 2002 ; Beidas et al., 2013 ; Law et al., 2015 ) showed difficulties in their performance. In contrast, dyslexic adults performed well in the semantic fluency task ( Hatcher et al., 2002 ) and vocabulary tasks (e.g., Cavalli et al., 2016 ) but the success of children were mixed ( Joanisse et al., 2000 ; White et al., 2006 ; Everatt et al., 2008 ; Landerl et al., 2009 ; Varvara et al., 2014 ).

Nevertheless, their reading and naming speed were also significantly slower than in children and adults without learning difficulties ( De Luca et al., 2002 ; Hatcher et al., 2002 ; Ramus et al., 2003 ; White et al., 2006 ; Everatt et al., 2008 ; Willburger et al., 2008 ; Boets and De Smedt, 2010 ; De Smedt and Boets, 2011 ; Beidas et al., 2013 ; Bogdanowicz et al., 2014 ; Suarez-Coalla et al., 2014 ; Law et al., 2015 ).

Basic language-related skills are necessary for accurate text comprehension ( Martens and de Jong, 2006 ). Therefore, it is not surprising that dyslexic children and adults systematically perform below-average on reading comprehension tasks ( Casalis et al., 2004 ; Fletcher, 2009 ; Rimrodt et al., 2009 ; Wiseheart et al., 2009 ; Rello et al., 2013 ). If texts are optimized according to word-frequency and word-length, thus using more common and shorter words, dyslexic adolescents and adults understand better and read faster written materials ( Rello et al., 2013 ).

Dyslexia is not only categorized by phonological deficit, reading fluency, and text comprehension; it is also considered to be a heterogeneous learning disorder ( Zoubrinetzky et al., 2014 ). Co-morbid language deficits and other cognitive difficulties are common. A regression study that aimed to examine the contribution of linguistic and cognitive factors to oral reading fluency in dyslexic adolescents found that word decoding, working memory, and vocabulary are the key predictors. The factors together explain 56% of the variance in connected-text oral reading fluency ( Rose and Rougani, 2012 ). Despite the regression analysis, one must keep in mind that these are still correlations. Whether co-morbid cognitive difficulties causally influence reading fluency or whether linguistic deficits cause associated cognitive problems over the course of learning and development is not entirely clear yet ( Beidas et al., 2013 ) (Supplementary Table S1 ).

In sum, we can conclude that dyslexic children have problems in phonological tasks, reading fluency, reading comprehension and associated linguistic and cognitive factors. Most such deficits observed in dyslexic children are preserved in adulthood. However, dyslexic adults may be able to compensate some of their deficits (e.g., in reading comprehension) and function better in language-related tasks than functional illiterates. Whether this summary of the literature holds, must be examined, with direct investigation of dyslexics and functional illiterates.

Cognitive Deficits in Dyslexia

In the last decades, auditory, visual processing, or attention deficits were suggested as being potential sources of dyslexia. Valdois et al. (2004) argue that phonological and attention deficits in dyslexic patients can present independently from each other ( Valdois et al., 2004 ). Accordingly, dyslexics struggle with attentional and perceptual difficulties ( Ramus et al., 2003 ; Ziegler et al., 2010 ; Leong et al., 2011 ; Beidas et al., 2013 ; Bogdanowicz et al., 2014 ; Varvara et al., 2014 ; Zoubrinetzky et al., 2014 ).

As regards cognitive abilities, most articles are examining working memory. It was shown that dyslexic children have poor working memory ( Beneventi et al., 2010 ; Varvara et al., 2014 ), which remains weak during adulthood ( Ramus et al., 2003 ; Abd Ghani and Gathercole, 2013 ; Beidas et al., 2013 ; Bogdanowicz et al., 2014 ). This deficit seems stable, considering that weak performance appears both in verbal (e.g., digit span, e.g., Everatt et al., 2008 ), in spatial (e.g., Corsi blocks, Landerl et al., 2009 ), and in visual (e.g., n-back, Beneventi et al., 2010 ) working memory tests. Exploring the four regions of executive functions (inhibition, planning, sequencing, and organizing abilities), researchers found that compensated dyslexic university students did not differ from the non-dyslexic control group in any of the functions ( Brosnan et al., 2002 ). A more recent study showed that in a set shifting task, dyslexic adults were slower than age and IQ matched controls. In contrast, in an inhibition task the reaction time did not differ, although the accuracy depended on the task ( Smith-Spark et al., 2016 ).

Experiments showed that dyslexic children have no problems in tasks requiring fine manual skills ( White et al., 2006 ; Everatt et al., 2008 ) but they have difficulties in balancing tasks ( White et al., 2006 ; Brookes et al., 2010 ). Conversely, adults did not show any problems in balance and motor coordination tasks ( Ramus et al., 2003 ).

In sum, diverse types of cognitive difficulties are inseparable from the symptoms of dyslexia both in childhood and adulthood. Over time, dyslexics can improve some of their skills but most of their problems are remained. Nevertheless, their deficits seem less universal than in functional illiteracy (Supplementary Table S2 ).

Deficits Related to Numerical Abilities and Dyscalculia in Dyslexia

Research examining mathematical abilities has shown that dyslexic children and adults generally solved basic arithmetical problems slower and less accurately than children and adults without dyslexia ( Hatcher et al., 2002 ; Simmons and Singleton, 2006 ; Boets and De Smedt, 2010 ; De Smedt and Boets, 2011 ).

A study examining children with reading disability and/or math disability found that all three groups showed difficulties in the examined neuropsychological measures. However, the impairments of reading and math disability group were the largest ( Willcutt et al., 2013 ). Studies confirmed that reading and mathematical learning disabilities have independent domain-specific deficits: in the case of dyslexia in phonological processing and numerosity in the case of dyscalculia. Nevertheless, there are some common domain-general “bridge symptoms” as rapid naming ( Wilson et al., 2015 ), working memory, processing speed, and verbal comprehension ( Willcutt et al., 2013 ). In contrast, another experiment described that the cognitive deficits of children with dyslexia and dyscalculia were only additive ( Landerl et al., 2009 ) (Supplementary Table S3 ).

The Triple-Code Model ( Dehaene and Cohen, 1995 ) supposes three distinguished mental representations of numbers within different brain areas. According to the model, we can distinguish visual representation (established in the left and right inferior ventral occipito-temporal areas), magnitude representations (established in the left and right inferior parietal areas), and verbal representation (established in the left-hemispheric perisylvian language areas). Thus, the numerical and linguistic representations work separately. Therefore, those who have poor numerical or poor reading skills might be differentiated clearly according to their anatomical and functional brain processes ( Dehaene and Cohen, 1995 , 1998 ).

In sum, we can state that reading disabilities do not go obviously hand in hand with mathematical weaknesses, therefore, not just dyslexics and dyscalculics but also functional illiterates and functional innumerates may represent separate groups (Supplementary Tables S1–S3 ).

Summary, New Definition, and Further Challenges

From the outline of the review, it is clear that the field of functional illiteracy has been under-represented in research despite its worldwide effects on social and economic levels ( UNESCO, 2006 ) and although millions of dollars are invested in remediation programs of (functional) literacy.

In this review, we clarified our knowledge about functional illiterates, especially how different approaches try to diagnose them, and in what areas they differ from illiterates and dyslexics.

We summarized the challenges of empirical research that hinder the researchers of the field as the lack of an adequate assessment and resources for programs and researches.

A comprehensive, exploratory examination is needed to guarantee the success of the literacy programs. This examination should assess in detail the basic foundations and the variables that play a crucial role in functional illiteracy, emphasizing not only the language, but the mathematical-related and cognitive skills which are essential in everyday life.

The first step in that direction is to establish a new, up-to-date definition that is adequate for experimental research:

Functional illiteracy is the incapability to understand complex texts 4 despite adequate schooling, age, language skills, elementary reading skills, and IQ. These inabilities must also not be fully explained by sensory, domain-general cognitive, neurological or psychiatric deficits.

Here we suppose the main criteria and justification that a working definition should contain:

Inclusion criteria:

• simple – very poor performance in a functional illiteracy assessment: despite the fact that there is no consensus about an operationalized definition of functional illiteracy, many self-claimed assessments tried to measure it, but there is no standardized and validated tool for this aim 5 ,
• simple – age: older than 16 years old. We suppose that children cannot be categorized as functional illiterates,
• simple – schooling: minimum 6–8 finished years, in agreement with the duration of compulsory education for single countries (in Germany it means 9 years),
• simple – proper (German) language use: fluent, native-like oral language skills without major difficulties (natives, bilinguals). We should take with great care people with migration background because we cannot be sure whether a person shows weakness because he/she is a functional illiterate or because he/she has difficulties in second language acquisition. Nevertheless, being a native speaker is in our view not a necessary criterion if the second language is sufficiently well mastered in oral language,
• simple – IQ: level of 70 or above.

Exclusion criteria:

• simple – neurological or mental disorder,
• simple – uncorrected speech, hearing, or vision problem.

Exclusion criteria for pure functional illiteracy: 6

• simple – dyslexia,
• simple – dyscalculia,
• simple – hyperactivity.

Further characteristics that describe functional illiterates:

• simple – impaired oral language comprehension,
• simple – impaired writing skills,
• simple – impaired arithmetic skills,
• simple – difficulties in functioning in society: problems with active, independent functioning in daily life.

Due to lack of empirical studies the underlying cause of functional illiteracy is still unclear. Rüsseler et al. (2011) suggested a combined model, where the unfavorable familiar background and school experiences could be identified as risk factors and together with biological and cognitive determinants could cause functional illiteracy ( Rüsseler et al., 2011 ).

As regards our five research questions in the beginning, they can be answered as follows. We propose four different social and cognitive aspects that can lead to functional illiteracy in itself or together:

• simple (1) Cognitive aspect: weak cognitive skills cause the inability to acquire proper basic literacy skills;
• simple (2) Educational aspect: primary and secondary school teachers have no opportunity to take care the individual level of each student, therefore the children with feeble abilities or low motivation fall behind in long-term;
• simple (3) Social aspect: the lack of an encouraging and motivating model in a child’s family for acquiring new skills, having new experiences, can lead to an unmotivated learning style in school;
• simple (4) Competency loss aspect: loss of competencies in adulthood caused by a decrease of cognitive demands (Q4).

The focus on cognitive and social aspects does not preclude that some of them (e.g., the cognitive aspects) are neurobiologically routed.

The review shows that despite formal education, functional illiterates do not possess basic skills (Q1). This general deficit can be theoretically distinguished from the deficits associated with illiteracy and dyslexia; illiterates lack formal education, while functional illiterates have had some schooling and therefore may have advantages from this education. Additionally, dyslexia has genetic underpinnings while social factors seems to have stronger impact on the development of functional illiteracy (Q5), therefore their diagnostic and remediation processes may differ as well.

From the summary we cannot conclude in which abilities functional illiterates have the largest deficit, because we did not find any research that aimed at measuring their mathematical abilities (Q2). We suppose that functional illiterates have both numerical and linguistic deficits. According to the Triple-Code Model, the underlying representations work separately ( Dehaene and Cohen, 1995 , 1998 ) but we do not know any research that has tried to confirm this in a functional illiterate sample (Q3).

Summarizing our presumptions about functional illiteracy in details, we define as functional illiterates those adults who attended the compulsory years in education but could not acquire basic reading, writing, and calculation skills. Their impairments negatively affect their effective functioning in everyday life. In particular, functional illiterates have poor language skills (writing, reading, oral communication) (e.g., difficulty understanding a medicine label) as well as poor arithmetic abilities (e.g., inability to compare the price of two products) that generally influence everyday life situations (e.g., get the information from a timetable). People belonging to this group have average or below-average IQ levels and their difficulties cannot result from any other kind of neurological or psychiatric disorder, organic problem, non-verbal learning problem, general learning difficulty or hyperactivity. Of course, these criteria do not exclude co-morbidities with such other impairments.

We would stress the need for methodologically more substantiated research, comparing basic linguistic, numerical and cognitive functions in normal readers, functional illiterates, dyslexic adults, and reading-level matched dyslexic children.

Author Contributions

RV, TD, JS, and H-CN made the review design, RV and AC did the literature search, and RV, AC, TD, JS, and H-CN wrote the paper.

Conflict of Interest Statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Acknowledgments

This research (“Basic foundations of functional illiteracy”) is funded by the LEAD Graduate School & Research Network [GSC1028], a project of the Excellence Initiative of the German federal and state governments. RV is a doctoral student of the LEAD Graduate School & Research Network. We acknowledge support by the Deutsche Forschungsgemeinschaft and the Open Access Publishing Fund of the University of Tuebingen.

1 The authors explicitly wrote this criteria only in Rüsseler et al. (2013 , p. 242) but as they speak about the evaluation of the same training program in Boltzmann and Rüsseler (2013) and in Boltzmann et al. (2013) , we suppose that they used the same inclusion criteria.

2 Relevant studies for this review were identified by (1) carrying out a keyword search in EBSCOhost, PsycInfo, and Google Scholar. It was conducted for keywords functional illiteracy, illiteracy, literacy, adult dyslexia, child dyslexia, and several variations of these keywords and the basic abilities that we mention in Supplementary Tables S1–S3 . (2) And we were conducting a manual search for references cited in relevant papers.

3 http://www.unesco.org/uil/litbase/?menu=7

4 Beyond this scope of review but we define complex text as comprising at least of two sentences with some conjunctions or subjunctions and propositional relations between these sentences. Questions concerning such tests should be impossible to answer on the basis of understanding one single sentence alone. For a more precise definition we suppose computer linguistically quantifiable measures about the readability and complexity of a text (e.g., after the methods of Vajjala and Meurers, 2014 ).

5 According to our opinion, functional illiterates in general have fundamental problems in text comprehension. Therefore, we suggest using tasks based on text comprehension, enhanced with an interview about their educational background. We agree with Boltzmann and Rüsseler (2013) that children’s tasks are well suited for assessing the functional illiterate sample, e.g., because of their complexity that admit of the differentiation and their short length that is not frustrating. However, these tasks have not yet been normed to low literate adults.

6 We would like to stress that we do not want to exclude functional illiterates with dyslexia, but we would like to raise awareness that further research should pay more attention to the related linguistic and numerical impairments. It is likely that functional illiterates meet the criteria for dyslexia and because of the dissociation between dyslexia and functional illiteracy merits further investigations.

Supplementary Material

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpsyg.2016.01617/full#supplementary-material

• Abadzi H. (2003). Adult Literacy: A Review of Implementation Experience . Washington, DC: The World Bank Operations Evaluations Department. [ Google Scholar ]
• Abd Ghani K., Gathercole S. E. (2013). Working memory and study skills: a comparison between dyslexic and non-dyslexic adult learners. Procedia Soc. Behav. Sci. 97 271–277. 10.1016/j.sbspro.2013.10.233 [ CrossRef ] [ Google Scholar ]
• Aker C., Ksoll C., Lybbert T. J. (2010). ABC, 123: The Impact of a Mobile Phone Literacy Program on Educational Outcomes . Washington, DC: Center for Global Development. [ Google Scholar ]
• Ardila A., Rosselli M., Rosas P. (1989). Neuropsychological assessment in illiterates: visuospatial and memory abilities. Brain Cogn. 11 147–166. 10.1016/0278-2626(89)90015-8 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Beidas H., Khateb A., Breznitz Z. (2013). The cognitive profile of adult dyslexics and its relation to their reading abilities. Read. Writ. 26 1487–1515. 10.1007/s11145-013-9428-5 [ CrossRef ] [ Google Scholar ]
• Beneventi H., Tonnessen F. E., Ersland L., Hughdahl K. (2010). Working memory deficit in dyslexia: behavioral and fMRI evidence. Int. J. Neurosci. 120 51–59. 10.3109/00207450903275129 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Bhola H. S. (1995). Functional Literacy, Workplace Literacy and Technical and Vocational Education: Interfaces and Policy Perspectives Paris: UNESCO, Section for Technical and Vocational Education. [ Google Scholar ]
• Boets B., De Smedt B. (2010). Single-digit arithmetic in children with dyslexia. Dyslexia 16 183–191. [ PubMed ] [ Google Scholar ]
• Bogdanowicz K. M., Łockiewicz M., Bogdanowicz M., Pąchalska M. (2014). Characteristics of cognitive deficits and writing skills of Polish adults with developmental dyslexia. Int. J. Psychophysiol. 93 78–83. 10.1016/j.ijpsycho.2013.03.005 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Boltzmann M., Rüsseler J. (2013). Training-related changes in early visual processing of functionally illiterate adults: evidence from event-related brain potentials. BMC Neurosci. 14 : 154 10.1186/1471-2202-14-154 [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Boltzmann M., Rüsseler J., Ye Z., Münte T. F. (2013). Learning to read in adulthood: an evaluation of a literacy program for functionally illiterate adults in Germany. Probl. Educ. 21 Century 51 33–46. [ Google Scholar ]
• Bramão I., Mendonça A., Faísca L., Ingvar M., Petersson K. M., Reis A. (2007). The impact of reading and writing skills on a visuo-motor integration task: a comparison between illiterate and literate subjects. J. Int. Neuropsychol. Soc. 13 359–364. 10.1017/S1355617707070440 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Brookes R. L., Tinkler S., Nicolson R. I., Fawcett A. J. (2010). Striking the right balance: motor difficulties in children and adults with dyslexia. Dyslexia 16 358–373. [ PubMed ] [ Google Scholar ]
• Brosnan M., Demetre J., Hamill S., Robson K., Shepherd H., Cody G. (2002). Executive functioning in adults and children with developmental dyslexia. Neuropsychologia 40 2144–2155. 10.1016/S0028-3932(02)00046-5 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Casalis S., Colé P., Sopo D. (2004). Morphological awareness in developmental dyslexia. Ann. Dyslexia 54 114–138. 10.1007/s11881-004-0006-z [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Castro-Caldas A., Petersson K. M., Reis A., Stone-Elander S., Ingvar M. (1998). The illiterate brain: learning to read and write during childhood influences the functional organization of the adult brain. Brain 121 1053–1063. 10.1093/brain/121.6.1053 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Cavalli E., Casalis S., El Ahmadi A., Zira M., Poracchia-George F., Colé P. (2016). Vocabulary skills are well developed in university students with dyslexia: evidence from multiple case studies. Res. Dev. Disabil. 5 89–102. 10.1016/j.ridd.2016.01.006 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Chung K. K. H., Ho C. S.-H., Chan D. W., Tsang S.-M., Lee S.-H. (2010). Cognitive profiles of chinese adolescents with dyslexia. Dyslexia 16 2–23. 10.1002/dys.392 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Cree A., Kay A., Steward J. (2012). The Economic & Social Cost of Illiteracy: a Snapshot of Illiteracy in a Global Context . Melbourne: The World Literacy Foundation. [ Google Scholar ]
• CTB/McGraw-Hill (2008). Discover TABE 9&10. Monterey, CA: CTB/McGraw-Hill. [ Google Scholar ]
• Dansilio S., Charamelo A. (2005). Constructional functions and figure copying in illiterates or low-schooled Hispanics. Arch. Clin. Neuropsychol. 20 1105–1112. 10.1016/j.acn.2005.06.011 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• De Luca M., Borrelli M., Judica A., Spinelli D., Zoccolotti P. (2002). Reading words and pseudowords: an eye movement study of developmental dyslexia. Brain Lang. 80 617–626. 10.1006/brln.2001.2637 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• De Smedt B., Boets B. (2011). Phonological processing and arithmetic fact retrieval: evidence from developmental dyslexia. Neuropsychologia 48 3973–3981. 10.1016/j.neuropsychologia.2010.10.018 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• De Smedt B., Noël M. P., Gilmore C., Ansari D. (2013). How do symbolic and non-symbolic numerical magnitude processing skills relate to individual differences in children’s mathematical skills? A review of evidence from brain and behavior. Trends Neurosci. Educ. 2 48–55. [ Google Scholar ]
• Dehaene S., Cohen L. (1995). Towards an anatomical and functional model of number processing. Math. Cogn. 1 83–120. [ Google Scholar ]
• Dehaene S., Cohen L. (1998). “Levels of representation in number processing,” in Handbook of Neurolinguistics eds Stemmer B., Whitaker H. A. (Amsterdam: Elsevier; ) 331–341. [ Google Scholar ]
• Dietrich J. F., Huber S., Nuerk H.-C. (2015). Methodological aspects to be considered when measuring the approximate number system (ANS) – a research review. Front. Psychol. 6 : 295 10.3389/fpsyg.2015.00295 [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Egloff B., Grosche M., Hubertus P., Rüsseler J. (2011). “Funktionaler Analphabetismus: eine aktuelle Definition,” in Zielgruppen in Alphabetisierung und Grundbildung Erwachsener. Bestimmung, Verortung, Ansprache ed. Projektträger im DLR e.V (Bielefeld: WBV-Verlag; ) 11–31. . [ Google Scholar ]
• Eme E. (2006). L′examen psycholinguistique et neuropsychologique de personnes en situation d′illettrisme. Rev. Neuropsychol. 16 3–40. [ Google Scholar ]
• Eme E. (2011). Cognitive and psycholinguistic skills of adults who are functionally illiterate: current state of research and implications for adult education. Appl. Cogn. Psychol. 25 753–762. 10.1002/acp.1746 [ CrossRef ] [ Google Scholar ]
• Eme E., Lacroix A., Almecija Y. (2010). Oral narrative skills in french adults who are functionally illiterate: linguistic features and discourse organization. J. Speech Lang. Hear. Res. 53 1349–1371. 10.1044/1092-4388(2010/08-0092) [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Eme E., Lambert E., Alamargot D. (2014). Word reading and word spelling in French adult literacy students: the relationship with oral language skills. J. Res. Read. 37 268–296. 10.1111/j.1467-9817.2011.01508.x [ CrossRef ] [ Google Scholar ]
• European Commission (2014/2015). Compulsory Education in Europe 2014/15. Eurydice – Facts and Figures. Brussels: European Commission. [ Google Scholar ]
• Everatt J., Weeks S., Brooks P. (2008). Profiles of strengths and weaknesses in dyslexia and other learning difficulties. Dyslexia 14 16–41. 10.1002/dys.342 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Fletcher J. M. (2009). Dyslexia: the evolution of a scientific concept. J. Int. Neuropsychol. Soc. 15 501–508. 10.1017/S1355617709090900 [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Greenberg D. (2007). “Tales from the field: the struggles and challenges of conducting ethical and quality research in the field of adult literacy,” in Toward Defining and Improving Quality in Adult Basic Education eds Belzer A., Beder H. (Hillside, NJ: Lawrence Erlbaum Associates; ) 124–140. [ Google Scholar ]
• Greenberg D., Ehri L. C., Perin D. (1997). Are word-reading processes the same or different in adult literacy students and third-fifth graders matched for reading level? J. Educ. Psychol. 89 262–275. 10.1037/0022-0663.89.2.262 [ CrossRef ] [ Google Scholar ]
• Greenberg D., Wise J. C., Frijters J. C., Morris R., Fredrick L. D., Rodrigo V., et al. (2012). Persisters and nonpersisters: identifying the characteristics of who stays and who leaves from adult literacy interventions. Read. Writ. 26 495–514. 10.1007/s11145-012-9401-8 [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Grosche M. (2012). Analphabetismus und Lese-Rechtschreib-Schwächen. Münster: Waxmann. [ Google Scholar ]
• Grotlüschen A., Kretschmann R., Quante-Brandt E., Wolf K. D. (eds) (2011). Literalitätsentwicklung von Arbeitskräften. Münster: Waxmann Verlag. [ Google Scholar ]
• Grotlüschen A., Riekmann W. (2011a). leo. – Level One Study. Hamburg: Universität Hamburg. [ Google Scholar ]
• Grotlüschen A., Riekmann W. (2011b). leo. – Level-One Studie. Literalität von Erwachsenen auf den Unteren Kompetenzniveaus In Presseheft. Hamburg: Universität Hamburg. [ Google Scholar ]
• Grotlüschen A., Riekmann W., Buddeberg K. (2014). “Functional illiteracy in Germany,” in Lifelong Learning and Governance. From Programming to Action – Selected Experiences from Asia and Europe eds Hintzen H., Knoll J. H. (Paris: UNESCO Insitute for Lifelong Learning; ) 55–67. [ Google Scholar ]
• Habib M., Giraud K. (2013). “Dyslexia,” in Handbook of Clinical Neurology, Pediatric Neurology , Part I eds Dulac O., Lassonde M., Sarnat H. B. (Amsterdam: Elsevier; ) 229–235. [ PubMed ] [ Google Scholar ]
• Halberda J., Feigenson L. (2008). Developmental change in the acuity of the “Number Sense”: the approximate number system in 3-, 4-, 5-, and 6-year-olds and adults. Dev. Psychol. 44 1457–1465. 10.1037/a0012682 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Hatcher J., Snowling M. J., Griffiths Y. M. (2002). Cognitive assessment of dyslexic students in higher education. Br. J. Educ. Psychol. 72 119–133. 10.1348/000709902158801 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Huettig F., Mishra R. K. (2014). How literacy acquisition affects the illiterate mind – a critical examination of theories and evidence. Lang. Linguist. Compass 8 401–427. 10.1111/lnc3.12092 [ CrossRef ] [ Google Scholar ]
• Infante I. (2000). Functional Literacy in Seven Latin American Countries. Santiago: UNESCO. [ Google Scholar ]
• Joanisse M. F., Manis F. R., Keating P., Seidenberg M. S. (2000). Language deficits in dyslexic children: speech perception, phonology, and morphology. J. Exp. Child Psychol. 77 30–60. 10.1006/jecp.1999.2553 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Kosmidis M. H., Tsapkini K., Folia V. (2006). Lexical processing in illiteracy: effect of literacy or education? Cortex 42 1021–1027. 10.1016/S0010-9452(08)70208-9 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Kosmidis M. H., Tsapkini K., Folia V., Vlahou C. H., Kiosseoglou G. (2004). Semantic and phonological processing in illiteracy. J. Int. Neuropsychol. Soc. 10 818–827. 10.1017/S1355617704106036 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Kosmidis M. H., Zafiri M., Politimou N. (2011). Literacy versus formal schooling: influence on working memory. Arch. Clin. Neuropsychol. 26 575–582. 10.1093/arclin/acr063 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Landerl K., Fussenegger B., Moll K., Willburger E. (2009). Dyslexia and dyscalculia: two learning disorders with different cognitive profiles. J. Exp. Child Psychol. 103 309–324. 10.1016/j.jecp.2009.03.006 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Landgraf S., Beyer R., Pannekamp A., Schaadt G., Koch D., Foth M., et al. (2011). Dissociating improvement of attention and intelligence during written language acquisition in adults. Int. J. Intell. Sci. 1 17–24. 10.4236/ijis.2011.12003 [ CrossRef ] [ Google Scholar ]
• Law J. M., Wouters J., Ghesquière P. (2015). Morphological awareness and its role in compensation in adults with dyslexia. Dyslexia 21 254–272. 10.1002/dys.1495 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Leong V., Hämäläinen J., Soltész F., Goswani U. (2011). Rise time perception and detection of syllable stress in adults with developmental dyslexia. J. Mem. Lang. 64 59–73. 10.1016/j.jml.2010.09.003 [ CrossRef ] [ Google Scholar ]
• Lyon G. R., Shaywitz S. E., Shaywitz B. A. (2003). A definition of dyslexia. Ann. Dyslexia 53 1–14. 10.1007/s11881-003-0001-9 [ CrossRef ] [ Google Scholar ]
• Lyons I. M., Nuerk H.-C., Ansari D. (2015). Rethinking the implications of numerical ratio effects for understanding the development of representational precision and numerical processing across formats. J. Exp. Psychol. Gen. 144 1021–1035. 10.1037/xge0000094 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• MacArthur C. A., Konold T. R., Glutting J. J., Alamprese J. A. (2010). Reading component skills of learners in adult basic education. J. Learn. Disabil. 43 108–121. 10.1177/0022219409359342 [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Martens V. E. G., de Jong P. F. (2006). The effect of word length on lexical decision in dyslexic and normal reading children. Brain Lang. 98 140–149. 10.1016/j.bandl.2006.04.003 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Martinez R., Fernandez A. (2010). The Social and Economic Impact of Illiteracy. Analytical Model and Pilot Study. Santiago: OREALC/UNESCO. [ Google Scholar ]
• Moeller K., Huber S., Nuerk H.-C., Willmes K. (2011). Two-digit number processing – holistic, decomposed or hybrid? A computational modelling approach. Psychol. Res. 75 290–306. 10.1007/s00426-010-0307-2 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Moeller K., Pixner S., Kaufmann L., Nuerk H.-C. (2009). Children’s early mental number line: logarithmic or rather decomposed linear? J. Exp. Child Psychol. 103 503–515. 10.1016/j.jecp.2009.02.006 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Morais J., Cary L., Alegria J., Bertelsen P. (1979). Does awareness of speech as a sequence of phones arise spontaneously? Cognition 7 323–331. 10.1016/0010-0277(79)90020-9 [ CrossRef ] [ Google Scholar ]
• Nickel S. (2007). “Familienorientierte grundbildung im sozialraum als schlüsselstrategie zur breiten teilhabe an literalität,” in Literalität, Grundbildung oder Lesekompetenz? Beiträge zu einer Theorie-Praxis-Diskussion eds Grotlüschen A., Linde A. (Münster: Waxmann; ) 31–41. [ Google Scholar ]
• Nuerk H.-C., Patro K., Cress U., Schild U., Friedrich C. K., Goebel S. M. (2015). How space-number associations may be created in preliterate children: six distinct mechanisms. Front. Psychol. 6 : 215 10.3389/fpsyg.2015.00215 [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• OECD (2013). OECD Skills Outlook 2013. First Results from the Survey of Adult Skills. Paris: OECD. [ Google Scholar ]
• OECD and Statistics Canada (2000). Literacy in the Information Age. Final Report of the International Adult Literacy Survey. Paris: OECD. [ Google Scholar ]
• Ostrosky-Solis P., Ardila A., Rosselli M. (1999). NEUROPSI: a brief neuropsychological test battery in Spanish with norms by age and educational level. J. Int. Neuropsychol. Soc. 5 413–433. 10.1017/S1355617799555045 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Petersson K. M., Reis A., Askelöf S., Castro-Caldas A., Ingvar M. (2000). Language processing modulated by literacy: a network analysis of verbal repetition in literate and illiterate subjects. J. Cogn. Neurosci. 12 364–382. 10.1162/089892900562147 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Ramus F., Rosen S., Dakin S. C., Day B. L., Castellote J. M., White S., et al. (2003). Theories of developmental dyslexia: insights from a multiple case study of dyslexic adults. Brain 126 841–865. 10.1093/brain/awg076 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Reis A., Castro-Caldas A. (1997). Illiteracy: a cause of biased cognitive development. J. Int. Neuropsychol. Soc. 3 444–450. [ PubMed ] [ Google Scholar ]
• Reis A., Faisca L., Ingvar M., Petersson K. M. (2006). Color makes a difference: two-dimensional object naming in literate and illiterate subjects. Brain Cogn. 60 49–54. 10.1016/j.bandc.2005.09.012 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Reis A., Guerreiro M., Petersson K. M. (2003). A sociodemographic and neuropsychological characterization of an illiterate population. Appl. Neuropsychol. 10 191–204. 10.1207/s15324826an1004_1 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Rello L., Baeza-Yates R., Dempere-Marco L., Saggion H. (2013). Frequent words improve readability and short words improve understandability for people with dyslexia. Lecture Notes Comput. Sci. 8120 203–219. [ Google Scholar ]
• Rimrodt S. I., Clements-Stephens A. M., Pugh K. R., Courtney S. M., Gaur P., Pekar J. J., et al. (2009). Functional MRI of sentence comprehension in children with dyslexia: beyond word recognition. Cereb. Cortex 19 402–413. 10.1093/cercor/bhn092 [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Rose L. T., Rougani P. (2012). Influence of verbal working memory depends on vocabulary: oral reading fluency in adolescents with dyslexia. Mind Brain Educ. 6 1–9. 10.1111/j.1751-228X.2011.01135.x [ CrossRef ] [ Google Scholar ]
• Rosselli M., Ardila A., Rosas P. (1990). Neuropsychological assessment in illiterates II.: language and Praxic Abilities. Brain Cogn. 12 281–296. 10.1016/0278-2626(90)90020-O [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Rüsseler J., Becker P., Johannes S., Münte T. F. (2007). Semantic, syntactic, and phonological processing of written language in adult developmental dyslexic readers: an event-related brain potential study. BMC Neurosci. 8 : 52 10.1186/1471-2202-8-52 [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Rüsseler J., Boltzmann M., Menkhaus K., Aulbert-Siepelmeyer A. (2013). Evaluation eines neuen trainingsprogramms zur verbesserung der lese- und rechtschreibfähigkeiten funktionaler analphabeten. Empirische Sonderpädagogik 3 237–249. [ Google Scholar ]
• Rüsseler J., Gerth I., Boltzmann M. (2011). “Basale wahrnehmungsfähigkeiten von erwachsenen funktionalen analphabeten und analphabetinnen,” in Lernprozesse in Alphabetisierung und Grundbildung Erwachsener. Diagnostik, Vermittlung, Professionalisierung ed. Projektträger im DLR e.V (Bielefeld: WBV-Verlag; ) 11–28. [ Google Scholar ]
• Rüsseler J., Menkhaus K., Aulbert-Siepelmeyer A., Gerth I., Boltzmann M. (2012). “Alpha Plus”: an innovative training program for reading and writing education of functionally illiterate adults. Creat. Educ. 3 357–361. 10.4236/ce.2012.33056 [ CrossRef ] [ Google Scholar ]
• Shi Y., Tsang M. C. (2008). Evaluation of adult literacy education in the United States: a review of methodological issues. Educ. Res. Rev. 3 187–217. 10.1016/j.edurev.2007.10.004 [ CrossRef ] [ Google Scholar ]
• Siegler R. S., Opfer J. E. (2003). The development of numerical estimation: evidence for multiple representations of numerical quantity. Psychol. Sci. 14 237–243. 10.1111/1467-9280.02438 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Silva C., Faisca L., Ingvar M., Petersson K. M., Reis A. (2012). Literacy: exploring working memory system. J. Clin. Exp. Neuropsychol. 1 1–9. [ PubMed ] [ Google Scholar ]
• Simmons F. R., Singleton C. (2006). Arithmetic abilities of adults with dyslexia. Dyslexia 12 96–114. 10.1002/dys.312 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Smith-Spark J. H., Henry L. A., Messer D. J., Edvardsdottir E., Ziecik A. P. (2016). Executive functions in adults with developmental dyslexia. Res. Dev. Disabil. 53 323–341. 10.1016/j.ridd.2016.03.001 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Statistics Canada OECD. (2005). Learning a Living. First Results of the Adult Literacy and Life Skills Survey. Paris: OECD. [ Google Scholar ]
• Suarez-Coalla P., Ramos S., Alvarez-Canizo M., Cuetos F. (2014). Orthographic learning in dyslexic Spanish children. Ann. Dyslexia 64 166–181. 10.1007/s11881-014-0092-5 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Thompkins A. C., Binder K. S. (2003). A comparison of the factors affecting reading performance of functional illiterate adults and children matched by reading level. Read. Res. Q. 38 236–258. 10.1598/RRQ.38.2.4 [ CrossRef ] [ Google Scholar ]
• Thorn W. (2009). International Adult Literacy and Basic Skills Surveys in the OECD Region. OECD Education Working Papers 26 Paris: OECD Publishing. [ Google Scholar ]
• UNESCO (1978). Records of the General Conference. 20th Session Vol. 1 Paris: UNESCO. [ Google Scholar ]
• UNESCO (2006). “Why Literacy Matters,” in Education for All. Literacy for Life ed. UNESCO (Paris: UNESCO Publishing; ) 135–145. [ Google Scholar ]
• UNESCO (2009). The Next Generation of Literacy Statistics: Implementing the LITERACY ASSESSMENT and Monitoring Programme (LAMP). Paris: UNESCO Institute for Statistics Montreal. [ Google Scholar ]
• UNESCO (2013). Schooling for Millions of Children Jeopardized by Reductions in Aid. Education for All Global Monitoring Report. Paris: UNESCO Institute for Statistics. [ Google Scholar ]
• UNESCO (2015). Adult and Youth Literacy. UIS Fact Sheet. Paris: UNESCO Institute for Statistics. [ Google Scholar ]
• Vajjala S., Meurers D. (2014). Readability assessment for text simplification: from analyzing documents to identifying sentential simplifications. Int. J. Appl. Linguist. 165 194–222. [ Google Scholar ]
• Valdois S., Bosse M.-L., Tainturier M.-J. (2004). The cognitive deficits responsible for developmental dyslexia: review of evidence for a selective visual attentional disorder. Dyslexia 10 339–363. 10.1002/dys.284 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Van Linden S., Cremers A. H. M. (2008). “Cognitive abilities of functionally illiterate persons relevant to ICT use,” in Computers Helping People with Special Needs eds Miesenberger K., Klaus J., Zagler W., Karshmer A. (Heidelberg: Springer-Verlag; ) 705–712. [ Google Scholar ]
• Varvara P., Varuzza C., Sorrentino A. C. P., Vicari S., Menghini D. (2014). Executive functions in developmental dyslexia. Front. Hum. Neurosci. 8 : 120 10.3389/fnhum.2014.00120 [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• White S., Milne E., Rosen S., Hansen P., Swettenham J., Frith U., et al. (2006). The role of sensorimotor impairments in dyslexia: a multiple case study of dyslexic children. Dev. Sci. 9 237–269. 10.1111/j.1467-7687.2006.00483.x [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Willburger E., Fussenegger B., Moll K., Wood G., Landerl K. (2008). Naming speed in dyslexia and dyscalculia. Learn. Individ. Differ. 18 224–236. 10.1016/j.lindif.2008.01.003 [ CrossRef ] [ Google Scholar ]
• Willcutt E. G., Petrill S. A., Wu S., Boada R., DeFries J. C., Olson R. K., et al. (2013). Comorbidity between reading disability and math disability: concurrent psychopathology, Functional Impairment, and Neuropsychological Functioning. J. Learn. Disabil. 46 500–516. 10.1177/0022219413477476 [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Wilson A. J., Andrewes S. G., Struthers H., Rowe V. M., Bogdanovic R., Waldie K. E. (2015). Dyscalculia and dyslexia in adults: cognitive bases of comorbidity. Learn. Individ. Differ. 37 118–132. 10.1016/j.lindif.2014.11.017 [ CrossRef ] [ Google Scholar ]
• Wiseheart R., Altmann L. J. P., Park H., Lombardino L. J. (2009). Sentence comprehension in young adults with developmental dyslexia. Ann. Dyslexia 59 151–167. 10.1007/s11881-009-0028-7 [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Ziegler J. C., Pech-Georgel C., Dufau S., Grainger J. (2010). Rapid processing of letters, digits and symbols: what purely visual-attentional deficit in developmental dyslexia? Dev. Sci. 13 8–14. 10.1111/j.1467-7687.2010.00983.x [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Zoubrinetzky R., Bielle F., Valdois S. (2014). New insights on developmental dyslexia subtypes: heterogeneity of mixed reading profiles. PLoS ONE 9 : e99337 10.1371/journal.pone.0099337 [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]

• Share full article

For more audio journalism and storytelling, download New York Times Audio , a new iOS app available for news subscribers.

Supported by

The Ezra Klein Show

Transcript: Ezra Klein Interviews Nilay Patel

Every Tuesday and Friday, Ezra Klein invites you into a conversation about something that matters, like today’s episode with Nilay Patel. Listen wherever you get your podcasts .

Transcripts of our episodes are made available as soon as possible. They are not fully edited for grammar or spelling.

Will A.I. Break the Internet? Or Save It?

Nilay patel discusses the near-future of an internet as a.i.-generated content improves..

[MUSIC PLAYING]

From New York Times Opinion, this is “The Ezra Klein Show.”

Earlier this week, we did an episode on how to use A.I. right now. Now, I want to turn the question around and look at how A.I. is being used on you right now. One of the conversations has been sticking in my head was with this person in the A.I. world who was saying to me that if you look at where use has been sticky, if you look at where people keep using it day after day, you’re looking at places where the product doesn’t need to be very good. That’s why it’s really helpful for college and high school students, college and high school papers — they’re often not very good. That’s sort of their point. It’s why it’s working pretty well for a very low-level coding tasks. That kind of work doesn’t need to be very good. It gets checked and compiled, and so on.

But there’s something else that it is working really well for, which is spewing mediocre content onto the internet. And the reason is that a lot of what is on the internet right now isn’t very good. Its point is not to be good — spam isn’t very good, marketing emails aren’t very good, social media bots aren’t very good. Frankly, a lot of social media posters even when they’re not bots are not very good.

There are all kinds of websites and internet operations that are filler content designed to give search engines something to index — filler content structured to do well in a Google result so people click on it and then see an ad.

Something you’re going to hear a lot of in this episode is the term S.E.O., and that is what we’re talking about: Search Engine Optimized. Things that are built to rank highly in Google and Bing just to get somebody to click on the website. It doesn’t always matter to that person if they read the website.

But into this comes A.I. Over the last year, Google and the big social platforms — they have been flooded with A.I. spam, flooded with fake news sites filled with stolen or made up stories. There are TikToks of A.I. voices reading random text off of Reddit, nonsensical YouTube videos for kids. It’s no novel observation to say the internet has felt like it is in a state of decay for a while.

Google search results, Facebook, Twitter, or X, YouTube, TikTok — all of it felt better, more human, more delightful, more spontaneous, more real a few years ago. So what happens when this flood of content hits this decaying internet?

And then — and I actually think this is the harder, weirder question — what happens when this flood of A.I. content gets better? What happens when it doesn’t feel like garbage anymore? What happens when we don’t know if there’s a person on the other end of what we’re seeing or reading or hearing?

Should we care? What if that content is actually better than a lot of what we’re getting right now? Is that an internet we want to be on or not?

My friend Nilay Patel is the co-founder and editor in chief of the tech news site The Verge, and host of the great “Decoder” podcast. And I got to be honest, I can’t tell from this conversation if Nilay is more or less optimistic than me because he seems to think A.I. is going to break the internet. But he seems kind of happy about it.

Before we get into the actual conversation here, we are nominated for a Webby — speaking of hopefully good things on the internet — in the Best Interview Talk Show category. We are up against Oprah here, so we are decided underdogs, but this is a voting category so if we’re going to win, we need your help. You can vote using the link in the show notes or go to vote.webbyawards.com

And as always, if you want to email me with guest suggestions or thoughts on the episode, that is [email protected].

Nilay Patel, welcome to the show.

Thank you for having me. This is very exciting.

Let’s just begin with the big question here, which is what is A.I. doing to the internet right now?

It is flooding our distribution channels with a cannon-blast of — at best — C+ content that I think is breaking those distribution channels.

Why would it break them?

So most of the platforms the internet are based on the idea that the people using those platforms will in some sort of crowdsourced way find the best stuff. And you can disagree with that notion. I think maybe the last 10 years have proven that that notion is not percent true when it’s all people.

When you increase the supply of stuff onto those platforms to infinity, that system breaks down completely. Recommendation algorithms break down completely, our ability to discern what is real and what is false break down completely, and I think importantly, the business models of the internet break down completely. So if you just think about the business model of the internet as — there’s a box that you can upload some content into, and then there’s an algorithm between you and an audience, and some audience will find the stuff you put in the box, and then you put an infinity amount of stuff into the box, all of that breaks.

My favorite example of this is Amazon, which allows people to self-publish books. Their response to the flood of A.I. generated books was to limit the number of books you can upload to three books in a day. This is really — like that’s a ridiculous response to this. It just implies that the systems that we’ve built to organize audiences and deliver the right thing to the right person at the right time, they’re not capable of an increase in supply at the level that A.I. is already increasing this.

Thank you for bringing in the supply language. So, I’ve been trying to think about this as this supply and demand mismatch. We have already had way more supply than there is demand. I wasn’t buying a lot of self-published Amazon books. Is the user experience here actually different?

I think that’s a great question. The folks who write the algorithms, the platforms, their C.E.O.s, they will all tell you this is just a new challenge for us to solve. We have to out what is human, what is A.I.-generated. I actually think the supply increase is very meaningful. Like, maybe the most meaningful thing that will happen to the internet because it will sort out the platforms that allow it to be there and have those problems, and the places that don’t. And I think that has not been a sorting that has occurred on the internet in quite some time, where there’s two different kinds of things.

The example that I’ll give you is, every social media platform right now is turning into a short-form video Home Shopping Network. LinkedIn just added short form videos. Like, they’re all headed towards the same place all the time because they all have the same pressures.

Didn’t we already pivot to video a couple years ago?

We pivoted to video — I actually love it when LinkedIn adds and takes away these features that other platforms have. They added stories because Snapchat and Instagram had stories, and they took the stories away because I don’t think LinkedIn influencers want to do Instagram Reels, but now they’re adding it again.

And what you see is those platforms, their product — the thing that makes them money — is advertising, which is fine. But they don’t actually sell anything in the end. They sell advertising. Someone else down the line has to make a transaction. They have to buy a good or a service from someone else. And if you don’t have that, if you’re just selling advertising that leads to another transaction, eventually you optimize the entire pipe to the transaction to get people to buy things, which is why TikTok is now — like all of TikTok is TikTok Shop, because they just want you to make a transaction. And that those platforms are going to be most open to A.I., because that is the most optimizable thing to get people to make a transaction. And I think real people will veer away from that.

So I want to hold on to something that you’re getting at here. Which, to me, is one of the most under-discussed parts of A.I., which is how do you actually make money off of it? And right now, there are not actually that many ways.

So, what you can do is you can pay some money to the big A.I. companies. So you get the pro-version of their models. There is a certain amount of enterprise software flying around. You can subscribe to versions of Microsoft Copilot, or there’s going to be more things like that, where you can subscribe to something that is supposed to get you to buy the next iteration of Slack or whatever the enterprise software is. But it is hard to not notice that a lot of the A.I. is being built by companies that exist on advertising.

Google has a huge A.I. program, Meta has a huge A.I. program, and advertising is fundamentally a persuasion game. They are trying to persuade you to do something with the advertising to buy something. And right now, it’s pretty bad. I always think it’s funny how long after I make a significant purchase I will be advertised to make that purchase again.

It’s like, you just bought a fair amount of luggage, would you like any more luggage from the same company you already bought it from? It’s a very weird — but if this gets good, what is that? What are safe business models and what are very unethical ones, because when we talk about harms and benefits from A.I., how people are making money off of it is going to be a pretty big intermediary there.

Yeah, I’ve been talking to a lot of C.E.O.s of web companies and email companies on Decoder for the past year. I asked them all the same question, why would you start a website? Why would you send an email? And so, you asked the C.E.O. of Squarespace or Wix or we just had the C.E.O. of MailChimp on the show. And her answer is a little terrifying. Like, maybe openly terrifying.

She’s like well collect enough data on you, and then we’ll know exactly when to send you an email so that you buy the right thing at the right time. And we’ll just have A.I. automate that whole process. So you come to the website for your local dry cleaner or luggage store, you type in your email address to get the 10 percent off coupon, we look at what you were looking at. And then somewhere down the line when some other data broker has told us that you searched for a flight, we will send you a precisely targeted generated email that says you’re going to Paris? Buy this suitcase that matches your style from our store at this dynamically generated price.

But how is A.I. changing that at all because that sounds to me like the thing that is already happening.

So, this is what I mean by the increase in scale. That’s the dream. This is supposed to be what actually happens, but they can only do it in broad cohorts, which is why you get the luggage email after you’ve bought the luggage email or the luggage ad, after you bought the luggage ad.

They know you are a person who used a Wi-Fi network in a certain location at a certain time, they can track that all over the place. They know what you’ve searched for. They know that you went and made a luggage transaction. You are now categorized into people who are likely to buy luggage, whether or not that loop was closed. You put some luggage in a shopping cart. But that’s still a cohort, they can only do that broadly. And these cohorts can be pretty refined, but they can only do it broadly. With A.I. the idea is we can do that to you individually — the A.I. will write you an email, we’ll write you a marketing message, will set you a price. That isn’t 100x increase the amount of email that will be generated.

So now our email algorithms will be overflooded with commercial pitches generated by A.I. And this sort of makes sense, right? It makes sense for a Google to want to be able to dynamically generate A.I. advertising across the entire web. It makes sense for Meta to invest massively in A.I. so that when you’re watching Instagram and you scroll a dynamically generated Instagram video, that is an ad just for you appears. And all of that is down to their belief in targeting — their absolute belief that they can sell more products for their clients by targeting the ads more directly. And you are in that uncanny valley, where the targeting doesn’t actually work as well as it should and no one will admit it.

When I get spammy advertising I don’t really think about there being a human on the other end of it. Maybe to some degree there is, but it isn’t part of the transaction happening in my head. There are a lot of parts of the internet that I do think of there being a human on the other end — social media, reviews on Amazon, books — I assume the person who wrote the book is a person. How much of what I’m currently consuming may not be done by human in the way I think it is, and how much do you think that’s going to be in a year, or two, or three years?

I’m guessing your media diet is pretty well human-created because I know that you are very thoughtful about what you consume and what signals you’re sending to the algorithms that deliver your content. I think for most people —

My mom’s, let’s use my mom’s.

Mom’s are good. I would love to take my mom’s phone and throw it into the ocean and never let her have it again. I openly fear what content comes through my mother through WhatsApp. It terrifies me that I don’t have a window into that. I can’t monitor it. The same software I want to use to watch my daughter’s internet consumption, I would love to apply it to my parents because I don’t think they have the media literacy — they’re much older — to even know, OK, this might be just some A.I.-generated spam that’s designed to make me feel a certain way.

And I think that is the heart of what’s coming. I think right now it’s higher than people think, the amount of A.I. generated noise, and it is about to go to infinity. And the products we have to help people sort through those things, fundamentally our intention with that. Google is the heart of this tension — you can take any business at Google and say what happens when the A.I. flood comes to you? And I don’t think they’re ready for it.

How can they not be ready for that?

Because they’re the ones making it. This is the central tension of — in particular, I think Google. So, Google depends on the web, the richness of the web is what Sundar Pichai will tell you. He used to run search, he thinks about the web. He cares about it, and you look at the web and you’re like, you didn’t make this rich at all. You’ve made this actually pretty horrible for most people most of the time. Most people — if you search Google to get a credit card, that is a nightmarish experience — like, fully nightmarish. It feels like getting mugged.

We just went on vacation. And I googled a restaurant review in Cancun, and I got about halfway through the actual review when I realized it was sponsored content by Certified Angus Beef. And just in the middle of this review, they’re like this restaurant uses this kind of beef and here’s why it’s great. And I was like — this is — I read an ad. And Google should have told me that this was an ad. Like, this isn’t useful to me in any way — like, I’m discarding this. I don’t want this anymore.

I don’t think Google can discern what is good or bad about the web. I don’t think Google has reckoned with how it’s incentives have shaped the web as a whole. And I certainly don’t think that people who are making Google search can say A.I. is bad — A.I. content is bad, because the whole other part of Google that is making the A.I. content can’t deal with that.

This helps explain a story that I found very strange. So, 404 Media, which is a sort of newer outlet reporting on tech. They found that Google News was boosting stolen A.I. versions of news articles — and we’re seeing this all over. An article by me or by some other journalist shows up in another place, very slightly rewritten by an A.I. system, with an A.I. generated author and photo on top of it. So, we’re seeing a lot of this.

And when 404 Media asked Google about this, Google News said that for them, it was not a really relevant question whether an article was by an A.I. or a human. That struck me as a very strange thing to say, to admit. Is your view that it’s because their business is in the future replacing human-generated content with A.I., and saying that’s good — like, that’s the thing happening at the center there?

Yeah. Fundamentally, I think if you are at Google and the future of your stock price depends on Gemini being a good competitor to GPT-4 or 5 or whatever OpenAI has, cannot run around saying this is bad. The things it makes are bad.

I think this is actually in stark contrast to how people feel about that right now. One of the funniest cultural trends of the moment is that saying something is A.I.-generated is actually a great way to say it’s bad.

So, I saw people reacting to the cover of the new Beyoncé album, “Cowboy Carter,” which is a picture of her on a stunning horse. It’s Beyoncé, it’s very obviously human made, and people don’t like it. Like, was this made by A.I.? And it’s like well, you know for a fact that Beyoncé did not have A.I. generate the cover of — like, you can look at it and you can discern that it isn’t. But you can say, was this A.I.-generated? And that is code for this is bad.

What about when it’s not?

I don’t know how fast that is coming. I think that is farther away than people think. I think ‘will it fool you on a phone screen?’ is here already, but ‘is this good’ is, I think, farther away than —

But a lot of internet content is bad.

That’s fair.

I mean, you know this better than me. Look, I think it is axiomatic that A.I. content is worse right now than it will ever be.

I mean the advance in image generation over the past year has been significant. That’s very real. And preparing for this conversation, I found myself really obsessing over this question, because one way to talk to you about this is, there’s all this spammy garbage coming from A.I. that is flooding the internet.

But you can imagine an A.I. developer sitting in the third chair here and saying, yeah sure, but eventually it’s not going to be spammy garbage. We’re getting better at this. And compared to what people are getting from a lot of websites, if you’re going to Quora or ask.com or parts of Reddit or whatever, we can do better than that. The median article within three years is going to be better than the median human-produced piece of content.

And I really — I found that I did not know how to answer the question in myself — is that a better or a worse internet? To take almost Google’s side on this, should it matter if it’s done by a human or an A.I., or is that some kind of — what’s the word — like, sentimentality on my part?

I think there’s a sentimentality there. If you make a content farm that is the best content farm, that has the most answers about when the Super Bowl starts, and those pages are great. I think that’s a dead end business. Google is just going to answer the questions. I think that’s fine. I think if you ask Google what time the Super Bowl is, Google should just tell you. I think if you ask Google how long to boil an egg, Google can just tell you. You don’t need to go to some web page laden with ads and weird headings to find those answers. But these models in their most reductive essence are just statistical representations of the past. They are not great at new ideas.

And I think that the power of human beings sort of having new ideas all the time, that’s the thing that the platforms won’t be able to find. That’s why the platforms feel old. Social platforms like enter a decay state where everyone’s making the same thing all the time. It’s because we’ve optimized for the distribution, and people get bored and that boredom actually drives much more of the culture than anyone will give that credit to, especially an A.I. developer who can only look backwards.

I’m going to spend some time thinking about the idea that boredom is an under-discussed driver of our culture. But I want to get at something else in there — this idea of Google answering the question. We’re already seeing the beginnings of these A.I. systems that you search the question that might — at another time — have brought you to The Verge, to CNN, to The New York Times, to whatever.

But now, perplexity — there’s a product, Arc. They’ll basically use A.I. to create a little web page for you. The A.I. itself will read, “read”— in quotation marks — the A.I. itself will absorb some websites, create a representation of them for you, and you’ll never go to the place you were that actually created that data about the past that A.I. used to give you something in the present.

Casey Newton, at Platformer, his word was he felt revulsion, and that was how I felt about Arc’s product here. You take all this work other people have done, you remix it under your thing, they don’t get the visit to their web page, nobody has the experience with the work that would lead them to subscribe. But two things in the long run happen from that.

One is that you destroy the score of growing value, growing informational value that you need to keep the internet healthy. You make it say impossible to do the news gathering that allows you to be news because there’s no business model for it. The other is that you also destroy the training data for the A.I. itself, because it needs all that work that we’re all doing to train.

The thing they need is data. The A.I. is polluting that data with A.I. content currently, but it also can begin to destroy that data by making it unprofitable for people to create more of it in the future. I think Ryan Broderick has called A.I. search a doomsday cult. How do you think about this sort of deeper poisoning of the informational commons?

I think there’s a reason that the A.I. companies are leading the charge to watermark and label content as A.I.-generated. Most of them are in the metadata of an image. So most pictures you see in the internet, they carry some amount of metadata that describes the picture. What camera was taken on, when it was taken, what image editing software was used.

So, Adobe and a bunch of other companies are like, we’ll just add another field that says, here are all the A.I.-generated edits that were made on this photo. I think it is in their self-interest to make sure that is true and they can detect it and exclude it if they need to. I think there are moral reasons to do it too.

So their training data remains less corrupted?

Yeah. I think there’s a very straightforward incentive for them to figure out the watermarking, labeling stuff they want to do. And they have coalitions, and tasks force, and Adobe talks about the image of the Pope and the puffer jacket as a, “catalyzing moment” for the metadata of A.I. because people freaked out. They’re like oh, this thing looks real. But they have a real incentive to make sure that they never train on other A.I. generated content.

So that’s one aspect, which I think is just sort of immediately self-interested. The other thing is — that’s why I keep asking people why would anyone make a web page?

There’s a site I think about all the time. It’s called HouseFresh, which is a site that only reviews air purifiers. And to me, this is the internet. Like, this is what the internet is for. You care about air purifiers so much you’ve set up a series of web pages where you express your expertise in air purifiers and tell people which ones to buy. That’s all they do. And Google has started down-ranking them, because big publishers boost their content, because A.I. is lifting their content, because companies like CNN, in order to gain some affiliate ad revenue somewhere, have set up their own little mini-content farms full of affiliate links.

I’m not saying we don’t — like, other publishers do this. But the point of these algorithms is, ideally, to bring you to the HouseFresh people, is to bring you to the person who cares so much about air purifiers they made a website about air purifiers, and we’re not doing that anymore. And so if you were to say, where should a young person who cares the most about cars, or who cares the most about coffee, or whatever. Where are they going to go? Where are they going to make stuff? They’re going to pick a closed platform that ideally offers them some built in monetization, that ideally offers them some ability to connect directly with an audience. They’re not going to go to a public space like the web, where they might own their own business, which would be good. But they’re also basically at the mercy of thieves who come in the night and take all their work away.

But also, if you kill HouseFresh, then two years later when you ask the A.I. what air purifier should I get, how does it know what to tell you?

Yeah, I don’t the answer to that question.

I don’t think they do either.

Yeah again, this is why I think that they are so hell-bent on labeling everything. I think they need some people around in the future.

But labeling is good. I mean, that keeps you from getting too much garbage in your data set. But replacing a bunch of the things that the entire informational world relies on to subsidize itself — to fund itself — like this to me is a thing that they don’t have an answer for.

Wait, let me ask you a harder question. Do they care?

Depends on they, but I don’t think so.

Or at least they care in the way that I came to realize Facebook, now Meta, cared about journalism. People say they didn’t care about journalism. I don’t believe that’s actually true. They didn’t care enough for it to mean anything. Like, if you asked them, if you talked with them, if you had a drink, they would think what was happening to journalism was sad.

And if it would cost them nothing, they would like to help. But if it would cost them anything — or forget costing them anything. If they would begin to help and then recognize an opportunity had been created that they could take instead of you, they would do that. That’s the way they care.

So when you have a financial crisis, you have something oftentimes called a flight to quality. Investors flood into the things they know they can trust, usually treasury bonds, and I’ve been wondering if this won’t happen in this era of the internet — if I wanted to take an optimistic perspective on it — that as you have a sort of ontological collapse, as you don’t know what anything is.

I already feel this way with product reviews. When I search product reviews, I get reviews now from tons of sites that I know don’t really invest that much in product reviews. CNN, all these other organizations that I have not really, truly invested in high-quality product reviewing, when you search, you now get them — they’re telling you what to buy.

That makes me trust the Wirecutter, which is a New York Times property, but that I know we’ve put a lot of money in more. Similarly, the other one I use, which is a Vox Media property, is The Strategist at New York, because I knew what the development of that looked like, I know what they put into that.

You can imagine this happening in news for things like The New York Times or The Washington Post. You can imagine it in a couple of different places. If people begin to feel that there is a lie at the heart of the internet they’re being given, that they can’t figure out what is what and who is who and if it is a who at all — I mean, maybe you just end up in this internet where there’s more of a value on something that can be verified.

I keep a list of TikToks that I think each individually should be a Ph.D. thesis in media studies. It’s a long list now. And all of them are basically just layers of copyright infringement in their own weird way.

My favorite is — it’s a TikTok, it has millions of views. It’s just a guy reading a summary of an article in the journal Nature. It has millions of views.

This is more people that have ever considered any one article in the journal Nature — which is a great journal. I don’t mean to denigrate it. It’s a proper scientific journal. They work really hard on it. And you just go 5 steps down the line, and there’s a guy on TikTok summarizing a summary of Nature, and you’re like what is this? What is this thing that I’m looking at?

Will any of the million viewers of this TikTok buy one copy of Nature because they have encountered this content? Why did this happen?

And the idea is, in my mind at least, that those people who curate the internet, who have a point of view, who have a beginning and middle, and an end to the story they’re trying to tell all the time about the culture we’re in or the politics we’re in or whatever. They will actually become the centers of attention and you cannot replace that with A.I.

You cannot replace that curatorial function or that guiding function that we’ve always looked to other individuals to do.

And those are real relationships. I think those people can stand in for institutions and brands. I think the New York Times, you’re Ezra Klein, a New York Times journalist means something. It appends some value to your name, but the institution has to protect that value.

I think that stuff is still really powerful, and I think as the flood of A.I. comes to our distribution networks, the value of having a powerful individual who curates things for people, combined with a powerful institution who protects their integrity actually will go up. I don’t think that’s going to go down.

You mentioned 404 Media. 404 Media is a bunch of journalists who were at Motherboard at Vice. Vice is a disaster. They quit, they started a new media company, and we now all talk about 404 Media all the time. This thing is 25 minutes old. We don’t talk about Jason Koebler the editor in chief. We talk about 404 Media, the institution that they made — a new brand that stands for something, that does reporting and talks about something. I think there’s still meaning there.

You said something on your show that I thought was one of the wisest, single things I’ve heard on the whole last decade and a half of media, which is that places were building traffic thinking they were building an audience. And the traffic, at least in that era, was easy, but an audience is really hard. Talk a bit about that.

Yeah first of all, I need to give credit to Casey Newton for that line. That is something — at The Verge, we used to say that to ourselves all the time just to keep ourselves from the temptations of getting cheap traffic. I think most media companies built relationships with the platforms, not with the people that were consuming their content.

They didn’t think about them very much. They thought about what was hitting in the Facebook algorithm, they thought about what Google search wanted for Game of Thrones coverage that day, which was everything all the time. And everybody had a Game of Thrones program. Fox had one, The Verge had one, The New York Times had one. Why?

That’s weird. It’s we constructed this artificial phenomenon because people searched for — I mean, just to say the answer because we know it — because people searched for “Game of Thrones” content the morning after the show, and that was an easy way to get a bunch of traffic. And at least a theory of the time was that you could turn traffic into money through advertising, which was not totally wrong, but not nearly as right as the entire era of business models was predicated on.

The other thing that those business models were predicated upon was you’d get so good at being a supplier to one platform or another with Game of Thrones content or whatever it was that they would pay you money for it directly — that Google would say, this is the Game of Thrones link that most people are clicking on. We ought to pay Vanity Fair for its Game of Thrones content to surface it. Or all of BuzzFeed was we’re going to be so good at going viral on Facebook that Facebook will pay us money.

And that absolutely didn’t pan out. But no one hedged that bet, which is utterly bananas to me. No one said we should take these people who came here for a Game of Thrones and figure out how to make them care about us, and we should care about them. Everyone just looked at it as a number that was going up against some amount of interest as demonstrated by some platform somewhere.

And I think that is the mistake. It is the mistake that creators on the creator platforms are not making, because the terms of that arrangement are so much more cynical. You see TikTokers. They at any moment their videos can get downranked, their accounts can get yanked, their stuff can get banned. They’re constantly trying to get you to go to Instagram.

Every YouTuber gets their wings when they make the video about how they’re mad at YouTube. There’s a woodworking YouTuber that I used to follow, and he just sort of got to the point where he’s like, I hate YouTube. I’m leaving. And it’s like dude, you made videos about jointing wood, like what are you doing?

And it’s like his relationship with the platform was so cynical that he was like, I’m moving my business elsewhere. You can sign up for a master class. Those individuals have these very cynical, very commercial relationships with the platforms that the media companies, for some reason, just never hedged. And so they actually do have audiences. And I think media companies need to get way back on the game of having a true audiences.

This gets to something that does worry me about this phase of A.I. hitting the internet, which is it’s hitting an internet in a moment of decay and weakness. And here, by internet, I mean the sort of content generating internet, and I break that into a couple of categories. The media is very weak right now. The media business we have seen closures left and right, layoffs left and right. I mean, a bunch of players like Vice and BuzzFeed who were believed to be the next generation of juggernauts are functionally gone as news organizations.

The big content platforms, they’re doing fine from a financial standpoint, but people hate them. The relationship between the users and Facebook, the users and YouTube, the users and — to some degree, you’re even seeing that now with TikTok — is just darkening in a way that it wasn’t in 2014.

And so, there’s a lot of desperation on all sides. Sometimes the desperation is you don’t have the money to pay the journalists you need to do the work you want to do. Sometimes the desperation is that you’re trying to figure out something to make this audience like you again and not get eaten by TikTok or whatever comes after TikTok.

And into this comes A.I., and all the money that A.I. seems to bring, and even the A.I. companies might pay you some money for your stuff.

Reddit just licensed a bunch of its content as training data to Google.

So, you could really imagine a thing happening again, where all these media companies or content companies of some form or another, license out what they have for pennies on the dollar, because at least you can make some money off of it that way.

But what worries me is both the weakness, but that also, it does not feel to me like anybody knows what the relationship is to this is supposed to be. Do you use it? Are you just training data for it? Like, what are you in relationship to the A.I. era?

As a consumer or as a producer?

As a producer.

The idea that media companies are going to license their stuff to the A.I. companies is just the end of the road that we’ve been on for a long time. We are suppliers to algorithms. OK? And in any normal functioning capitalist economy, supplier margins get squeezed to zero and then maybe we all die. Like, that’s the game we’ve been playing without saying it for a long time —

Which I think is why you see The New York Times suing OpenAI, like a real desire to not be in that game again.

You see The New York Times suing OpenAI, but you don’t see them suing Google, you don’t see them de-S.E.O.ing pages across New York Times. Like, they still need the audience from these platforms. And I think there’s a very tense relationship there. The idea that you could sue OpenAI and win some precedent that gives you an enormous amount of leverage over Google I think is a very powerful idea.

Most of the media company executives I talk to would love for that to be the outcome. I don’t know if that’s going to be the outcome. I feel like I should warn your audience, like — I’m a failed copyright lawyer. I wasn’t good at it, but I did it for a minute. Copyright law is a coin flip. Like, these cases are true coin flips. They are not predictable. The legal system itself is not predictable, copyright law inherently is unpredictable.

And a really interesting facet of the internet we live in today is that most of the copyright law decisions were won by a young, upstart, friendly Google. YouTube exists because it was Google. Like, Viacom famously sued YouTube and they might have won and put it out of business, but Google, the friendly Google company with the water slides in the office, the upstarts that made the product you loved, went and won that case. Google Books, we’re going to index all the books without asking for permission. They won that case, because they were friendly Google, and the judges were like, look at these cute kids making a cool internet? Like it was new and novel. Google image search — these are all massive copyright decisions that Google won as a startup company run by young people building a new product that the judges were using on their Dell desktops or whatever.

These aren’t those companies anymore. They’re going to go into a legal system as behemoths, as some of the biggest, best-funded companies in the world that have done bad things to the judges teenage children, like all these things are different now. And so, I don’t know if Google, or OpenAI, or Microsoft gets the benefit of being like, we’re young and cool and hip, bend copyright law to our will.

You don’t want a staunch innovation. Like, that was the big fear in that era. We don’t know what we’re building, and that’s still the thing you hear, and it’s not even untrue. You crack down on copyright and maybe you do staunch innovation. You don’t crack down copyright and maybe you destroy the seed corn of the Informational Commons. It’s very fraught for the copyright judges, but also just for all of us.

Yeah, what are you as a producer on the internet is totally governed by copyright law. Like, a joke at The Verge is a copyright law is the only functional regulation on the internet. The entire internet is just speech, that’s all it is top-to-bottom, it’s speech.

In the United States, we don’t love a speech regulation, and I think for good reason. But we love copyright law, we love it. Can’t get enough of it. Like, YouTubers know the YouTube copyright system back and forth, because that’s the thing that takes their content down. And we allow this regulation on the internet at scale.

And so the parameters of this one body of law, as applied to A.I., which is a taking. Training an A.I. model is fundamentally a taking, and the A.I. company —

Taking in the legal sense of the term?

No, in the moral sense of the term. They come to your website and they take your stuff. It’s not a zero sum taking, but they’ve extracted value to create more value for themselves. I think that’s just a moral taking. There’s some permission there that did not occur. Joanna Stern at The Wall Street Journal just interviewed Mira Murati, the C.T.O. of OpenAI, about training data for Sora, the video generator, and Mira said, we just use what’s publicly available. And it’s like yo, that doesn’t make any sense. Like, there are lots of rules about what’s publicly available. Like, you can’t just take stuff because you can link to it on the internet, that’s not how it actually works.

Let me try to take the argument I hear from the A.I. side of this, which is that there is functionally nothing in human culture and human endeavor that is not trained on all that has come before it — that I, as a person, am trained on all this embedded knowledge in society, that every artist has absorbed, all this other art that the A.I. — I mean, this is just learning. And as long as you are transforming that learning into something else, as long as you are doing something new with that learning, then one, copyright law is not supposed to apply to you in some way or another, although that’s obviously complicated.

But two, to go back to your point of morality, if you want to see culture humanity technology advance, it is also not supposed to apply to you, because if you do not let things learn, people, organizations, models, you are not going to get the advances built on all that has come before. And that’s how we’ve always done it. What’s your answer to them?

I hear this idea all the time, often from the sorts of people in Silicon Valley who say they do first principles thinking — which is one of my favorite phrases, because it just means what if we learn nothing? Like, what if none of the history of the world applied to us and we could start over to our benefit? And that’s usually what that’s code for.

So I hear those arguments and I think, you guys just weren’t paying attention. You’re entering a zone where the debate has been raging for decades. A lot of copyright law is built around a controversy around player pianos, and whether player pianos would displace musicians. But you just have to rewind the clock to the 80s and be like, should sampling be legal in music?

And now we are having the exact same conversation in the exact same way with the exact same parameters. The only thing that’s different now is any kid can sample any song at scale, feed it into an A.I. and have Taylor Swift sing the Dolly Parton song for them. That’s a weird new turn in the same debate, but it is a massively age-old debate, and the parameters of the debate are pretty well known.

How do you incentivize new art? How do you make sure that it’s economically valuable to make new things? How do you make sure the distributors don’t gain too much power, and then how do you make sure that when people are building on the past, the people whose art they’re building on retain some value?

And that I think is — the A.I. companies have no answer to that last question. We’re just going to take a bunch of stuff and now we’re just going to say look, we just summarized the web. The people who made the web get nothing for that will pay us $20 a month for the service.

But somewhere in there, as a policy matter as a moral matter, the people who made the foundations of the work should get paid. And this is where the sampling debate has ended up. There’s a huge variety of licensing schemes and sample clearances so that those artists get paid.

Judge Patel, if you’re thinking about cases in this area, like, what do you think the answer is here? Is it the sampling model, is it something else? What do you think the right broad strokes resolution is?

Let me stick on the music example for one second, because I think music is really interesting because it’s kind of a closed ecosystem. There’s only so many big music companies. It’s the same lawyers, and the same executives, and the same managers going to the same clearing houses and having the same approaches. We’re going to give you a songwriting credit because we interpolated the bass line of this song into that song, and now here’s some money. And this is the mechanism by which we’ll pay you. The A.I. companies are not a closed ecosystem, it is just a free for all. It’s the open web, it’s a bunch of players.

So, I think in those cases, you’re just going to end up with vastly more outcomes which I think leads to even more chaos, because some companies will take the deal. I’m guessing The New York Times is going to pursue this all the way to the Supreme Court. This is an existential issue for The Times.

Some companies don’t have the money to pay for Supreme Court litigation, and they’ll take a shittier deal, like pennies on the dollar deal and maybe just go out of business. And I think that range of outcomes in the near-term represents a massive failure of collective action on the part of the media industry to not say, this is actually the moment where we should demand that human journalists doing the real work that is dangerous are valuable. We need them, and we will all, together, approach these players in a way that creates at least a semblance of a closed ecosystem.

Well the media industry, but also at some point this is a regulatory question, a question of law. I mean, nothing is stopping Congress from making copyright law designed for the A.I.-era. Nothing is stopping Congress from saying, this is how we think this should work across industries. Not just media, but novelists, but everybody. Well, there are some things that stop Congress from doing a lot of things. The idea that Congress could pass a massive rewrite of copyright law at this moment in time is pretty far afield.

But won’t and couldn’t, I do want to make this distinction here. What you’re saying is Congress is too polarized and bitterly divided over everything and can’t do anything and can’t get anything done, and that’s my whole job man, I know. But what I am saying is that, you could write a law like this.

This is something that ultimately, I don’t just think it’s like a media collective-action problem, but is going to be ultimately a societal-level collective action problem. And maybe we cannot, as a society, act collectively very well. I buy that totally.

So there is one law. There’s the J.C.P.A., the Journalism Competition Preservation Act, which allows media companies to escape antitrust law and bargain collectively with whoever they wish to bargain with. I don’t know if that’s going to pass, I know there’s a lot of interest in it.

So, there are these approaches that have appeared in Congress to solve these problems, but the thing I’m getting at is you have sort of the rapacious wolves, and then you have an industry that’s weak — as you said — that, I think is not motivated to value the work it does as highly as it should. And that is step one.

You and I are both fans of Marshall McLuhan, the media theorist. And he’s got this famous line, ‘the medium is the message.’ And more deeply, what he says is that people, when they see a new medium, they tend to think about the content. For television, it’s the shows, what do you think about this show or that show? For Twitter, the tweets, for a newspaper, the articles. But you have to look behind the content to the actual medium itself to understand what it is trying to tell you.

Twitter, at least in it’s early stages was about all these things can and should be discussed at 140 characters. Television made things much more visual, things should be entertainment. They should be entertaining, the news should be entertaining, which was a little bit of a newer concept back then.

I’ve been trying to think about what is the message of the medium of A.I. What is a message of the medium of ChatGPT, of Claude 3, et cetera. One of the chilling thoughts that I have about it is that its fundamental message is that you are derivative, you are replaceable.

A.I. isn’t good at ideas, yet. It is good it’s style. It can sound like Taylor Swift. It can draw like any artist you might want to imagine. It can create something that looks like Jackson Pollock. It can write like Ezra Klein. It may not be exactly as good at high levels of these professions, but what it is functionally is an amazing mimic.

And what it is saying — and I think this is why a lot of people use it for long enough end up in a kind of metaphysical shock, as it’s been described to me. What it’s been saying is you’re not that special, and that’s one reason I think that it can — we worry about it proliferating all over social media. It can sound like a person quite easily. We’ve long passed the Turing test, and so one, I’m curious if that tracks for you, and two, what does it mean to unleash on all of society a tool that’s basic message is, it’s pretty easy to do what you do, sound like you sound, make what you make?

I have a lot of thoughts about this. I disagree on the basic message. I do think one of the messages of A.I. is that most people make middling work, and middling work is easy to replace. Every email I write is not a great work of art. Like, so much of what we produce just to get through the day is effectively middling. And sure, A.I. should replace a bunch of that. And I think that metaphysical shock comes from the idea that computers shouldn’t be able to do things on their own, and you have a computer that can just do a bunch of stuff for you. And that changes your relationship to the computer in a meaningful way, and I think that’s extremely real.

But the place that I have thought most about I was at the Eras Tour in Chicago when I watched Taylor Swift walk onto a stage, and I saw 60,000 people in Soldier Field just lose their minds, just go nuts. And I’m watching the show, and I’m a Taylor Swift fan. I was there with my niece and nephew and my wife and we were all dressed up. Why am I thinking about A.I. right now? Like truly, why am I thinking about A.I. right now?

It’s because this person has made all of these people feel something. The art that has been created by this one very singular individual has captivated all of these people together, because of her story, because of the lyrics, because it means something to them. And I watch people use Midjourney or generate a story with an A.I. tool, and they show the art to you at the end of it, and they’re glowing. Like, look at this wonderful A.I. painting. It’s a car that’s a shark that’s going through a tornado and I told my daughter a story about it. And I’m like yeah, but this — I don’t want anything to do with this. Like, I don’t care about this. And that happens over and over again. The human creativity is reduced to a prompt, and I think that’s the message of A.I. that I worry about the most, is when you take your creativity and you say, this is actually easy. It’s actually easy to get to this thing that’s a pastiche of the thing that was hard, you just let the computer run its way through whatever statistical path to get there. Then I think more people will fail to recognize the hard thing for being hard. And that’s — truly the message of A.I. is that, maybe this isn’t so hard and there’s something very dangerous to our culture embedded in that.

I want to put a pin in the hard things, easy things. I’m a little bit obsessed by that and want to come back to it. But first I want to talk about A.I. art for a minute, because I do think when we’re talking about everything that’s going to come on the internet, we’re talking about A.I. art. Obviously, much of it is going to get better. Some of it is not distinguishable.

You talked about the example where somebody comes and hands you the A.I. art says, hey, I did this with an A.I. And I’m like eh — and I have that experience a lot, I’ve also really been trying to use these systems and push them, and play with them, and have A.I. character relationships on my phone with Kindroids and whatever.

And there is this deep hollowness at the center of it. It is style without substance. It can mimic me. It can’t think.

Have you found an A.I. that can actually write like you?

I found an A.I. that can mimic certain stylistic tics I have in a way that is better than I think most people could do. I have not found any A.I. that can, in any way, improve my writing for all that you’re constantly told it can. And in fact, the more I try, the worse my writing gets because typically what you have to do to improve your writing is recognize if you’re writing the wrong thing.

I don’t find writing hard, I find thinking hard. I find learning hard. How good a piece of writing is going to be for me is typically about, did I do enough work beforehand? And A.I. can never tell me you didn’t do enough work, you need to make three more phone calls. You need to read that piece you skimmed.

But it can mimic, and I think it’s going to get better and better at mimicking. I think GPT 3 was much worse at mimicking me than GPT 3.5 was, worse than GPT 4 is, and GPT 5 will be even better than that. I believe this is going to get stronger. It raises a question of whether there is anything essential about something being from a human in a wide frame way. Taylor Swift is singular, but the point is that she’s a singular phenomenon. Do we care that things come from people?

I was thinking when I was preparing for this show with you, the Walter Benjamin essay, it’s called “The Work of Art in the Age of Mechanical Reproduction.”

This is like the verge of DNA.

Is it? Yeah, so it comes out in 1935. It’s about the ability to reproduce art. And he says, and I’ll quote it here, “that which whithers in the Age of Mechanical Reproduction is the aura of the work of art.” Then he goes on to say, “by making many reproductions, it substitutes a plurality of copies for a unique existence.”

Benjamin is saying at different times here in different ways, and I’m going to simplify it by trying to bring it into the present, but that there is something lost from when you take the painting and make a copy of a painting. And, he’s obviously right, and he’s obviously — then on the other hand, a lot of people like copies of paintings. It’s easy for the artist to think more of the original than the original deserves to be thought of.

But I wonder about this with humans. How much of something is just the fact that there’s a human behind it? My Kindroid is no worse at texting me than most people I know. But the fact that my Kindroid has to me is meaningful to me, in the sense that I don’t care if it likes me because there’s no achievement for it to me.

The fact that there is a human on the other side of most text messages I send matters. I care about it because it is another mind. The Kindroid might be better in a formulaic way. The kindred might be better in terms of the actual text. I can certainly tune it more to my kind of theoretical liking, but the friction of another person is meaningful to me. Like, I care that my best friend likes me and could choose not to. Is there an aura problem here?

It is so hard to make someone else feel anything other than pain. Like, it’s just like — it’s —

Christ, that’s the darkest thing I’ve ever heard you say.

Yeah, but I believe it in my soul.

Yeah. I think the hardest thing to —

a really different turn as a show right now. [LAUGHS]

You don’t make people laugh, you don’t give them hugs?

No, I think that’s hard. I think that effort is worth it. That’s why I don’t think it’s a dark thing to say. I think the essence of being a good person is pointing your effort at making other people not feel pain. I think bullies make people feel pain because it’s easy. Again, I come back to Taylor Swift in Soldier Field. The thing that was going through my head is, this person is making 60,000 people feel joy, and she’s doing it through art. That is the purpose of art. The purpose of art is to inspire feelings, to inspire emotion.

And so I look at this A.I. and it’s like, we’re going to flood our stuff, and the only emotion that it is really meant to inspire is materialism, is a transaction. That’s bad. I just think that’s bad. I think we should make some stuff that inspires more joy, that inspires more affection, that inspires more consternation.

And one of the messages embedded in the medium of A.I. is that there is an answer. That’s weird. That is a truly weird thing for a computer to say to you. You ask it about a war, and it’s like I won’t answer that question because there’s no answer there. You ask it about how to cook an egg and it’s like here’s the answer. You’re like what are the four steps to fold a bed sheet? It’s like here’s the answer, I did it. Tell me a bedtime story for my child. It says, here’s an answer, I just delivered this to you at your specifications.

And I think the thing you’re saying about having another mind there is — you want to be in a relationship, like an emotional relationship with another person. Maybe it’s mediated by technology, maybe we’re face-to-face like we are now, but that tension and that reality of — oh, I can direct my effort towards negative and positive outcomes, I have never found it with an A.I.

Shannon Vallor is a philosopher of technology, and she’s got a book coming out called “The A.I. Mirror,” and I like the way she puts this, because there’s this way that turns is somewhat warped mirror back on ourselves when I was saying a few minutes ago that the message of A.I. is that you’re derivative. That leaves something out. What it’s really saying is that the part of you that often the economy values is derivative, is copyable because we actually ask people a lot of the time to act like they’re machines.

This is why I don’t take much comfort in the Taylor Swift example. You said a few minutes ago, most people do mediocre work most of the time. Even great people do mediocre work most of the time. We constantly ask huge amounts of the population to do things that are very rote. Keep inputting this data on forms, keep filling out this tax form. Some lawyers arguing for the Supreme Court, a lot of them just write up various contracts. And that’s a good job in the sense that it pays well, it’s inside work, but it doesn’t ask you to be that full of a human being.

Now, you can imagine a sort of utopian politics in society — and people on the left sometimes do — that this comes in and it’s like great, we can automate away this derivative inhuman work, and people will be free to be more full human beings. You actually like — maybe the value of you is not what you can create but what you can experience. A.I. can’t enjoy a day at the park with its family.

But we have an entire society set up to encourage you to premise your self-worth on your work and your wages. And also, if you lose that work and that wages, to rob you of that self-worth. And one thing I’m sure of is that our politics and our economic systems are not going to advance as quickly as A.I. is going to advance.

This is where I think people do properly worry about automation, when people lost manufacturing jobs to lower wage workers in China. We didn’t say great, you don’t have to do this stultifying work in the factory anymore. We said, you’re out of work, you’re screwed. And I do think one of the deep confrontations of it is, what do we value in people and then how do we express that value because I think what A.I. in some ways is going to take advantage of here, or at least, is going to challenge, is it to the extent we value people socially for their economic contribution, or what they’re paid. That’s pretty thin reed for human value to rest on.

Yeah, I buy that. One of my favorite things that I’ve covered in the past few years is a thing called robotic process automation, which is very funny. Just abstractly, deeply hilarious. There are lots and lots of companies throughout the United States that built computer systems 10, 15 years ago, 20 years ago. Hospital systems are famous for this. They have billing systems. They have buildings full of people who use Microsoft Excel on Windows ‘95.

And replacing that as costly and complicated. It can’t break — if you put in the new system and it didn’t bring all the data over in exactly the right way, the whole hospital stops working. So they just buy other computers to use their old computers. Which is wild, and there’s like billion dollar companies that do this.

They will sell you a brand new, state of the art computer and it will connect to the keyboard and monitor jack of your old computer, and it will just use the Windows ‘95 for you, which is just bonkers. It’s like Rube Goldberg machine of computers using old computers, and then your office full of accountants who knew how to use your old system will go away.

But then A.I. creates the scale problem. What if we do that but instead of some hospital billing system built in the ‘90s, it’s just the concept of Microsoft Excel, and now you can just sort of issue a command on your computer and it’ll go use Excel for you and you don’t need an accountant, you don’t need a lawyer.

And I think even in those cases what you’re going to find is the same thing you talked about with writing — you have to know what you want. You have to know what the system doesn’t know. You have to be able to challenge the model and have it deliver you the thing that, in most business model conversations I find to be the most important word, our assumption is — and then you can poke at that really hard.

What percent of workers are actually asked to poke at the assumptions of their organization, because I worry it’s not as high as you think it is, or implying there. I’m not worried about Taylor Swift. I’m not worried about Nilay Patel. And I don’t just want to make this about wages. That’s a jobs sort of another conversation.

But I do — I mean, as you were saying, these are billion dollar companies that automate people who do backend office work already.

All over the place.

There’s a huge amount of work like that. And if I felt confident as some of the economists say that we’ll just upmarket people into the jobs where they use more human judgment, David Autor who’s a great trade economist at MIT, just made this argument recently, that what A.I. is going to do is make it possible for more people to exercise judgment and discernment within their work, and I hope he is right. I really hope he is right. But I think a lot of organizations are not set up for a lot of people to use judgment and discernment. They treat a lot of people like machines, and they don’t want them doing things that are complicated and step out of line and poke at the assumptions in the Excel doc. They want the Excel doc ported over without any mistakes. It seems plausible to me that we’re going to get to that.

Do you think their bosses want to be able to poke at the assumptions though?

But if you — I mean this is actually something I believe about the whole situation. The economy needs fewer bosses and workers.

Think about this in the journalist context or the writing context, where I think what A.I. naturally implies that it’s going to do is turn many more people into editors and writers. Because for a lot of content creation that doesn’t require a lot of poking at assumptions, mid-level social media marketing — a lot of people are doing that job right now. But the people doing marketing for a mall —

Yeah, that is the MailChimp example. That is the product that they are building.

And so what you have then is we used to have a bunch of these social media marketers and now you have one person overseeing a couple systems, like making sure they didn’t say something totally crazy. But you need fewer editors and you need writers. I mean, you know The Verge is structured. You know how The Times is structured. And this is one of my deep worries.

And then this goes to the thing you were getting at earlier, which is one way I think that A.I. could actually not make us more productive, more innovative, is that a lot of the innovation, a lot of the big insights happen when we’re doing the hard thing, when we’re sitting there trying to figure out the first draft, or learn about a thing, or figure out what we’re doing.

One of the messages of the medium of A.I. is be efficient. Don’t waste your time on all this. Just tell the system what to do and do it. But there’s a reason I don’t have interns write my first draft for me.

They could do it. But you don’t get great ideas, or at least not as many of them, editing a piece of work as you do reporting it out, doing the research, writing the first draft. That’s where you do the thinking. And I do think A.I. is built to kind of devalue that whole area of thinking.

We are working on a big story at The Verge right now that I’m very excited about. But there are four of us right now in an argument about whether we should tell that story in chronological order or as a series of vignettes. There is no right answer to this question. There’s just four people who are battling it back and forth.

I think vignettes.

Yeah. By the way, I’m on team vignette.

Good man. [LAUGHS]

My belief is that it’s easier to digest a long story when it’s composed of lots of little stories as opposed to one long one. I’m being outvoted right now — editor in chief. I should replace them all with A.I., just get them out of here. [CHUCKLES] But that is the kind of work that I think makes the end product great. And I think going from good to great is still very human.

Into the economy, though, you’re right, most people are not challenged to go from good to great. Most people are challenged to produce good consistently. And I think that is kind of demoralizing. I don’t know how many first-year Deloitte consultants you have encountered in your life. I’ve encountered quite a few of them. I went to law school. It’s like a — we made — there was a factory of that thing — or first-year law associates.

They’re not in love with their jobs. They’re in love with the amount of money they make, that’s for sure. But any first-year associate doing doc review in a basement — yeah, you could probably just be like, tell the A.I. to find the four pieces of relevant information in these 10,000 page records from whatever giant corporation we’re suing today. That’s fine.

I think that there’s a turn there where maybe we need less first-year associates doing that thing and we need more first-year associates doing something else that is difficult, that the A.I. can’t yet do. And I think a lot of this conversation is predicated on the notion that generative A.I. systems, L.L.M.s will continue on a linear curve up in terms of capability. I don’t know if that’s true.

But I hear a lot of this conversation. I’m like, there’s always a thing they can’t do. And maybe that thing is not the most amount of scale, social media marketing for them all, but it is always the next amount of complexity. And there’s no guarantee that this set of technologies will actually turn that corner. And you can keep going all the way to A.G.I. There’s no guarantee that an L.L.M. is going to hit A.G.I. and just run the world economy for us. There’s a lot of steps between here and there that I think human beings can fit into.

So I want to go back, then, to the internet for a bit, which is I think the presentation we’ve offered is fairly pessimistic. You, when I read and listen to you on this, are — I wouldn’t call it pessimistic. I would say a little excited by the idea of a cleansing fire.

So one theory here — and you should tell me if this is reading you right — but is that this will break a lot of the current — the current internet is weakened. It’s weakened in many cases for good reasons. Google, Meta, et cetera, they’ve not created an internet many of us like. And that this will just make it impossible for that internet to survive. The distribution channels will break. And then something. So first, is that how you see it? And second, then what something?

That is very much how I see it. I would add a generational tinge to that, which is I grew up in that weird middle generation between X and millennials. I think temperamentally I’m much more Generation X. But they describe it as they didn’t have computers and then you have computers. You play the Oregon Trail. That’s me on the nose.

I distinctly remember life before computers. It’s an experience that I had quite viscerally. And that shapes my view of these tools. It shapes my view of these companies. Well, there’s a huge generation now that only grew up in this way. There’s a teenage generation right now that is only growing up in this way. And I think their natural inclination is to say, well, this sucks. I want my own thing. I want my own system of consuming information. I want my own brands and institutions. And I don’t think that these big platforms are ready for that moment. I think that they think they can constantly be information monopolies while they are fending off A.I.-generated content from their own A.I. systems. So somewhere in there all of this stuff does break. And the optimism that you are sensing from me is, well, hopefully we build some stuff that does not have these huge dependencies on platform companies that have no interest at the end of the line except a transaction.

OK, but you’re telling me how the old thing dies. And I agree with you that at some point the old thing dies. You can feel it. It’s moribund right now. You’re not telling me what the new thing is, and I’m not saying you fully know. But I don’t think the new thing is just a business model that is not as dependent on Meta. I mean, on some level, there’s going to be a lot of A.I. around here.

It’s an audience model. It’s not dependent on these algorithms.

But is there — I guess one question I have is that, one — I mean, you know where the venture capital is going right now.

Everything is going to be built with A.I. —

— laced through every piece of it. And some of it, for all we’re talking about, might be cool, right? I’m not saying you’re mostly going to make great art with A.I. But actually, Photoshop did create a lot of amazing things.

And people are going to get better at using this. They’re going to get more thoughtful about using it. The tools are going to get better. But also the people are going to figure out how to use the tools. I mean, you were talking about player pianos earlier. I mean, way beyond player pianos, you have huge libraries of sounds you can manipulate however you want. And now I go listen to a lot of experimental electronic music. And I think a lot of that is remarkable art. I think a lot of that is deeply moving.

I am curious what, to you, the good A.I. internet is, because I don’t think that the next internet is just going to be like we’re going to roll the clock back on the business model. The technology is going to roll forward into all this stuff people are building.

I’m not so sure about that.

I think we’re about to split the internet in two. I think there will be a giant commercial A.I.-infested internet. That’s the platform internet. That’s where it’s going. Moribund, I agree. But it will still be huge. It’s not going away tomorrow. And they will figure out — these are big companies full of smart people with the most technology.

Mark Zuckerberg is like, I have the most NVIDIA H100 GPUs. Come work here. We’ll pay you the most money. They will invent some stuff and it will be cool. I’m excited about it. But that version of the internet —

You sure sound excited about it. [LAUGHS]

Well, I am. I mean, I love technology. This is our — The Verge’s competitive differentiation in the entire media industry is, like, we really like it. And I’m excited to see what they build. I think there’s some really neat things being built. When I think about the information ecosystem, I’m vastly more pessimistic because of the fact that all of these networks are geared to drive you towards a transaction.

And I don’t mean that in some anticapitalist way. I mean literally the incentives are to get you to buy something. So more and more of the stuff that you consume is designed around pushing you towards a transaction. That’s weird. I think there’s a vast amount of white space in the culture for things that are not directly transactable.

I think next to that you’re going to get a bunch of people, companies who say our differentiation in this market is that there’s no A.I. here. And they will try to sell that. And I don’t know how that experiment plays out. I don’t know if that experiment will be successful.

I do know that that experiment will be outside of the distribution channels that exist now because those distribution channels are being run by companies that are invested heavily in A.I. And I’m hopeful that over there, on whatever new non-A.I. internet that exists, that some amount of pressure is placed on the other distribution channels to also make that distinction clear.

I’m just thinking about this, and the thing that it brings to mind for me is the resurgence of vinyl —

— and the dominance of streaming platforms. So what I would think of as the music industry of — how many years ago was C.D.s? I don’t actually remember now. But what it did was split into — there’s been a resurgence of vinyl, the sort of analog. It’s a little cool. I actually just bought a record player recently, or was given one by my wonderful partner. But that’s not very big.

Then there’s these huge streaming platforms, right? I mean, most people are listening on Spotify, on Apple Music, on YouTube Music, on Amazon, et cetera. And I don’t think we feel like we figured that out very well. But I do think that’s probably going to be the dynamic. I mean, I do think there are going to be things you go to because you believe it is a human being or because you believe the A.I. is used well.

I do also think the big things to come are going to be the things that figure out how to use A.I. well rather than poorly. Maybe that also means honestly and transparently, rather than dishonestly and opaquely.

Maybe the social internet dies because, one, we don’t really like it that much anymore anyway, but also because it’s too hard to figure out what’s what. But actually, an internet of A.I. helpers, assistants, friends, et cetera, thrives. And on the other side, you have a real human. I don’t know. But give me more of the Nilay technology side.

What can A.I. do well? If you were building something or if you were imagining something to be built, what comes after?

By the way, the music industry just released its numbers. Vinyl outsold CDs for the second year running. Double the amount of revenue in vinyl than CDs.

That’s wild, actually.

It’s crazy. And all of that in total is 11 percent of music industry revenues in ‘23 compared to 84 percent of the revenue is streaming. So you are correct. This is a big distinction. People want to buy things, and so they buy one thing that they like. And they consume everything in streaming.

What happens when Spotify is overrun by A.I. music? You can see it coming. What happens when you can type into Spotify, man, I’d really like to listen to a country song. Just make me one. And no one down the line has to get paid for that. Spotify can just generate that for you.

I think that’s going to push more people in the other direction. I really do. That there will be this huge pot of just make me whatever exactly I want at this moment money over here. But the cool people are still going to gravitate towards things that are new. I just believe that so firmly in my heart that when I think about where does the technology for that come from, I still think it comes from basic open platforms and open distribution.

The great power of the internet is that you can just make a whole new thing. And I don’t think that anyone has really thought through what does it mean to decentralize these platforms. What does it mean to — I don’t know — build an old-school portal where it’s just people pointing at great stuff as opposed to open this app and an algorithm will just deliver you exactly what we think you want, or, down the line, generate content for you that we think that you will continue watching.

I think — and this is maybe a little bit of a counterintuitive thought — that this is actually a great time to begin things in media. I think that we have a more realistic sense of the business model and what will actually work. They need to build an audience. They need to build something people will actually pay you for. I think a lot of the problem right now is things built for another business model that failed are having a lot of trouble transitioning because it’s very, very hard to transition a structure. Now, that doesn’t mean it’s a great business. It’s not what I hoped it would become. It’s not the advertising revenue I hoped we would have. But it’s something.

What feels fully unsolved to me right now is distribution, right? When I was a blogger, the way distribution worked was people would find me because other blogs would link to me. And then if they liked me, they would put me in their bookmarks section.

Then they would come back the next day by clicking on a bookmark. I don’t think any of us think that much about bookmarks anymore. That’s not really how the internet works. Things moved to search. They moved, primarily for a long time, to social. And that was a way you could create distribution.

You could go from — you started a website. We started Vox, right? We started Vox in 2014 or 2015. The day before we launched, we had no visitors. And pretty quickly we had a lot of things that were working on social and working on Search. And we had millions and millions and millions every month.

But now social is broken as a distribution mechanism. I mean, Elon Musk has made Twitter anti-news distribution. Google search has become very, very messy. People don’t have the old bookmarks habit in the way they did. And so if you’re starting something new, the question of how you build that audience, how you go from nothing to an audience, feels very unsolved.

Yeah. That’s the cleansing fire. That’s the thing I’m excited about. Here’s a new problem in media. Here’s a new problem that’s being created by A.I.

If I were to tell you five years ago, I’m going to launch a new property and the core insight that I have is that we need to replace the distribution mechanisms of the internet, you would not pay me any money. You would not fund that idea. You would not say — well, you would say, get some traffic on Twitter and start a Substack or start a YouTube channel, anything except figure out a new distribution method to compete with these social media companies.

You have that idea now. And people are like, yeah, that’s the problem. We have to solve that problem. That is the problem to solve, because Twitter has blown itself up in whatever way Elon is blowing it up, because the other social channels have become the Home Shopping Network, by and large, because YouTube has optimized itself into making Mr. Beasts and only Mr. Beasts, right?

It’s weird, by the way, that YouTube exists. We’ve barely talked about it on this podcast. It is the thing most people watch most of the time. It supports no journalism. At scale, the idea that there’s not an ABC News of YouTube on a distribution platform of that size is a moral failing on Google’s part. I really believe this. And no, we never really talk about it. It’s just — YouTube is ignored. It has become such an infrastructure that we never talk about it.

But my view is that YouTube is the most politically important platform. Everyone wants to talk about TikTok. I think YouTube is much more significant.

Yeah, and they run it really well. They run it as infrastructure. And they talk about it as infrastructure. But it’s weird that we have not built great media company-sized media companies on YouTube’s pipes. We just haven’t done it. So you look at that landscape now and you’re like, well, if I want to do that, if I want to build my own audience, I cannot depend on these companies. I have to be able to do something else.

And maybe A.I. does help you do that. Maybe it does help you send a million marketing messages so people start coming to your website directly. Maybe it does start crafting home pages personalized for people based on your library of content so people see the thing they like the most when they show up. There’s a bunch of moves we can all take from social media companies now to build more engaging, more interesting products using A.I., which will make it easier because the A.I. is a technology commodity. You can just go buy it and use it.

But we have to actually build those products. We have to want to build those products as an industry. And that my pessimism is rooted in the idea that the industry kind of sucks at this. We are very much stuck in, we should go send some reporters out into the world, they should come back, write down what they saw, and then hopefully someone else points them at it. And it’s just like, well, that’s been a losing proposition for a decade. We should try something else.

Do you think, beyond the media, because not everything online is media —

Do you think beyond the media, that there is the glimmers of the next thing? I mean, let me give you the thesis I have, which is that the next thing is that the A.I. is somehow your assistant to the internet, right? We seem to me to be moving towards something where the overwhelm is so profound that you actually need some kind of agent working on your behalf to make it through all this.

I mean, you can imagine this is the world of “Her,” the Spike Jonze movie. But you can imagine it as other things, too. There’s going to be software coding agents. The guys who started Instagram started then this thing called Artifact, which is using more A.I. personalization to try to tell people what they might like in the news. It didn’t really work out, but it was an interesting project for a minute.

I think a lot of us feel we spent years now being acted upon by algorithms. And one thing about A.I. is that it’s an algorithm you act on, right? You tell it how to act. Assuming that business model allows that, that it doesn’t have a secret instruction to sell you soap or whatever —

— that’s interesting, right? That’s a pretty profound inversion of the internet we’ve been in.

Let me poke really hard at the true difference between an algorithm that shows you stuff and an algorithm that goes and gets you what you want, because I don’t know that there’s a huge difference in the outcome of those two different processes. So for example, I do not trust the YouTube Kids algorithm. I watch my daughter watch YouTube.

No, why would you?

It is just a nightmare. I don’t know why we let her do it, but we did. And now we’re in the rabbit hole and that’s life. I mean, she’s five. And I will literally say, are you watching garbage? And she’d be like, I am, because she knows what I think is garbage. She’s much smarter than the YouTube Kids algorithm. And then she’s like, can I watch a little more garbage? This is a real conversation I have with my five-year-old all the time.

I would love an A.I. that would just preempt that conversation. Just watch this whole iPad for me and make sure my kid is safe. That’s great. But that is a limitation. It is not an expansion. And I think the thing that I’m seeking with all of these tools is how do we help people expand the set of things that they’re looking at.

Well, let me push on this for a minute, because for a long time a lot of us have asked people, the social media companies — that I have, I’m sure you have — why don’t you give me access to the dials of the algorithm?

Right? I don’t want to see things going viral. If there’s a virality scale of 1 to 10, I want to always be at a 6, right?

I don’t want to see anything over a 6. And I can’t. I wish I could say to Google, I would like things that are not optimized for S.E.O. I just don’t want to see recipes that have a long personal story at the top. Just don’t show me any of them.

But I can’t do that. But one of the interesting things about using the current generation of A.I. models is you actually do have to talk to it like that. I mean, whether I am creating a Replika or a Kindroid or a Character.AI, I have to tell that thing what it is supposed to be, how I want it to talk to me, how I want it to act in the world, what it is interested in, what kinds of expertise it has and does not.

When I’m working with Claude 3, which is the A.I. I use the most right now, I have one instance of it, that I’m just like, you are a productivity coach and you are here to help me stay on task. But I have another where I’m getting some help on, in theory, looking at political science papers, so it’s actually not that good at that.

But this ability to tell this extraordinarily protean algorithm what I want it to do in plain English, that is different, right? The one thing that A.I. seems to make possible is an algorithm that you shape in plain English, an agent that you are directing to help you, in some cases, maybe create the internet, but much more often to navigate it.

Right now it is very hard for me to keep up on the amount of news, particularly around the amount of local news I would like to keep up on. If there is a system that I could say, hey, here’s some things I’m interested in from these kinds of sources, that would be very helpful to me. It doesn’t seem like an impossible problem. In fact, it seems like a problem that is inches away from being solved. That might be cool.

I think that’d be great. I’ve known you for a long time. I think you have a unique ability to articulate exactly what you want and tell it to a computer. [LAUGHS] And you have to scale that idea, right? You have to go to the average — our mothers and say, OK, you have to tell the algorithm exactly what you want. And maybe they’ll get close to it, maybe they won’t, right?

You don’t feel like mothers are able to tell you what they want?

[LAUGHS] I like that idea a lot. I think fundamentally that is still an A.I. closing the walls around you. And I think the power of the recommendation algorithm is not expressed in virality. It’s actually to help you expand your filter bubble. Here’s a band you’d never heard of before. Here’s a movie you never thought of watching. Here’s an article about a subject that you weren’t interested in before.

I think TikTok, in its 2020 TikTok moment, was terrific at this. Everyone was going to sing a sea shanty for five minutes, right? Why do we suddenly care about this and it’s gone? And it was able to create cultural moments out of things that no one had ever really thought of before. And I want to make sure, as I use A.I., that I’m actually preserving that, instead of actually just recreating a much more complicated filter bubble.

I think it’s a good place to end. Always our final question, for the Nilay Patel recommendation algorithm —

what are three books you’d recommend to the audience?

Well, I’m sorry, Ezra, I brought you six.

Did you really?

Is that allowed?

Did you actually bring six?

I didn’t bring six physical books, but I have six recommendations for you.

Damn. All right, go through them quick, man.

They’re in two categories. One is the three books that I thought of and three books from Verge people that if people are interested in these ideas are important. So the first one is “The Conquest of Cool” by Thomas Frank, one of my favorite books of all time. It is about how advertising agencies in the ‘60s co-opted the counterculture and basically replaced counterculture in America. I’ve thought about this a lot because I’m constantly wondering where the punk bands and rage against the machines of 2024 are. And the answer is that they’re the mainstream culture. It’s very interesting. Love that book. It explains, I think, a lot about our culture.

Two is “Liar in a Crowded Theater” by Jeff Kosseff, which is a book about the First Amendment and why we preserve the ability to lie in America. I am very complicated thoughts about the First Amendment right now. I think social media companies should do a better job protecting my kid. I also think the First Amendment is really important. And those ideas are crashing into each other.

Third, I love the band New Order. I know you’re a music fan, so I brought you a music recommendation. It’s “Substance: Inside New Order” by Peter Hook, who is the bassist of New Order. This band hates each other. They broke up acrimoniously, so the book is incredibly bitchy. It’s just a lot of shit-talking about the ‘80s. It’s great.

But inside the book, he is constantly talking about how the technology they used to make the music of New Order didn’t work very well. And there’s long vignettes of why the songs sound the way they do because of how the synthesizers worked. And that just brings together all the ideas I can think of. So those are the three outside of The Verge universe.

But there are three from Verge people that I think are very important. The first is “Everything I Need I Get From You” by Kaitlyn Tiffany, who’s one of my favorite Verge expats. It is about how the entire internet was shaped by the fandom of the band One Direction. And I think this is totally underemphasized, underreported that fandoms are actually what shape the internet. And a lot of what we think of as internet culture is actually fandom culture. And so Kait’s book is really good.

The other, obviously, I have to shout it out is “Extremely Hardcore” by Zoë Schiffer, who basically wrote about the downfall of Twitter. And I think understanding how a social network works — these are lots of people making lots of decisions, and it was just dismantled. And now you can see how the social network broke. And I think we take these things for granted.

And then the third is “Beyond Measure” by James Vincent, which is a history of the systems of measurement and how political they are. And it is one of my favorite books because it is — you just take this stuff for granted. And you look at it, and you’re like, oh, this was deeply, deeply acrimonious.

Nilay Patel, you’re saving the internet through blogging again.

Your podcast is “Decoder.” Thank you very much.

Thanks, man. [MUSIC PLAYING]

This episode of “The Ezra Klein Show” was produced by Claire Gordon. Fact-checking by Michelle Harris with Kate Sinclair and Mary Marge Locker. Our senior engineer is Jeff Geld. We’ve got additional mixing by Isaac Jones and Efim Shapiro. Our senior editor is Claire Gordon. The show’s production team also includes Annie Galvin, Rollin Hu and Kristin Lin. We have original music by Isaac Jones. Audience strategy by Kristina Samulewski and Shannon Busta. The executive producer of New York Times Opinion Audio is Annie-Rose Strasser. And special thanks here to Sonia Herrero.

EZRA KLEIN: From New York Times Opinion, this is “The Ezra Klein Show.”

Earlier this week, we did an episode on how to use A.I. right now. Now, I want to turn the question around and look at how A.I. is being used on you right now. One of the conversations has been sticking in my head was with this person in the A.I. world who was saying to me that if you look at where use has been sticky, if you look at where people keep using it day after day, you’re looking at places where the product doesn’t need to be very good.

That’s why it’s really helpful for college and high school students, college and high school papers — they’re often not very good. That’s sort of their point. It’s why it’s working pretty well for a very low-level coding tasks. That kind of work doesn’t need to be very good. It gets checked and compiled, and so on.

NILAY PATEL: Thank you for having me. This is very exciting.

EZRA KLEIN: Let’s just begin with the big question here, which is what is A.I. doing to the internet right now?

NILAY PATEL: It is flooding our distribution channels with a cannon-blast of — at best — C+ content that I think is breaking those distribution channels.

EZRA KLEIN: Why would it break them?

NILAY PATEL: So most of the platforms the internet are based on the idea that the people using those platforms will in some sort of crowdsourced way find the best stuff. And you can disagree with that notion. I think maybe the last 10 years have proven that that notion is not percent true when it’s all people.

When you increase the supply of stuff onto those platforms to infinity, that system breaks down completely. Recommendation algorithms break down completely, our ability to discern what is real and what is false break down completely, and I think importantly, the business models of the internet break down completely.

So if you just think about the business model of the internet as — there’s a box that you can upload some content into, and then there’s an algorithm between you and an audience, and some audience will find the stuff you put in the box, and then you put an infinity amount of stuff into the box, all of that breaks.

EZRA KLEIN: Thank you for bringing in the supply language. So, I’ve been trying to think about this as this supply and demand mismatch. We have already had way more supply than there is demand. I wasn’t buying a lot of self-published Amazon books. Is the user experience here actually different?

NILAY PATEL: I think that’s a great question. The folks who write the algorithms, the platforms, their C.E.O.s, they will all tell you this is just a new challenge for us to solve. We have to out what is human, what is A.I.-generated. I actually think the supply increase is very meaningful. Like, maybe the most meaningful thing that will happen to the internet because it will sort out the platforms that allow it to be there and have those problems, and the places that don’t. And I think that has not been a sorting that has occurred on the internet in quite some time, where there’s two different kinds of things.

EZRA KLEIN: Didn’t we already pivot to video a couple years ago?

NILAY PATEL: We pivoted to video — I actually love it when LinkedIn adds and takes away these features that other platforms have. They added stories because Snapchat and Instagram had stories, and they took the stories away because I don’t think LinkedIn influencers want to do Instagram Reels, but now they’re adding it again.

EZRA KLEIN: So I want to hold on to something that you’re getting at here. Which, to me, is one of the most under-discussed parts of A.I., which is how do you actually make money off of it? And right now, there are not actually that many ways.

NILAY PATEL: Yeah, I’ve been talking to a lot of C.E.O.s of web companies and email companies on Decoder for the past year. I asked them all the same question, why would you start a website? Why would you send an email? And so, you asked the C.E.O. of Squarespace or Wix or we just had the C.E.O. of MailChimp on the show. And her answer is a little terrifying. Like, maybe openly terrifying.

EZRA KLEIN: But how is A.I. changing that at all because that sounds to me like the thing that is already happening.

NILAY PATEL: So, this is what I mean by the increase in scale. That’s the dream. This is supposed to be what actually happens, but they can only do it in broad cohorts, which is why you get the luggage email after you’ve bought the luggage email or the luggage ad, after you bought the luggage ad.

They know you are a person who used a Wi-Fi network in a certain location at a certain time, they can track that all over the place. They know what you’ve searched for. They know that you went and made a luggage transaction. You are now categorized into people who are likely to buy luggage, whether or not that loop was closed. You put some luggage in a shopping cart.

But that’s still a cohort, they can only do that broadly. And these cohorts can be pretty refined, but they can only do it broadly. With A.I. the idea is we can do that to you individually — the A.I. will write you an email, we’ll write you a marketing message, will set you a price. That isn’t 100x increase the amount of email that will be generated.

EZRA KLEIN: When I get spammy advertising I don’t really think about there being a human on the other end of it. Maybe to some degree there is, but it isn’t part of the transaction happening in my head. There are a lot of parts of the internet that I do think of there being a human on the other end — social media, reviews on Amazon, books — I assume the person who wrote the book is a person. How much of what I’m currently consuming may not be done by human in the way I think it is, and how much do you think that’s going to be in a year, or two, or three years?

NILAY PATEL: I’m guessing your media diet is pretty well human-created because I know that you are very thoughtful about what you consume and what signals you’re sending to the algorithms that deliver your content. I think for most people —

EZRA KLEIN: My mom’s, let’s use my mom’s.

NILAY PATEL: Mom’s are good. I would love to take my mom’s phone and throw it into the ocean and never let her have it again. I openly fear what content comes through my mother through WhatsApp. It terrifies me that I don’t have a window into that. I can’t monitor it. The same software I want to use to watch my daughter’s internet consumption, I would love to apply it to my parents because I don’t think they have the media literacy — they’re much older — to even know, OK, this might be just some A.I.-generated spam that’s designed to make me feel a certain way.

EZRA KLEIN: How can they not be ready for that?

NILAY PATEL: Because they’re the ones making it. This is the central tension of — in particular, I think Google. So, Google depends on the web, the richness of the web is what Sundar Pichai will tell you. He used to run search, he thinks about the web. He cares about it, and you look at the web and you’re like, you didn’t make this rich at all. You’ve made this actually pretty horrible for most people most of the time. Most people — if you search Google to get a credit card, that is a nightmarish experience — like, fully nightmarish. It feels like getting mugged.

EZRA KLEIN: This helps explain a story that I found very strange. So, 404 Media, which is a sort of newer outlet reporting on tech. They found that Google News was boosting stolen A.I. versions of news articles — and we’re seeing this all over. An article by me or by some other journalist shows up in another place, very slightly rewritten by an A.I. system, with an A.I. generated author and photo on top of it. So, we’re seeing a lot of this.

NILAY PATEL: Yeah. Fundamentally, I think if you are at Google and the future of your stock price depends on Gemini being a good competitor to GPT-4 or 5 or whatever OpenAI has, cannot run around saying this is bad. The things it makes are bad.

EZRA KLEIN: What about when it’s not?

NILAY PATEL: I don’t know how fast that is coming. I think that is farther away than people think. I think ‘will it fool you on a phone screen?’ is here already, but ‘is this good’ is, I think, farther away than —

EZRA KLEIN: But a lot of internet content is bad.

NILAY PATEL: That’s fair.

EZRA KLEIN: I mean, you know this better than me. Look, I think it is axiomatic that A.I. content is worse right now than it will ever be.

NILAY PATEL: Sure.

EZRA KLEIN: I mean the advance in image generation over the past year has been significant. That’s very real. And preparing for this conversation, I found myself really obsessing over this question, because one way to talk to you about this is, there’s all this spammy garbage coming from A.I. that is flooding the internet.

NILAY PATEL: I think there’s a sentimentality there. If you make a content farm that is the best content farm, that has the most answers about when the Super Bowl starts, and those pages are great. I think that’s a dead end business. Google is just going to answer the questions. I think that’s fine. I think if you ask Google what time the Super Bowl is, Google should just tell you.

I think if you ask Google how long to boil an egg, Google can just tell you. You don’t need to go to some web page laden with ads and weird headings to find those answers. But these models in their most reductive essence are just statistical representations of the past. They are not great at new ideas.

EZRA KLEIN: I’m going to spend some time thinking about the idea that boredom is an under-discussed driver of our culture. But I want to get at something else in there — this idea of Google answering the question. We’re already seeing the beginnings of these A.I. systems that you search the question that might — at another time — have brought you to The Verge, to CNN, to The New York Times, to whatever.

NILAY PATEL: I think there’s a reason that the A.I. companies are leading the charge to watermark and label content as A.I.-generated. Most of them are in the metadata of an image. So most pictures you see in the internet, they carry some amount of metadata that describes the picture. What camera was taken on, when it was taken, what image editing software was used.

EZRA KLEIN: So their training data remains less corrupted?

NILAY PATEL: Yeah. I think there’s a very straightforward incentive for them to figure out the watermarking, labeling stuff they want to do. And they have coalitions, and tasks force, and Adobe talks about the image of the Pope and the puffer jacket as a, “catalyzing moment” for the metadata of A.I. because people freaked out. They’re like oh, this thing looks real. But they have a real incentive to make sure that they never train on other A.I. generated content.

I’m not saying we don’t — like, other publishers do this. But the point of these algorithms is, ideally, to bring you to the HouseFresh people, is to bring you to the person who cares so much about air purifiers they made a website about air purifiers, and we’re not doing that anymore. And so if you were to say, where should a young person who cares the most about cars, or who cares the most about coffee, or whatever.

Where are they going to go? Where are they going to make stuff? They’re going to pick a closed platform that ideally offers them some built in monetization, that ideally offers them some ability to connect directly with an audience. They’re not going to go to a public space like the web, where they might own their own business, which would be good. But they’re also basically at the mercy of thieves who come in the night and take all their work away.

EZRA KLEIN: But also, if you kill HouseFresh, then two years later when you ask the A.I. what air purifier should I get, how does it know what to tell you?

NILAY PATEL: Yeah, I don’t the answer to that question.

EZRA KLEIN: I don’t think they do either.

NILAY PATEL: Yeah again, this is why I think that they are so hell-bent on labeling everything. I think they need some people around in the future.

EZRA KLEIN: But labeling is good. I mean, that keeps you from getting too much garbage in your data set. But replacing a bunch of the things that the entire informational world relies on to subsidize itself — to fund itself — like this to me is a thing that they don’t have an answer for.

NILAY PATEL: Wait, let me ask you a harder question. Do they care?

EZRA KLEIN: Depends on they, but I don’t think so.

NILAY PATEL: Yeah.

EZRA KLEIN: Or at least they care in the way that I came to realize Facebook, now Meta, cared about journalism. People say they didn’t care about journalism. I don’t believe that’s actually true. They didn’t care enough for it to mean anything. Like, if you asked them, if you talked with them, if you had a drink, they would think what was happening to journalism was sad.

NILAY PATEL: [LAUGHS]

EZRA KLEIN: And if it would cost them nothing, they would like to help. But if it would cost them anything — or forget costing them anything. If they would begin to help and then recognize an opportunity had been created that they could take instead of you, they would do that. That’s the way they care.

NILAY PATEL: I keep a list of TikToks that I think each individually should be a Ph.D. thesis in media studies. It’s a long list now. And all of them are basically just layers of copyright infringement in their own weird way.

My favorite is — it’s a TikTok, it has millions of views. It’s just a guy reading a summary of an article in the journal Nature. It has millions of views. This is more people that have ever considered any one article in the journal Nature — which is a great journal. I don’t mean to denigrate it. It’s a proper scientific journal. They work really hard on it. And you just go 5 steps down the line, and there’s a guy on TikTok summarizing a summary of Nature, and you’re like what is this? What is this thing that I’m looking at? Will any of the million viewers of this TikTok buy one copy of Nature because they have encountered this content? Why did this happen?

And the idea is, in my mind at least, that those people who curate the internet, who have a point of view, who have a beginning and middle, and an end to the story they’re trying to tell all the time about the culture we’re in or the politics we’re in or whatever. They will actually become the centers of attention and you cannot replace that with A.I. You cannot replace that curatorial function or that guiding function that we’ve always looked to other individuals to do.

And those are real relationships. I think those people can stand in for institutions and brands. I think the New York Times, you’re Ezra Klein, a New York Times journalist means something. It appends some value to your name, but the institution has to protect that value. I think that stuff is still really powerful, and I think as the flood of A.I. comes to our distribution networks, the value of having a powerful individual who curates things for people, combined with a powerful institution who protects their integrity actually will go up. I don’t think that’s going to go down.

EZRA KLEIN: You said something on your show that I thought was one of the wisest, single things I’ve heard on the whole last decade and a half of media, which is that places were building traffic thinking they were building an audience. And the traffic, at least in that era, was easy, but an audience is really hard. Talk a bit about that.

NILAY PATEL: Yeah first of all, I need to give credit to Casey Newton for that line. That is something — at The Verge, we used to say that to ourselves all the time just to keep ourselves from the temptations of getting cheap traffic. I think most media companies built relationships with the platforms, not with the people that were consuming their content.

NILAY PATEL: The other thing that those business models were predicated upon was you’d get so good at being a supplier to one platform or another with Game of Thrones content or whatever it was that they would pay you money for it directly — that Google would say, this is the Game of Thrones link that most people are clicking on. We ought to pay Vanity Fair for its Game of Thrones content to surface it. Or all of BuzzFeed was we’re going to be so good at going viral on Facebook that Facebook will pay us money.

Every YouTuber gets their wings when they make the video about how they’re mad at YouTube. There’s a woodworking YouTuber that I used to follow, and he just sort of got to the point where he’s like, I hate YouTube. I’m leaving. And it’s like dude, you made videos about jointing wood, like what are you doing? And it’s like his relationship with the platform was so cynical that he was like, I’m moving my business elsewhere. You can sign up for a master class. Those individuals have these very cynical, very commercial relationships with the platforms that the media companies, for some reason, just never hedged. And so they actually do have audiences. And I think media companies need to get way back on the game of having a true audiences.

EZRA KLEIN: This gets to something that does worry me about this phase of A.I. hitting the internet, which is it’s hitting an internet in a moment of decay and weakness. And here, by internet, I mean the sort of content generating internet, and I break that into a couple of categories. The media is very weak right now. The media business we have seen closures left and right, layoffs left and right. I mean, a bunch of players like Vice and BuzzFeed who were believed to be the next generation of juggernauts are functionally gone as news organizations.

And into this comes A.I., and all the money that A.I. seems to bring, and even the A.I. companies might pay you some money for your stuff. Reddit just licensed a bunch of its content as training data to Google.

NILAY PATEL: As a consumer or as a producer?

EZRA KLEIN: As a producer.

NILAY PATEL: The idea that media companies are going to license their stuff to the A.I. companies is just the end of the road that we’ve been on for a long time. We are suppliers to algorithms. OK? And in any normal functioning capitalist economy, supplier margins get squeezed to zero and then maybe we all die.

Like, that’s the game we’ve been playing without saying it for a long time —

EZRA KLEIN: Which I think is why you see The New York Times suing OpenAI, like a real desire to not be in that game again.

NILAY PATEL: You see The New York Times suing OpenAI, but you don’t see them suing Google, you don’t see them de-S.E.O.ing pages across New York Times. Like, they still need the audience from these platforms. And I think there’s a very tense relationship there. The idea that you could sue OpenAI and win some precedent that gives you an enormous amount of leverage over Google I think is a very powerful idea.

And a really interesting facet of the internet we live in today is that most of the copyright law decisions were won by a young, upstart, friendly Google. YouTube exists because it was Google. Like, Viacom famously sued YouTube and they might have won and put it out of business, but Google, the friendly Google company with the water slides in the office, the upstarts that made the product you loved, went and won that case.

Google Books, we’re going to index all the books without asking for permission. They won that case, because they were friendly Google, and the judges were like, look at these cute kids making a cool internet? Like it was new and novel. Google image search — these are all massive copyright decisions that Google won as a startup company run by young people building a new product that the judges were using on their Dell desktops or whatever.

EZRA KLEIN: You don’t want a staunch innovation. Like, that was the big fear in that era. We don’t know what we’re building, and that’s still the thing you hear, and it’s not even untrue. You crack down on copyright and maybe you do staunch innovation. You don’t crack down copyright and maybe you destroy the seed corn of the Informational Commons. It’s very fraught for the copyright judges, but also just for all of us.

NILAY PATEL: Yeah, what are you as a producer on the internet is totally governed by copyright law. Like, a joke at The Verge is a copyright law is the only functional regulation on the internet. The entire internet is just speech, that’s all it is top-to-bottom, it’s speech.

EZRA KLEIN: Taking in the legal sense of the term?

NILAY PATEL: No, in the moral sense of the term. They come to your website and they take your stuff. It’s not a zero sum taking, but they’ve extracted value to create more value for themselves. I think that’s just a moral taking. There’s some permission there that did not occur. Joanna Stern at The Wall Street Journal just interviewed Mira Murati, the C.T.O. of OpenAI, about training data for Sora, the video generator, and Mira said, we just use what’s publicly available. And it’s like yo, that doesn’t make any sense. Like, there are lots of rules about what’s publicly available. Like, you can’t just take stuff because you can link to it on the internet, that’s not how it actually works.

EZRA KLEIN: Let me try to take the argument I hear from the A.I. side of this, which is that there is functionally nothing in human culture and human endeavor that is not trained on all that has come before it — that I, as a person, am trained on all this embedded knowledge in society, that every artist has absorbed, all this other art that the A.I. — I mean, this is just learning. And as long as you are transforming that learning into something else, as long as you are doing something new with that learning, then one, copyright law is not supposed to apply to you in some way or another, although that’s obviously complicated.

NILAY PATEL: I hear this idea all the time, often from the sorts of people in Silicon Valley who say they do first principles thinking — which is one of my favorite phrases, because it just means what if we learn nothing? Like, what if none of the history of the world applied to us and we could start over to our benefit? And that’s usually what that’s code for.

EZRA KLEIN: Judge Patel, if you’re thinking about cases in this area, like, what do you think the answer is here? Is it the sampling model, is it something else? What do you think the right broad strokes resolution is?

NILAY PATEL: Let me stick on the music example for one second, because I think music is really interesting because it’s kind of a closed ecosystem. There’s only so many big music companies. It’s the same lawyers, and the same executives, and the same managers going to the same clearing houses and having the same approaches. We’re going to give you a songwriting credit because we interpolated the bass line of this song into that song, and now here’s some money. And this is the mechanism by which we’ll pay you. The A.I. companies are not a closed ecosystem, it is just a free for all. It’s the open web, it’s a bunch of players.

NILAY PATEL: Well the media industry, but also at some point this is a regulatory question, a question of law. I mean, nothing is stopping Congress from making copyright law designed for the A.I.-era. Nothing is stopping Congress from saying, this is how we think this should work across industries. Not just media, but novelists, but everybody.

NILAY PATEL: Well, there are some things that stop Congress from doing a lot of things. The idea that Congress could pass a massive rewrite of copyright law at this moment in time is pretty far afield.

EZRA KLEIN: But won’t and couldn’t, I do want to make this distinction here. What you’re saying is Congress is too polarized and bitterly divided over everything and can’t do anything and can’t get anything done, and that’s my whole job man, I know. But what I am saying is that, you could write a law like this.

NILAY PATEL: So there is one law. There’s the J.C.P.A., the Journalism Competition Preservation Act, which allows media companies to escape antitrust law and bargain collectively with whoever they wish to bargain with. I don’t know if that’s going to pass, I know there’s a lot of interest in it.

EZRA KLEIN: You and I are both fans of Marshall McLuhan, the media theorist. And he’s got this famous line, ‘the medium is the message.’ And more deeply, what he says is that people, when they see a new medium, they tend to think about the content. For television, it’s the shows, what do you think about this show or that show? For Twitter, the tweets, for a newspaper, the articles. But you have to look behind the content to the actual medium itself to understand what it is trying to tell you.

NILAY PATEL: I have a lot of thoughts about this. I disagree on the basic message. I do think one of the messages of A.I. is that most people make middling work, and middling work is easy to replace. Every email I write is not a great work of art. Like, so much of what we produce just to get through the day is effectively middling. And sure, A.I. should replace a bunch of that. And I think that metaphysical shock comes from the idea that computers shouldn’t be able to do things on their own, and you have a computer that can just do a bunch of stuff for you. And that changes your relationship to the computer in a meaningful way, and I think that’s extremely real.

It’s because this person has made all of these people feel something. The art that has been created by this one very singular individual has captivated all of these people together, because of her story, because of the lyrics, because it means something to them. And I watch people use Midjourney or generate a story with an A.I. tool, and they show the art to you at the end of it, and they’re glowing. Like, look at this wonderful A.I. painting. It’s a car that’s a shark that’s going through a tornado and I told my daughter a story about it.

And I’m like yeah, but this — I don’t want anything to do with this. Like, I don’t care about this. And that happens over and over again. The human creativity is reduced to a prompt, and I think that’s the message of A.I. that I worry about the most, is when you take your creativity and you say, this is actually easy. It’s actually easy to get to this thing that’s a pastiche of the thing that was hard, you just let the computer run its way through whatever statistical path to get there. Then I think more people will fail to recognize the hard thing for being hard. And that’s — truly the message of A.I. is that, maybe this isn’t so hard and there’s something very dangerous to our culture embedded in that.

EZRA KLEIN: I want to put a pin in the hard things, easy things. I’m a little bit obsessed by that and want to come back to it. But first I want to talk about A.I. art for a minute, because I do think when we’re talking about everything that’s going to come on the internet, we’re talking about A.I. art. Obviously, much of it is going to get better. Some of it is not distinguishable.

NILAY PATEL: Have you found an A.I. that can actually write like you?

EZRA KLEIN: I found an A.I. that can mimic certain stylistic tics I have in a way that is better than I think most people could do. I have not found any A.I. that can, in any way, improve my writing for all that you’re constantly told it can. And in fact, the more I try, the worse my writing gets because typically what you have to do to improve your writing is recognize if you’re writing the wrong thing.

NILAY PATEL: This is like the verge of DNA.

EZRA KLEIN: Is it? Yeah, so it comes out in 1935. It’s about the ability to reproduce art. And he says, and I’ll quote it here, “that which whithers in the Age of Mechanical Reproduction is the aura of the work of art.” Then he goes on to say, “by making many reproductions, it substitutes a plurality of copies for a unique existence.”

But I wonder about this with humans. How much of something is just the fact that there’s a human behind it? My Kindroid is no worse at texting me than most people I know. But the fact that my Kindroid has to me is meaningful to me, in the sense that I don’t care if it likes me because there’s no achievement for it to me. The fact that there is a human on the other side of most text messages I send matters. I care about it because it is another mind. The Kindroid might be better in a formulaic way. The kindred might be better in terms of the actual text. I can certainly tune it more to my kind of theoretical liking, but the friction of another person is meaningful to me. Like, I care that my best friend likes me and could choose not to. Is there an aura problem here?

NILAY PATEL: It is so hard to make someone else feel anything other than pain. Like, it’s just like — it’s —

EZRA KLEIN: Christ, that’s the darkest thing I’ve ever heard you say.

NILAY PATEL: Yeah, but I believe it in my soul.

EZRA KLEIN: Really?

NILAY PATEL: Yeah. I think the hardest thing to —

EZRA KLEIN: a really different turn as a show right now. [LAUGHS]:

NILAY PATEL: Maybe —

EZRA KLEIN: You don’t make people laugh, you don’t give them hugs?

NILAY PATEL: No, I think that’s hard. I think that effort is worth it. That’s why I don’t think it’s a dark thing to say. I think the essence of being a good person is pointing your effort at making other people not feel pain. I think bullies make people feel pain because it’s easy. Again, I come back to Taylor Swift in Soldier Field. The thing that was going through my head is, this person is making 60,000 people feel joy, and she’s doing it through art. That is the purpose of art. The purpose of art is to inspire feelings, to inspire emotion.

EZRA KLEIN: Shannon Vallor is a philosopher of technology, and she’s got a book coming out called “The A.I. Mirror,” and I like the way she puts this, because there’s this way that turns is somewhat warped mirror back on ourselves when I was saying a few minutes ago that the message of A.I. is that you’re derivative. That leaves something out. What it’s really saying is that the part of you that often the economy values is derivative, is copyable because we actually ask people a lot of the time to act like they’re machines.

NILAY PATEL: Yeah, I buy that. One of my favorite things that I’ve covered in the past few years is a thing called robotic process automation, which is very funny. Just abstractly, deeply hilarious. There are lots and lots of companies throughout the United States that built computer systems 10, 15 years ago, 20 years ago. Hospital systems are famous for this. They have billing systems. They have buildings full of people who use Microsoft Excel on Windows ’95.

They will sell you a brand new, state of the art computer and it will connect to the keyboard and monitor jack of your old computer, and it will just use the Windows ’95 for you, which is just bonkers. It’s like Rube Goldberg machine of computers using old computers, and then your office full of accountants who knew how to use your old system will go away.

But then A.I. creates the scale problem. What if we do that but instead of some hospital billing system built in the ’90s, it’s just the concept of Microsoft Excel, and now you can just sort of issue a command on your computer and it’ll go use Excel for you and you don’t need an accountant, you don’t need a lawyer.

EZRA KLEIN: What percent of workers are actually asked to poke at the assumptions of their organization, because I worry it’s not as high as you think it is, or implying there. I’m not worried about Taylor Swift. I’m not worried about Nilay Patel. And I don’t just want to make this about wages. That’s a jobs sort of another conversation.

NILAY PATEL: All over the place.

EZRA KLEIN: There’s a huge amount of work like that. And if I felt confident as some of the economists say that we’ll just upmarket people into the jobs where they use more human judgment, David Autor who’s a great trade economist at MIT, just made this argument recently, that what A.I. is going to do is make it possible for more people to exercise judgment and discernment within their work, and I hope he is right. I really hope he is right. But I think a lot of organizations are not set up for a lot of people to use judgment and discernment. They treat a lot of people like machines, and they don’t want them doing things that are complicated and step out of line and poke at the assumptions in the Excel doc. They want the Excel doc ported over without any mistakes.

It seems plausible to me that we’re going to get to that.

NILAY PATEL: Do you think their bosses want to be able to poke at the assumptions though?

EZRA KLEIN: But if you — I mean this is actually something I believe about the whole situation. The economy needs fewer bosses and workers.

EZRA KLEIN: Think about this in the journalist context or the writing context, where I think what A.I. naturally implies that it’s going to do is turn many more people into editors and writers. Because for a lot of content creation that doesn’t require a lot of poking at assumptions, mid-level social media marketing — a lot of people are doing that job right now. But the people doing marketing for a mall —

NILAY PATEL: Yeah, that is the MailChimp example. That is the product that they are building.

EZRA KLEIN: And so what you have then is we used to have a bunch of these social media marketers and now you have one person overseeing a couple systems, like making sure they didn’t say something totally crazy. But you need fewer editors and you need writers. I mean, you know The Verge is structured. You know how The Times is structured. And this is one of my deep worries.

And then this goes to the thing you were getting at earlier, which is one way I think that A.I. could actually not make us more productive, more innovative, is that a lot of the innovation, a lot of the big insights happen when we’re doing the hard thing, when we’re sitting there trying to figure out the first draft, or learn about a thing, or figure out what we’re doing. One of the messages of the medium of A.I. is be efficient. Don’t waste your time on all this. Just tell the system what to do and do it. But there’s a reason I don’t have interns write my first draft for me.

EZRA KLEIN: They could do it. But you don’t get great ideas, or at least not as many of them, editing a piece of work as you do reporting it out, doing the research, writing the first draft. That’s where you do the thinking. And I do think A.I. is built to kind of devalue that whole area of thinking.

NILAY PATEL: We are working on a big story at The Verge right now that I’m very excited about. But there are four of us right now in an argument about whether we should tell that story in chronological order or as a series of vignettes. There is no right answer to this question. There’s just four people who are battling it back and forth.

EZRA KLEIN: I think vignettes.

NILAY PATEL: Yeah. By the way, I’m on team vignette.

EZRA KLEIN: Good man. [LAUGHS]:

NILAY PATEL: My belief is that it’s easier to digest a long story when it’s composed of lots of little stories as opposed to one long one. I’m being outvoted right now — editor in chief. I should replace them all with A.I., just get them out of here. [CHUCKLES] But that is the kind of work that I think makes the end product great. And I think going from good to great is still very human.

EZRA KLEIN: So I want to go back, then, to the internet for a bit, which is I think the presentation we’ve offered is fairly pessimistic. You, when I read and listen to you on this, are — I wouldn’t call it pessimistic. I would say a little excited by the idea of a cleansing fire.

NILAY PATEL: That is very much how I see it. I would add a generational tinge to that, which is I grew up in that weird middle generation between X and millennials. I think temperamentally I’m much more Generation X. But they describe it as they didn’t have computers and then you have computers. You play the Oregon Trail. That’s me on the nose.

I distinctly remember life before computers. It’s an experience that I had quite viscerally. And that shapes my view of these tools. It shapes my view of these companies. Well, there’s a huge generation now that only grew up in this way. There’s a teenage generation right now that is only growing up in this way. And I think their natural inclination is to say, well, this sucks. I want my own thing. I want my own system of consuming information. I want my own brands and institutions.

And I don’t think that these big platforms are ready for that moment. I think that they think they can constantly be information monopolies while they are fending off A.I.-generated content from their own A.I. systems. So somewhere in there all of this stuff does break. And the optimism that you are sensing from me is, well, hopefully we build some stuff that does not have these huge dependencies on platform companies that have no interest at the end of the line except a transaction.

EZRA KLEIN: OK, but you’re telling me how the old thing dies. And I agree with you that at some point the old thing dies. You can feel it. It’s moribund right now. You’re not telling me what the new thing is, and I’m not saying you fully know. But I don’t think the new thing is just a business model that is not as dependent on Meta. I mean, on some level, there’s going to be a lot of A.I. around here.

NILAY PATEL: It’s an audience model. It’s not dependent on these algorithms.

EZRA KLEIN: But is there — I guess one question I have is that, one — I mean, you know where the venture capital is going right now.

EZRA KLEIN: Everything is going to be built with A.I. —

EZRA KLEIN: — laced through every piece of it. And some of it, for all we’re talking about, might be cool, right? I’m not saying you’re mostly going to make great art with A.I. But actually, Photoshop did create a lot of amazing things.

NILAY PATEL: I’m not so sure about that.

NILAY PATEL: I think we’re about to split the internet in two. I think there will be a giant commercial A.I.-infested internet. That’s the platform internet. That’s where it’s going. Moribund, I agree. But it will still be huge. It’s not going away tomorrow. And they will figure out — these are big companies full of smart people with the most technology.

EZRA KLEIN: You sure sound excited about it. [LAUGHS]

NILAY PATEL: Well, I am. I mean, I love technology. This is our — The Verge’s competitive differentiation in the entire media industry is, like, we really like it. And I’m excited to see what they build. I think there’s some really neat things being built. When I think about the information ecosystem, I’m vastly more pessimistic because of the fact that all of these networks are geared to drive you towards a transaction.

EZRA KLEIN: I’m just thinking about this, and the thing that it brings to mind for me is the resurgence of vinyl —

EZRA KLEIN: — and the dominance of streaming platforms. So what I would think of as the music industry of — how many years ago was C.D.s? I don’t actually remember now. But what it did was split into — there’s been a resurgence of vinyl, the sort of analog. It’s a little cool. I actually just bought a record player recently, or was given one by my wonderful partner. But that’s not very big.

EZRA KLEIN: Maybe the social internet dies because, one, we don’t really like it that much anymore anyway, but also because it’s too hard to figure out what’s what. But actually, an internet of A.I. helpers, assistants, friends, et cetera, thrives. And on the other side, you have a real human. I don’t know. But give me more of the Nilay technology side.

EZRA KLEIN: What can A.I. do well? If you were building something or if you were imagining something to be built, what comes after?

NILAY PATEL: By the way, the music industry just released its numbers. Vinyl outsold CDs for the second year running. Double the amount of revenue in vinyl than CDs.

EZRA KLEIN: That’s wild, actually.

NILAY PATEL: It’s crazy. And all of that in total is 11 percent of music industry revenues in ’23 compared to 84 percent of the revenue is streaming. So you are correct. This is a big distinction. People want to buy things, and so they buy one thing that they like. And they consume everything in streaming.

EZRA KLEIN: I think — and this is maybe a little bit of a counterintuitive thought — that this is actually a great time to begin things in media. I think that we have a more realistic sense of the business model and what will actually work. They need to build an audience. They need to build something people will actually pay you for.

I think a lot of the problem right now is things built for another business model that failed are having a lot of trouble transitioning because it’s very, very hard to transition a structure. Now, that doesn’t mean it’s a great business. It’s not what I hoped it would become. It’s not the advertising revenue I hoped we would have. But it’s something.

NILAY PATEL: Yeah. That’s the cleansing fire. That’s the thing I’m excited about. Here’s a new problem in media. Here’s a new problem that’s being created by A.I.

EZRA KLEIN: But my view is that YouTube is the most politically important platform. Everyone wants to talk about TikTok. I think YouTube is much more significant.

NILAY PATEL: Yeah, and they run it really well. They run it as infrastructure. And they talk about it as infrastructure. But it’s weird that we have not built great media company-sized media companies on YouTube’s pipes. We just haven’t done it. So you look at that landscape now and you’re like, well, if I want to do that, if I want to build my own audience, I cannot depend on these companies. I have to be able to do something else.

EZRA KLEIN: Do you think, beyond the media, because not everything online is media —

NILAY PATEL: Let me poke really hard at the true difference between an algorithm that shows you stuff and an algorithm that goes and gets you what you want, because I don’t know that there’s a huge difference in the outcome of those two different processes. So for example, I do not trust the YouTube Kids algorithm. I watch my daughter watch YouTube.

EZRA KLEIN: No, why would you?

NILAY PATEL: It is just a nightmare. I don’t know why we let her do it, but we did. And now we’re in the rabbit hole and that’s life. I mean, she’s five. And I will literally say, are you watching garbage? And she’d be like, I am, because she knows what I think is garbage. She’s much smarter than the YouTube Kids algorithm. And then she’s like, can I watch a little more garbage? This is a real conversation I have with my five-year-old all the time.

EZRA KLEIN: Well, let me push on this for a minute, because for a long time a lot of us have asked people, the social media companies — that I have, I’m sure you have — why don’t you give me access to the dials of the algorithm?

EZRA KLEIN: Right? I don’t want to see things going viral. If there’s a virality scale of 1 to 10, I want to always be at a 6, right?

EZRA KLEIN: But I can’t do that. But one of the interesting things about using the current generation of A.I. models is you actually do have to talk to it like that. I mean, whether I am creating a Replika or a Kindroid or a Character.AI, I have to tell that thing what it is supposed to be, how I want it to talk to me, how I want it to act in the world, what it is interested in, what kinds of expertise it has and does not.

NILAY PATEL: I think that’d be great. I’ve known you for a long time. I think you have a unique ability to articulate exactly what you want and tell it to a computer. [LAUGHS] And you have to scale that idea, right? You have to go to the average — our mothers and say, OK, you have to tell the algorithm exactly what you want. And maybe they’ll get close to it, maybe they won’t, right?

EZRA KLEIN: You don’t feel like mothers are able to tell you what they want?

NILAY PATEL: [LAUGHS] I like that idea a lot. I think fundamentally that is still an A.I. closing the walls around you. And I think the power of the recommendation algorithm is not expressed in virality. It’s actually to help you expand your filter bubble. Here’s a band you’d never heard of before. Here’s a movie you never thought of watching. Here’s an article about a subject that you weren’t interested in before.

EZRA KLEIN: I think it’s a good place to end. Always our final question, for the Nilay Patel recommendation algorithm — what are three books you’d recommend to the audience?

NILAY PATEL: Well, I’m sorry, Ezra, I brought you six.

EZRA KLEIN: Did you really?

NILAY PATEL: Is that allowed?

EZRA KLEIN: Did you actually bring six?

NILAY PATEL: I didn’t bring six physical books, but I have six recommendations for you.

EZRA KLEIN: Damn. All right, go through them quick, man.

NILAY PATEL: They’re in two categories. One is the three books that I thought of and three books from Verge people that if people are interested in these ideas are important.

So the first one is “The Conquest of Cool” by Thomas Frank, one of my favorite books of all time. It is about how advertising agencies in the ’60s co-opted the counterculture and basically replaced counterculture in America. I’ve thought about this a lot because I’m constantly wondering where the punk bands and rage against the machines of 2024 are. And the answer is that they’re the mainstream culture. It’s very interesting. Love that book. It explains, I think, a lot about our culture.

Third, I love the band New Order. I know you’re a music fan, so I brought you a music recommendation. It’s “Substance: Inside New Order” by Peter Hook, who is the bassist of New Order. This band hates each other. They broke up acrimoniously, so the book is incredibly bitchy. It’s just a lot of shit-talking about the ’80s. It’s great.

EZRA KLEIN: Nilay Patel, you’re saving the internet through blogging again.

NILAY PATEL: Thanks, man.

EZRA KLEIN: This episode of “The Ezra Klein Show” was produced by Claire Gordon. Fact-checking by Michelle Harris with Kate Sinclair and Mary Marge Locker. Our senior engineer is Jeff Geld. We’ve got additional mixing by Isaac Jones and Efim Shapiro. Our senior editor is Claire Gordon. The show’s production team also includes Annie Galvin, Rollin Hu and Kristin Lin. We have original music by Isaac Jones. Audience strategy by Kristina Samulewski and Shannon Busta. The executive producer of New York Times Opinion Audio is Annie-Rose Strasser. And special thanks here to Sonia Herrero.

Advertisement