Artificial Intelligence And Its Implications

2y ago
520.22 KB
75 Pages
Last View : 2m ago
Last Download : 1y ago
Upload by : Hayden Brunner

Artificial Intelligence and Its Implications .

The Economics ofArtificial Intelligence

National Bureau ofEconomic ResearchConference Report

The Economics ofArtificial Intelligence:An AgendaEdited byAjay Agrawal, Joshua Gans,and Avi GoldfarbThe University of Chicago PressChicago and London

The University of Chicago Press, Chicago 60637The University of Chicago Press, Ltd., London 2019 by the National Bureau of Economic Research, Inc.All rights reserved. No part of this book may be used or reproducedin any manner whatsoever without written permission, except in thecase of brief quotations in critical articles and reviews. For moreinformation, contact the University of Chicago Press, 1427 E. 60th St.,Chicago, IL 60637.Published 2019Printed in the United States of America28 27 26 25 24 23 22 21 20 191 2 3 4 5ISBN-13: 978-0-226-61333-8 (cloth)ISBN-13: 978-0-226-61347-5 (e-book)DOI: 0001Library of Congress Cataloging-in-Publication DataNames: Agrawal, Ajay, editor. Gans, Joshua, 1968– editor. Goldfarb,Avi, editor.Title: The economics of artificial intelligence : an agenda / AjayAgrawal, Joshua Gans, and Avi Goldfarb, editors.Other titles: National Bureau of Economic Research conference report.Description: Chicago ; London : The University of Chicago Press,2019. Series: National Bureau of Economic Research conferencereport Includes bibliographical references and index.Identifiers: LCCN 2018037552 ISBN 9780226613338 (cloth : alk.paper) ISBN 9780226613475 (ebook)Subjects: LCSH: Artificial intelligence—Economic aspects.Classification: LCC TA347.A78 E365 2019 DDC 338.4/70063—dc23LC record available at This paper meets the requirements of ANSI/NISO Z39.48-1992(Permanence of Paper).DOI: 10.7208/chicago/9780226613475.003.0014

National Bureau of Economic ResearchOfficersKaren N. Horn, chairJohn Lipsky, vice chairJames M. Poterba, president and chiefexecutive officerRobert Mednick, treasurerKelly Horak, controller and assistantcorporate secretaryAlterra Milone, corporate secretaryDenis Healy, assistant corporate secretaryDirectors at LargePeter C. AldrichElizabeth E. BaileyJohn H. BiggsJohn S. ClarkesonKathleen B. CooperCharles H. DallaraGeorge C. EadsJessica P. EinhornMohamed El-ErianDiana FarrellJacob A. FrenkelRobert S. HamadaPeter Blair HenryKaren N. HornLisa JordanJohn LipskyLaurence H. MeyerKaren MillsMichael H. MoskowAlicia H. MunnellRobert T. ParryJames M. PoterbaJohn S. ReedMarina v. N. WhitmanMartin B. ZimmermanDirectors by University AppointmentTimothy Bresnahan, StanfordPierre-André Chiappori, ColumbiaAlan V. Deardorff, MichiganEdward Foster, MinnesotaJohn P. Gould, ChicagoMark Grinblatt, California, Los AngelesBruce Hansen, Wisconsin–MadisonBenjamin Hermalin, California, BerkeleySamuel Kortum, YaleGeorge Mailath, PennsylvaniaMarjorie B. McElroy, DukeJoel Mokyr, NorthwesternCecilia Rouse, PrincetonRichard L. Schmalensee, MassachusettsInstitute of TechnologyIngo Walter, New YorkDavid B. Yoffie, HarvardDirectors by Appointment of Other OrganizationsJean-Paul Chavas, Agricultural and AppliedEconomics AssociationMartin J. Gruber, American FinanceAssociationPhilip Hoffman, Economic HistoryAssociationArthur Kennickell, American StatisticalAssociationJack Kleinhenz, National Association forBusiness EconomicsRobert Mednick, American Institute ofCertified Public AccountantsPeter L. Rousseau, American EconomicAssociationGregor W. Smith, Canadian EconomicsAssociationWilliam Spriggs, American Federationof Labor and Congress of IndustrialOrganizationsBart van Ark, The Conference BoardDirectors EmeritiGeorge AkerlofJagdish BhagwatiDon R. ConlanRay C. FairFranklin FisherSaul H. HymansRudolph A. OswaldAndrew PostlewaiteJohn J. SiegfriedCraig Swan

Relation of the Directors to theWork and Publications of theNational Bureau of Economic Research1. The object of the NBER is to ascertain and present to the economics profession, and to thepublic more generally, important economic facts and their interpretation in a scientific mannerwithout policy recommendations. The Board of Directors is charged with the responsibility ofensuring that the work of the NBER is carried on in strict conformity with this object.2. The President shall establish an internal review process to ensure that book manuscripts proposed for publication DO NOT contain policy recommendations. This shall apply both to theproceedings of conferences and to manuscripts by a single author or by one or more co-authorsbut shall not apply to authors of comments at NBER conferences who are not NBER affiliates.3. No book manuscript reporting research shall be published by the NBER until the Presidenthas sent to each member of the Board a notice that a manuscript is recommended for publication and that in the President’s opinion it is suitable for publication in accordance with the aboveprinciples of the NBER. Such notification will include a table of contents and an abstract orsummary of the manuscript’s content, a list of contributors if applicable, and a response formfor use by Directors who desire a copy of the manuscript for review. Each manuscript shallcontain a summary drawing attention to the nature and treatment of the problem studied andthe main conclusions reached.4. No volume shall be published until forty-five days have elapsed from the above notificationof intention to publish it. During this period a copy shall be sent to any Director requestingit, and if any Director objects to publication on the grounds that the manuscript containspolicy recommendations, the objection will be presented to the author(s) or editor(s). In caseof dispute, all members of the Board shall be notified, and the President shall appoint an adhoc committee of the Board to decide the matter; thirty days additional shall be granted forthis purpose.5. The President shall present annually to the Board a report describing the internal manuscript review process, any objections made by Directors before publication or by anyone afterpublication, any disputes about such matters, and how they were handled.6. Publications of the NBER issued for informational purposes concerning the work of theBureau, or issued to inform the public of the activities at the Bureau, including but not limitedto the NBER Digest and Reporter, shall be consistent with the object stated in paragraph 1.They shall contain a specific disclaimer noting that they have not passed through the reviewprocedures required in this resolution. The Executive Committee of the Board is charged withthe review of all such publications from time to time.7. NBER working papers and manuscripts distributed on the Bureau’s web site are not deemedto be publications for the purpose of this resolution, but they shall be consistent with the objectstated in paragraph 1. Working papers shall contain a specific disclaimer noting that they havenot passed through the review procedures required in this resolution. The NBER’s web siteshall contain a similar disclaimer. The President shall establish an internal review process toensure that the working papers and the web site do not contain policy recommendations, andshall report annually to the Board on this process and any concerns raised in connection with it.8. Unless otherwise determined by the Board or exempted by the terms of paragraphs 6and 7, a copy of this resolution shall be printed in each NBER publication as described inparagraph 2 above.

ContentsI.AcknowledgmentsxiIntroductionAjay Agrawal, Joshua Gans, and Avi Goldfarb1AI as a GPT1. Artificial Intelligence and the ModernProductivity Paradox: A Clash ofExpectations and StatisticsErik Brynjolfsson, Daniel Rock, andChad SyversonComment: Rebecca Henderson232. The Technological Elements ofArtificial IntelligenceMatt Taddy613. Prediction, Judgment, and Complexity:A Theory of Decision-Making andArtificial IntelligenceAjay Agrawal, Joshua Gans, and Avi GoldfarbComment: Andrea Prat894. The Impact of Artificial Intelligenceon Innovation: An Exploratory AnalysisIain M. Cockburn, Rebecca Henderson,and Scott SternComment: Matthew Mitchell115vii

viiiContents5. Finding Needles in Haystacks: ArtificialIntelligence and Recombinant GrowthAjay Agrawal, John McHale,and Alexander Oettl6. Artificial Intelligence as the Next GPT:A Political-Economy PerspectiveManuel TrajtenbergII.149175Growth, Jobs, and Inequality7. Artificial Intelligence, Income, Employment,and MeaningBetsey Stevenson1898. Artificial Intelligence, Automation, and WorkDaron Acemoglu and Pascual Restrepo1979. Artificial Intelligence and Economic GrowthPhilippe Aghion, Benjamin F. Jones, andCharles I. JonesComment: Patrick Francois23710. Artificial Intelligence and Jobs:The Role of DemandJames Bessen11. Public Policy in an AI EconomyAustan Goolsbee12. Should We Be Reassured If Automationin the Future Looks Like Automationin the Past?Jason Furman29130931713. R&D, Structural Transformation,and the Distribution of IncomeJeffrey D. Sachs32914. Artificial Intelligence and Its Implicationsfor Income Distribution and UnemploymentAnton Korinek and Joseph E. Stiglitz34915. Neglected Open Questions in theEconomics of Artificial IntelligenceTyler Cowen391

ContentsIII.Machine Learning and Regulation16. Artificial Intelligence, Economics, andIndustrial OrganizationHal VarianComment: Judith Chevalier39917. Privacy, Algorithms, and Artificial IntelligenceCatherine Tucker42318. Artificial Intelligence and Consumer PrivacyGinger Zhe Jin43919. Artificial Intelligence and International TradeAvi Goldfarb and Daniel Trefler46320. Punishing Robots: Issues in the Economicsof Tort Liability and Innovation inArtificial IntelligenceAlberto Galasso and Hong LuoIV.ix493Machine Learning and Economics21. The Impact of Machine Learningon EconomicsSusan AtheyComment: Mara Lederman50722. Artificial Intelligence, Labor, Productivity,and the Need for Firm-Level DataManav Raj and Robert Seamans55323. How Artificial Intelligence and MachineLearning Can Impact Market DesignPaul R. Milgrom and Steven Tadelis56724. Artificial Intelligence andBehavioral EconomicsColin F. CamererComment: Daniel KahnemanContributorsAuthor IndexSubject Index587611615625

AcknowledgmentsThis volume contains chapters and ideas discussed at the first NBER Conference on the Economics of Artificial Intelligence, held in September 2017in Toronto. We thank all the authors and discussants for their contributions.Funds for the conference and book project were provided by the Sloan Foundation, the Canadian Institute for Advanced Research, and the CreativeDestruction Lab at the University of Toronto. At the Sloan Foundation,Danny Goroff provided guidance that improved the overall agenda. TheNBER digitization initiative, under the leadership of Shane Greenstein, wasa key early supporter. We thank our dean, Tiff Macklem. In addition, JimPoterba at the NBER has been generous, giving us the flexibility needed tobring this project together. Special thanks are due to Rob Shannon, DenisHealy, Carl Beck, and Dawn Bloomfield for managing the conference andlogistics and to Helena Fitz-Patrick for guiding the book through the editorial process. Finally we thank our families, Gina, Natalie, Rachel, Amelia,Andreas, Belanna, Ariel, Annika, Anna, Sam, and Ben.xi

14Artificial Intelligence and ItsImplications for IncomeDistribution and UnemploymentAnton Korinek and Joseph E. Stiglitz14.1IntroductionThe introduction of artificial intelligence (AI) is the continuation of along process of automation. Advances in mechanization in the late nineteenth and early twentieth centuries automated much of the physical laborperformed by humans. Advances in information technology in the mid- tolate twentieth century automated much of the standardized data processingthat used to be performed by humans. However, each of these past episodesof automation left large areas of work that could only be performed byhumans.Some propose that advances in AI are merely the latest wave in this longprocess of automation, and may in fact generate less economic growth thanpast technological advances (see, e.g., Gordon 2016). Others, by contrast,emphasize that AI critically differs from past inventions: as artificial intelligence draws closer and closer to human general intelligence, much of humanlabor runs the risk of becoming obsolete and being replaced by AI in alldomains. In this view, progress in artificial intelligence is not only a continuaAnton Korinek is associate professor of economics and business administration at theUniversity of Virginia and Darden GSB and a research associate of the National Bureau ofEconomic Research. Joseph E. Stiglitz is University Professor at Columbia University and aresearch associate of the National Bureau of Economic Research.This chapter was prepared as a background paper for the NBER conference The Economicsof Artificial Intelligence. We would like to thank our discussant Tyler Cowan as well as JayantRay and participants at the NBER conference for helpful comments. We also acknowledgeresearch assistance from Haaris Mateen as well as financial support from the Institute for NewEconomic Thinking (INET) and the Rewriting the Rules project at the Roosevelt Institute,supported by the Ford, Open Society, and the Bernard and Irene Schwartz Foundations. Foracknowledgments, sources of research support, and disclosure of the authors’ material financial relationships, if any, please see

350Anton Korinek and Joseph E. Stiglitztion, but the culmination of technological progress; it could lead to a courseof history that is markedly different from the implications of previous wavesof innovation, and may even represent what James Barrat (2013) has termed“Our Final Invention.”No matter what the long- run implications of AI are, it is clear that it hasthe potential to disrupt labor markets in a major way, even in the short andmedium run, affecting workers across many professions and skill levels.1 Themagnitude of these disruptions will depend on two important factors: thespeed and the factor bias of progress in AI.On the first factor, measured productivity has increased rather slowlyin recent years, even as the world seems to be captured by AI fever.2 If AIrelated innovations enter the economy at the same slow pace as suggestedby recent productivity statistics, then the transition will be slower than, forexample, the wave of mechanization in the 1950– 1970s, and the resulting disruptions may not be very significant. However, there are three possible alternatives: First, some suggest that productivity is significantly undermeasured,for example, because quality improvements are not accurately captured. Thebest available estimates suggest that this problem is limited to a few tenthsof a percentage point (see, e.g., the discussion in Groshen et al. [2017]).Furthermore, there are also unmeasured deteriorations in productivity, forexample, declines in service quality as customer service is increasingly automated. Second, the aggregate implications of progress in AI may follow adelayed pattern, similar to what happened after the introduction of computers in the 1980s. Robert Solow (1987) famously quipped that “you can see thecomputer age everywhere but in the productivity statistics.” It was not untilthe 1990s that a significant rise in aggregate productivity could be detected,after sustained investment in computers and a reorganization of businesspractices had taken place. Third, it is of course possible that a significantdiscontinuity in productivity growth occurs, as suggested, for example, byproponents of a technological singularity (see, e.g., Kurzweil 2005).On the second factor, the disruptions generated by AI- related innovations depend on whether they are labor- saving, using the terminology ofHicks (1932), that is, whether at a given wage the innovations lead to lessdemand for labor. Some suggest that artificial intelligence will mainly assisthumans in being more productive, and refer to such new technologies asintelligence-assisting innovation (IA), rather than AI. Although we agreethat most AI- related innovations are likely to be complementary to at leastsome jobs, we believe that in taking a broader perspective, progress in AI1. For example, Frey and Osborne (2017) warn that 47 percent of jobs in the US economy areat risk of being automated by advances in AI- related fields. Areas in which human intelligencehas recently become inferior to artificial intelligence include many applications of radiology,trading in financial markets, paralegal work, underwriting, driving, and so forth.2. For example, Google Trends reveals that search interest in the topic “artificial intelligence”has quadrupled over the past four years.

AI and Its Implications for Income Distribution and Unemployment351is more likely to substitute for human labor, or even to replace workersoutright, as we will assume in some of our formal models below.We believe that the primary economic challenge posed by the proliferationof AI will be one of income distribution. We economists set ourselves tooeasy a goal if we just say that technological progress can make everybodybetter off —we also have to say how we can make this happen. This chapter isan attempt to do so by discussing some of the key economic research issuesthat this raises.3In section 14.2 of this chapter, we provide a general taxonomy of therelationship between technological progress and welfare. We first observethat in a truly first- best economy—in which complete risk markets are available before a veil of ignorance about innovations is lifted—all individualswill share in the benefits of technological progress. However, since the realworld does not correspond to this ideal, redistribution is generally needed toensure that technological progress generates Pareto improvements. If markets are perfect and redistribution is costless, it can always be ensured thattechnological progress makes everybody better off. The same result holdsif the costs of redistribution are sufficiently low. In all these cases, therecan be political unanimity about the desirability of technological progress.However, if redistribution is too costly, it may be impossible to compensate the losers of technological progress, and they will rationally opposeprogress. Even worse, if the economy suffers from market imperfections,technological progress may actually move the Pareto frontier inwards, thatis, some individuals may necessarily be worse off. Finally, we observe thatthe first welfare theorem does not apply to the process of innovation, and asa result, privately optimal innovation choices may move the Pareto frontierinward.In section 14.3, we decompose the mechanisms through which innovationleads to inequality into two channels. First, inequality rises because innovators earn a surplus. Unless markets for innovation are fully contestable, thesurplus earned by innovators is generally in excess of the costs of innovationand includes what we call innovator rents. We discuss policies that affect thesharing of such rents, such as antitrust policies and changes in intellectualproperty rights. The second channel is that innovations affect market prices;they change the demand for factors such as different types of labor andcapital, which affects their prices and generates redistributions. For example,AI may reduce a wide range of human wages and generate a redistributionto entrepreneurs. From the perspective of our first- best benchmark withcomplete insurance markets, these factor price changes represent pecuniary externalities. We discuss policies to counter the effects of the resultingfactor price changes.3. An important, and maybe even more difficult, complementary question, which is beyondthe scope of this chapter, is to analyze the political issues involved.

352Anton Korinek and Joseph E. StiglitzIn section 14.4, we develop a simple formal model of worker- rep

Artificial Intelligence and Its Implications . The Economics of . Artifi cial Intelligence, Economics, and Industrial Organization 399 Hal Varian Comment: Judith Chevalier 17. . ference on the Economics of Artifi cial Intelligence, held in September 2017 in Toronto.

Related Documents:

Artificial Intelligence and Its Military Implications China Arms Control and Disarmament Association July 2019 What Is Artificial Intelligence? Artificial intelligence (AI) refers to the research and development of the theories, methods, technologies, and application systems for

Artificial Intelligence and Its Implications on Future Submarine Warfare With reference to any historical example, what are the implications for the future of the Royal Australian Navy? This essay will discuss artificial intelligence (AI) a

Artificial Intelligence and Its Implications for Income Distribution and Unemployment Anton Korinek and Joseph E. Stiglitz NBER Working Paper No. 24174 December 2017 JEL No. D63,E64,O3 ABSTRACT Inequality is one of the main challenges posed by the proliferation of artificial intelligence (

and artificial intelligence expert, joined Ernst & Young as the person in charge of its global innovative artificial intelligence team. In recent years, many countries have been competing to carry out research and application of artificial intelli-gence, and the call for he use of artificial

Jan 28, 2019 · Artificial intelligence and machine learning: implications for insurers . Artificial intelligence (AI) is rapidly becoming an important technology in the insurance industry, as firms and their regulators investigate its

This paper is about examining the history of artificial intelligence from theory to practice and from its rise to fall, highlighting a few major themes and advances. 'Artificial' intelligence The term artificial intelligence was first coined by John McCarthy in 1956 when he held the first academic conference on the subject.

Jan 25, 2018 · Third, AI has a perception problem because of its very name. The term “artificial intelligence” implies that the technology has or soon will have intelligence akin to human intelligence. And, ominously, that this will quickly transform in to artificial super -intelligence

hardware IP re-use and consistency accross product families and higher level programming language makes the development job far more convenient when dealing with the STM32 families. HIGH-PERFORMANCE HIGH DEGREE OF INTEGRATION AND RICH CONNECTIVITY STM32H7: highest performance STM32 MCUs with advanced features including DSP and FPU instructions based on Cortex -M7 with 1 to 2 Mbytes of .