The Economics of Artificial Intelligence

Home > Other > The Economics of Artificial Intelligence > Page 105
The Economics of Artificial Intelligence Page 105

by Ajay Agrawal


  Telang, R., 444, 450, 457, 457n42

  Somanchi, S., 444

  Teodoridis, F., 150, 161

  Song, J., 30

  Tetenov, A., 527

  Sonnenschein, H., 589n1

  Thaler, R., 604

  Sopher, B., 590n2

  Thomas, K., 443

  Sorensen, O., 161

  Thomas, P., 528

  Spencer, B. J., 473

  Thompson, W. R., 82

  Spezio, M., 594n10

  Thrnton, B., 598

  Spiegel, M., 589n1

  Tibshirani, R., 93, 525, 527

  Spiegel, Y., 412

  Tirole, J., 106, 112, 264, 268, 269

  Spier, K., 494, 499, 502

  Topol, E. J., 94, 95

  Spiess, J., 169n8, 511, 512

  Tory, J., 485n23

  Spindler, M., 522

  Toulis, P., 80

  Srivastava, N., 80

  Trajtenberg, M., 4, 39, 116, 119, 132, 150,

  Stahl, D. O., 593

  169, 176n2

  Stantcheva, S., 360

  Trefl er, D., 465, 467, 479, 485n23

  Stein, A., 494

  Troske, K. R., 562

  Stern, A., 325

  Tschantz, M. C., 433

  Stern, S., 122, 141

  Tucker, C. E., 148, 425, 426, 427, 428, 429,

  Stevenson, B., 194n5

  448, 449, 482, 483, 483n19

  Stigler, G. J., 439, 582

  Turing, A. M., 123, 385

  Stiglitz, J. E., 30n9, 354, 360, 362, 364,

  Tuzel, S., 210

  364n11, 366n14, 366n15, 368n19, 369,

  Tversky, A., 596

  370, 370n21, 371, 373, 376n27, 376n28,

  378, 412

  Ugander, J., 538

  Stillwell, D., 429

  Uyarra, E., 480

  Stinchcombe, M., 74

  Stivers, A., 442n4

  Vadlamannati, K. C., 483

  Stole, L., 268

  Valentinyi, Á., 204, 241n4

  Strehl, A., 527

  Van der Laan, M. J., 522, 527

  Streitwieser, M. L., 562

  Van Seijen, H., 62, 84

  Strotz, R. H., 427

  Varian, H. R., 93, 310, 410, 413, 424, 425,

  Stucke, M. E., 414

  440, 511

  Stutzman, F., 451

  Venables, A. J., 480

  Sufi , A., 208

  Vesteger, M., 420

  Summers, L. H., 175

  Vickrey, W., 567

  Sutskever, I., 68, 74, 125

  Vijverberg, W. P. M., 483

  Sutton, J., 467, 589n1

  Vincent, N., 270

  Swaffi

  eld, J., 293n2

  Vines, P., 446

  Swaminathan, A., 528

  Vinge, V., 238, 253, 253n12, 373n23

  Sweeney, L., 433

  Viscusi, K., 496, 498

  Swire, P. P., 441n2

  Vishnu, A., 151n1, 168

  Syverson, C., 23, 29, 43, 210, 319, 320

  Von Hippel, E., 499

  Author Index 623

  Von Weizacker, C. C., 376

  Wright, J. R., 594

  Von Winterfeldt, D., 595

  Wu, L., 43

  Vuong, Q., 514

  Xiao, M., 593n8

  Wager, S., 523, 525, 527, 528

  Xie, D., 241n4, 295

  Wagman, L., 416, 424, 441n2, 448, 459,

  Xu, H., 577

  483n19

  Wallach, I., 116

  Yakovlev, E., 538

  Wan, M., 534

  Yang, S., 42, 47

  Wang, J., 475, 477, 594n10

  Yellen, J., 378

  Warren, S. D., 431, 459

  Yeomans, M., 517

  Waseem, M., 270

  Yildirim, P., 107n9

  Wattal, S., 450

  Yu, M., 465, 479

  Weil, D., 459

  Yudkowsky, E., 253n12

  Weingast, B. R., 577

  Weiss, A., 354

  Zanna, L.-F., 373n24

  Weitzman, M., 149, 151, 157, 171, 171n9,

  Zeevi, A., 529

  172, 241

  Zeileis, A., 525

  Western, B., 322

  Zeira, J., 152, 198, 212n7, 238, 239, 241

  Westlake, S., 18

  Zervas, G., 577n5

  Whinston, M. D., 148

  Zhang, M. B., 210

  White, H., 74, 511

  Zheng, Y., 329

  Williams, H., 122, 364n11

  Zhou, D., 529, 581

  Williams, R., 124

  Zhou, X., 580

  Wilson, P. W., 593

  Zhu, X., 471

  Winter, S., 161

  Zierahn, U., 321

  Wolfers, J., 195n5

  Zubizarreta, J. R., 523

  Wooldridge, J. M., 522

  Zweimüller, J., 295

  Wright, G., 42n19

  Zwiebel, J., 268

  Wright, G. C., 303

  Subject Index

  Note: Page numbers followed by “f ” or “t” refer to fi gures or tables, respectively.

  adoption, technological: implications of

  351; inequality and, 320– 23; internal

  speed of, for job market and inequality,

  agreements and, 463; international

  310– 12

  macroeconomics and, 488; knowledge

  adversarial artifi cial intelligence, 401

  externalities and, 470– 71; likely produc-

  aggregate productivity statistics, technolo-

  tivity eff ects of, and acceleration of, 45–

  gies and, 26– 28

  46; longer- term prospects of, 381– 86;

  AI. See artifi cial intelligence (AI)

  market structure and, 262– 63; as “next

  AlphaGo (Google), 63

  big thing,” 175; political economy of,

  Amazon Go concept, 67

  11, 394– 95; prediction costs and, 92– 93;

  applied artifi cial intelligence, 208

  privacy and, 425– 26; privacy concerns

  artifi cial intelligence (AI), 1; and automa-

  and, 423– 24; for promoting trust in

  tion of production, 239– 50; as basis

  online marketplaces, 576– 81; recent

  for learning, 120– 21; benefi t of more,

  approach to, 93; for reducing search

  318– 20; bibliometric data on evolution

  frictions, 581– 83; return of Malthus

  of, 128– 32; capital shares and, 270– 74,

  and, 381– 86; revolution, international

  272– 73f; in context, 84– 85; defi ned,

  eff ects of, 393– 94; in Schumpeterian

  3– 4, 62– 67, 93, 122, 237, 468; econo-

  model with creative destruction, 276–

  mies of scale from data and, 468– 69;

  79; sectoral reallocation and, 263– 64;

  economies of scale from overhead

  statistics on, 465– 66, 466t; studies on

  of developing AI capabilities, 469–

  economic eff ect of, 556– 58; theory of

  70; evolution of, 122– 25; expected

  privacy in economics and, 424– 26; as

  productivity eff ects of, 45– 46; fi rm

  tool, 16– 17; world’s largest companies

  organization and, 264– 70; future of

  and exposure to, 465– 67, 467t. See also

  research on economics of, 17; as general

  machine learning (ML)

  purpose technology, 4– 7, 39– 41; in idea

  artifi cial intelligence capital, measuring,

  production function, 250– 52; impact

  46– 50

  of, on innovation, 125– 28; impact of

  artifi cial intelligence– general purpose tech-

  long- term decline in labor force partici-

  nology (GPT) era: education strategies

  pation rate and,
323– 25; implications

  for, 179– 82; human- enhancing innova-

  of, 349– 53; income distribution and,

  tions vs. human- replacing innovations

  625

  626 Subject Index

  artifi cial intelligence– general purpose tech-

  consumer attitude, 448– 49

  nology (GPT) era ( continued)

  consumer privacy: challenging issues in,

  for, 184– 85; professionalization of

  457– 59; consumer attitude and, 448– 49;

  personal services strategies for, 182– 84;

  consumer risk and, 443– 48; data risk

  top skills required for employment in,

  and, 442– 43; nature of problem of, 442;

  180– 81, 181t

  policy landscape in United States, 454–

  artifi cial intelligence revolution, inter-

  57; supply side actions and, 450– 54.

  national eff ects of, 393– 94

  See also privacy

  Atomwise, case of, 115– 16, 120, 154

  consumer surplus, 11; distribution of,

  automatic teller machines (ATMs), security

  391– 93

  policy and, 416

  contracting, 106– 7

  automation, 3– 4, 105– 6; basic model, 336–

  convolutional neural networks (CNNs),

  41; Baumol’s cost disease and, 241– 50;

  75– 76, 75n6

  to date, and capital shares, 270– 74;

  cooperation, evolution of, 414

  decline in labor share and, 329– 31;

  cost disease, Baumol’s, 8– 9, 238– 39; auto-

  deepening of, 198, 204– 5, 216– 17; eco-

  mation and, 241– 50

  nomic adjustment and, 208– 9; employ-

  creative destruction, 260– 61

  ment and, 190– 91; excessive, 224– 26;

  model of, 211– 14; of production, and

  data, 61; acquisition methods, 403– 4; de-

  artifi cial intelligence, 239– 50; produc-

  creasing marginal returns of, 406, 407f;

  tivity and, 210– 11; sector of economy

  economics of, 14; importance of, 13–

  aff ected by, 330– 33; studies on employ-

  14; important characteristics of, 404– 6;

  ment on, 555– 58; wages and, 200– 211;

  localization rules, trade policies and,

  winners, 190; work and, 200– 211; Zeira

  485– 86; persistence of predictive power

  model of growth and, 239– 41

  of, 427– 28; privileged access to govern-

  average treatment eff ects, 522– 24

  ment, trade policies and, 486– 87;

  types of, and generation of spillovers,

  bandits (algorithms), problem of, 528– 29

  431– 34

  Baumol’s cost disease. See cost disease,

  data access, 405– 6

  Baumol’s

  data generation, as pillar of artifi cial intel-

  BenchSci search technology, 153

  ligence, 62f, 65– 66

  buy/ make decisions (fi rm boundaries),

  data ownership, 405– 6

  107– 8

  data persistence, 426– 27; predictive power

  and, 427– 28

  capital accumulation, 198, 204– 5, 216

  data pipeline, 402

  capital shares, and automation to date,

  data pyramid, 404, 405f

  270– 74

  data repurposing, 428– 31

  causal inference, new literature on, 519– 34

  data security: challenging issues in, 457–

  Children’s Online Privacy Protection Act of

  59; policy landscape in United States,

  1998 (COPPA), 454

  454– 57

  cloud- computing facilities, 402– 3

  data spillovers, 431– 34

  cluster policies, 480– 81

  data warehouses, 402– 3

  clusters, regional, theory of, 479

  decision- making, baseline model for, 95–

  CNNs (convolutional neural networks), 75–

  103; complexity and, 103– 8

  76, 75n6

  deepening of automation, 198, 204– 5,

  collusion, strategies for facilitating, 413– 14

  216– 17

  combinatorial- based knowledge production

  Deep Genomics, 154

  function, 154– 61; potential uses of new,

  deep learning, 3, 71– 77, 400; as general pur-

  170– 71; with team production, 161– 67

  pose invention in method of invention,

  Communications Act (1986), 456

  139– 43; as general purpose technology,

  competition policy, innovations and, 141– 43

  133– 39; as new discovery tool, 167– 69;

  complexity, 103– 4

  patent systems and, 142

  Subject Index 627

  deep learning networks, 94

  GPT. See general purpose technology (GPT)

  deep learning techniques, 25

  Gramm- Leach- Bliley Act (GLBA), 454

  deep neural networks (DNNs), 25– 26, 61,

  growth: impact of artifi cial intelligence on,

  63; structure in, 76– 77

  7– 9. See also economic growth

  demand, importance of, 301– 2

  destruction, creative, 260– 61

  Health Insurance Portability and Account-

  diff erence- in-diff erence models, 530– 31

  ability Act of 1996 (HIPAA), 454

  digital information, 334– 35

  HEI (human- enhancing innovations),

  direct network eff ects, 412

  184– 85

  displacement eff ect, 8, 198, 208, 214

  heterogeneous treatment eff ects, 524– 28

  DNNs. See deep neural networks (DNNs)

  hierarchical Poisson factorization, 510

  domain structure, as pillar of artifi cial intel-

  HIPAA. See Health Insurance Portabil-

  ligence, 62f, 63– 65

  ity and Accountability Act of 1996

  double machine learning, 523– 24

  (HIPAA)

  HRI (human- replacing innovations), 184– 85

  economic growth: artifi cial intelligence and,

  human- enhancing innovations (HEI), 184– 85

  262– 70; prospects for technology-

  human- replacing innovations (HRI), 184– 85

  driven, 149– 53; Zeira model of auto-

  mation and, 239– 41. See also growth

  idea production function, artifi cial intelli-

  economics, impact of machine learning on

  gence in, 250– 52

  practice of, 15– 16

  implementation/ restructuring lags, as expla-

  education, factory model of, 180

  nation for Solow paradox, 31– 36

  Electronic Communications Privacy Act of

  incentive auctions, machine learning and,

  1986: (ECPA), 456

  569– 76

  employment: automation and, 190– 91;

  income, artifi cial intelligence and, 189– 90

  levels of, and new technologies, 220– 21;

  income distribution: artifi cial intelligence

  long- run vs. short run, 192– 94; studies

  and, 351; impact of AI on, 11

  on automation and, 555– 57; work out-

  income inequality: artifi cial intelligence and,

  side of, 194– 95

  320– 23; impact of AI on, 7– 8, 11– 12;

  evolution of cooperation, 414

  speed of technological adoption and,

  310– 12

  factory model of education, 180

  income redistribution, political e
conomy of,

  Federal Trade Commission (FTC), 454– 55

  394– 95

  fi rm boundaries (make/ buy decisions),

  indirect network eff ects, 412

  107– 8

  industrial regulation, trade policies and, 487

  fi rm- level data: need for, 558– 59; strategies

  inequality. See income inequality

  for collecting, 561– 62

  information technology (IT), 24

  fi rm- level research questions, 560– 61

  innovation, 115– 18; competition policy and,

  fi rms: artifi cial intelligence and, 262– 70;

  141– 43; early stage, 121– 22; impact

  impact of machine learning on, 12

  of artifi cial intelligence on, 125– 28;

  institutions and, 141– 43; management

  general purpose machine learning (GPML),

  and organization of, 140– 41; product

  67– 71

  liability and, 494

  general purpose technology (GPT), 2, 65,

  institutions, innovations and, 141– 43

  119– 20, 169– 70; artifi cial intelligence

  intelligence- assisting innovation (IA),

  as, 4– 7, 39– 41; deep learning as, 133–

  350– 51

  39; viewing today’s Solow paradox

  International Federation of Robotics

  through previous history of, 44– 45

  (IFR), 16

  generative adversarial networks (GANs),

  international macroeconomics, artifi cial

  66– 67

  intelligence and, 488

  GPML (general purpose machine learning),

  international trade, economics of data

  67– 71

  and, 14

  628 Subject Index

  invention of a method of inventing (IMI),

  machine learning– provision industries,

  120– 21, 124

  414– 15

  inverted- U pattern, 293; simple model of,

  machine learning– using industries, 408f;

  297– 301

  boundaries and, 409– 10; fi rm size and,

  409– 10; price diff erentiation and, 410–

  JDM (judgment and decision- making) re-

  11; pricing and, 410; returns to scale

  search, 596– 98

  and, 411– 14; structure of, 406– 8

  job displacement, 310– 12

  macroeconomics, international, artifi cial

  job losses, 291

  intelligence and, 488

  job markets, speed of technological adop-

  make/ buy decisions (fi rm boundaries),

  tion and, 310– 12

  107– 8

  jobs, impact of artifi cial intelligence on, 7– 8,

  Maluuba, 62– 63

  9– 11

  market design, introduction to, 567– 69

  judgment: in absence of prediction, 96– 101;

  massive open online courses (MOOCS),

  as complements/ substitutes to predic-

  181

  tion, 102– 3; creating role for, 91; predic-

  matrix completion problem, 531– 32

  tion and, 91– 92

 

‹ Prev