The Economics of Artificial Intelligence

Home > Other > The Economics of Artificial Intelligence > Page 103
The Economics of Artificial Intelligence Page 103

by Ajay Agrawal


  extremely noisy, but it does parallel processing. It is extraordinarily effi

  cient,

  but there is no magic there. So, it is diffi

  cult to imagine that, with suffi

  cient

  data in the future, there will remain things that only humans can do.

  The reason that we see so many limitations, I think, is that this fi eld is

  really at the very beginning. I mean we are talking about developments

  (i.e., deep learning) that took off eight years ago. That is nothing. You have

  to imagine what it might be like in fi fty years. Because the one thing that I

  fi nd extraordinarily surprising in what is happening in AI these days is that

  everything is happening faster than we expected. People were saying that it

  will take ten years for AI to beat Go. The interesting thing is it took less by

  an order of magnitude. This excess of speed at which this thing is developing

  and accelerating, I think, is very remarkable. So, setting limits is certainly

  premature.

  One point that was made yesterday was about the uniqueness of humans

  when it comes to evaluations. It was called judgment, but in my jargon it

  is “evaluation.” Evaluations of outcomes are, basically, the utility side of

  the decision function. I do not see why that should be reserved for humans.

  On the contrary, I would like to make the following argument: the main

  characteristic of people is that they are very noisy. You show them the same

  stimulus twice and they do not give you the same response twice. We have

  stochastic choice theory because there is so much variability in people’s

  choices conditional on the same stimuli. What can be done with AI is to

  create a program that observes an individual’s choices. That program will

  be better than people at a wide variety of things. In particular, it will make

  better choices for the individual. Why? Because it will be noise free. We know

  from the literature that Colin cited on predictions that there is an interesting

  tidbit. Take some clinicians and have them predict some criterion a large

  number of times. Then develop a simple equation that predicts, not the out-

  610 Daniel Kahneman

  come, but each clinician’s judgment. That model does better in predicting

  the outcome than the clinicians themselves.

  That is fundamental. It is telling you that one of the major limitations on

  human performance is not bias, it is just noise. I may be partly responsible

  for this as, when people now talk about error, they tend to think of bias as

  an explanation. That’s the fi rst thing that comes to mind when there is an

  error in human performance.

  In fact, most of the errors that people make are better viewed as random

  noise, and there is an awful lot of it. Admitting the existence of noise has

  implications for practice. One implication is obvious. You should replace

  humans by algorithms whenever possible. Even when the algorithm does not

  do very well, humans do so poorly and are so noisy that, just by removing

  the noise, you can do better than people. The other is that when you can-

  not replace the human by an algorithm, you try to have human simulate an

  algorithm. The idea is that, by enforcing regularity, process and discipline

  on judgment and on choice, you reduce the noise, and you improve perfor-

  mance because noise is so pernicious.

  Yann LeCun said yesterday that humans would always prefer emo-

  tional contact with other humans. That strikes me as probably wrong. It is

  extremely easy to develop stimuli to which people will respond emotionally.

  An expressive face that changes expressions, especially if it’s baby- shaped,

  gives cues that will make people feel very emotional. Robots will have these

  cues. Furthermore, it is already the case that AI reads faces better than

  people do. Undoubtedly, robots will be able to predict emotions and de-

  velopment in emotions far better than people can.

  I really can imagine that one of the major uses of robots will be taking care

  of the old. I can imagine that many old people will prefer to be taken care

  of by friendly robots that have a name, have a personality, and are always

  pleasant. They will prefer that to being taken care of by their children.

  I want to end on a story. A well- known novelist wrote me some time ago

  that he’s planning a novel. The novel is about a love triangle between two

  humans and a robot. What he wanted to know is how the robot would be

  diff erent from the people.

  I proposed three main diff erences. One is obvious: the robot will be much

  better at statistical reasoning and less enamored with stories and narra-

  tives than people are. The other is that the robot would have a much higher

  emotional intelligence. The third is that the robot would be wiser. Wisdom

  is breadth. Wisdom is not having too narrow a view. That is the essence of

  wisdom; it’s broad framing. A robot will be endowed with broad framing. I

  say that when it has learned enough, it will be wiser than we people because

  we do not have broad faming. We are narrow thinkers, we are noisy think-

  ers, and it is very easy to improve upon us. I do not think that there is very

  much that we can do that computer will not eventually be programmed to do.

  Contributors

  Daron Acemoglu

  Erik Brynjolfsson

  Department of Economics

  MIT Sloan School of Management

  Massachusetts Institute of

  100 Main Street, E62-414

  Technology

  Cambridge, MA 02142

  50 Memorial Drive

  Cambridge, MA 02142-1347

  Colin F. Camerer

  Department of Economics

  Philippe Aghion

  California Institute of Technology

  Collège de France

  1200 East California Boulevard

  3 Rue D’Ulm

  Pasadena, CA 91125

  75005 Paris, France

  Judith Chevalier

  Ajay Agrawal

  Yale School of Management

  Rotman School of Management

  135 Prospect Street

  University of Toronto

  New Haven, CT 06520

  105 St. George Street

  Toronto, ON M5S 3E6, Canada

  Iain M. Cockburn

  School of Management

  Susan Athey

  Boston University

  Graduate School of Business

  595 Commonwealth Avenue

  Stanford University

  Boston, MA 02215

  655 Knight Way

  Stanford, CA 94305

  Tyler Cowen

  Department of Economics

  James Bessen

  George Mason University

  Technology & Policy Research

  4400 University Drive

  Initiative

  Fairfax, VA 22030

  Boston University School of Law

  765 Commonwealth Avenue

  Patrick Francois

  Boston, MA 02215

  Vancouver School of Economics

  University of British Columbia

  IONA Building, 6000 Iona Drive

  Vancouver, BC V6T 2E8, Canada

  611

  612 Contributors

  Jason Furman

  Daniel Kahneman

  Harvard Kennedy School

  Woodrow Wilson School

&
nbsp; 79 John F. Kennedy Street

  Princeton University

  Cambridge, MA 02138

  Princeton, NJ 08544-1013

  Alberto Galasso

  Anton Korinek

  Rotman School of Management

  University of Virginia

  University of Toronto

  Monroe Hall 246

  105 St. George Street

  248 McCormick Road

  Toronto, ON M5S 3E6, Canada

  Charlottesville, VA 22904

  Joshua Gans

  Mara Lederman

  Rotman School of Management

  Rotman School of Management

  University of Toronto

  University of Toronto

  105 St. George Street

  105 St. George Street

  Toronto, ON M5S 3E6, Canada

  Toronto, Ontario M5S 3E6, Canada

  Avi Goldfarb

  Hong Luo

  Rotman School of Management

  Harvard Business School

  University of Toronto

  Morgan Hall 241

  105 St. George Street

  Soldiers Field Road

  Toronto, ON M5S 3E6, Canada

  Boston, MA 02163

  Austan Goolsbee

  John McHale

  University of Chicago Booth School

  108 Cairnes Building

  of Business

  School of Business and Economics

  5807 S. Woodlawn Avenue

  National University of Ireland

  Chicago, IL 60637

  Galway H91 TK33, Ireland

  Rebecca Henderson

  Paul R. Milgrom

  Harvard Business School

  Department of Economics

  Morgan Hall 445

  Stanford University

  Soldiers Field Road

  579 Serra Mall

  Boston, MA 02163

  Stanford, CA 94305-6072

  Ginger Zhe Jin

  Matthew Mitchell

  Department of Economics

  Rotman School of Management

  University of Maryland

  University of Toronto

  3115F Tydings Hall

  105 St. George Street

  College Park, MD 20742-7211

  Toronto, ON M5S 3E6, Canada

  Benjamin F. Jones

  Alexander Oettl

  Department of Management and

  Scheller College of Business

  Strategy

  Georgia Institute of Technology

  Kellogg School of Management

  800 West Peachtree Street NW

  Northwestern University

  Atlanta, GA 30308

  2211 Campus Drive

  Evanston, IL 60208

  Andrea Prat

  Columbia Business School

  Charles I. Jones

  Uris Hall 624

  Graduate School of Business

  3022 Broadway

  Stanford University

  New York, NY 10027-6902

  655 Knight Way

  Stanford, CA 94305-4800

  Contributors 613

  Manav Raj

  Chad Syverson

  Stern School of Business

  University of Chicago Booth School of

  New York University

  Business

  44 West Fourth Street

  5807 S. Woodlawn Avenue

  New York, NY 10012

  Chicago, IL 60637

  Pascual Restrepo

  Matt Taddy

  Department of Economics

  University of Chicago Booth School of

  Boston University

  Business

  270 Bay State Road

  5807 S. Woodlawn Avenue

  Boston, MA 02215

  Chicago, IL 60637

  Daniel Rock

  Steven Tadelis

  MIT Sloan School of Management

  Haas School of Business

  100 Main Street, E62-365

  University of California, Berkeley

  Cambridge, MA 02142

  545 Student Services Building

  Berkeley, CA 94720

  Jeff rey D. Sachs

  Center for Sustainable Development,

  Manuel Trajtenberg

  Earth Institute

  Eitan Berglas School of Economics

  Columbia University

  Tel Aviv University

  535 West 116th Street, MC 4327

  Tel Aviv 69978, Israel

  New York, NY 10027

  Daniel Trefl er

  Robert Seamans

  Rotman School of Management

  Stern School of Business

  University of Toronto

  New York University

  105 St. George Street

  44 West 4th Street, KMC 7-58

  Toronto, ON M5S 3E6, Canada

  New York, NY 10012

  Catherine Tucker

  Scott Stern

  MIT Sloan School of Management

  MIT Sloan School of Management

  100 Main Street, E62-536

  100 Main Street, E62-476

  Cambridge, MA 02142

  Cambridge, MA 02142

  Hal Varian

  Betsey Stevenson

  School of Information

  Gerald R. Ford School of Public Policy

  University of California, Berkeley

  University of Michigan

  102 South Hall

  5224 Weill Hall

  Berkeley, CA 94720-4600

  735 South State Street

  Ann Arbor, MI 48109-3091

  Joseph E. Stiglitz

  Columbia University

  Uris Hall 212

  3022 Broadway

  New York, NY 10027

  Author Index

  Abadie, A., 531

  Angermueller, C., 168

  Ablon, L., 448

  Aral, S., 43

  Abrahamson, Z., 420

  Arntz, M., 321

  Abramovitz, M., 32

  Aronoff , M., 497

  Acemoglu, D., 23n1, 43, 89, 90, 105, 127,

  Arrow, K., 9, 43, 110, 118, 148, 364n11, 366,

  141, 152, 197, 198, 201, 202, 203,

  412

  203n4, 204, 204n5, 205, 206, 208, 210,

  Arthur, B. W., 150

  211, 212n7, 219, 220, 220n11, 223, 224,

  Asher, S., 525

  225n14, 238, 240, 243, 271, 283, 293n3,

  Athey, S., 68, 425, 448, 449, 510, 514, 515,

  376n27, 376n28, 554

  516, 517, 519, 523, 524, 525, 527, 528,

  Acquisti, A., 410, 416, 424, 440n2, 444, 448,

  529, 530, 531, 532, 533, 534, 536, 538

  451, 457, 457n42, 459, 483n19

  Atkeson, A., 42n19

  Adee, S., 426

  Autor, D. H., 7, 23n1, 30, 89, 198, 202n3,

  Agarwal, A., 82n12, 83

  203, 208, 220n11, 238, 239n3, 240, 271,

  Aghion, P., 122, 172, 262, 262n19, 263, 265,

  309, 322, 475, 555

  267, 268, 373n23, 465, 477, 479, 495

  Axelrod, R., 413

  Agrawal, A., 5, 39, 90, 97n8, 150, 161, 167,

  Ayres, R., 201

  241, 425, 464, 501

  Azoulay, P., 475, 477

  Aguiar, M., 386n36

  Airoldi, E., 80

  Babcock, L., 592n5

  Akerlof, G., 378

  Bai, J., 532

  Akerman, A., 7, 303

  Baker, D., 366n14

  Alexopoulos, M., 555

  Baker, G., 107

  Allen, R. C., 204, 209

  Banbura, M., 537

  Alloway, T., 29

  Barkai, S., 271

  Alon, T., 43

  Barrat, J., 350

  Alpaydin, E., 92

  Baslandze, S., 263, 264

  Altman, S., 325

  Bastani, H., 529

  Alvarez- Cuadrado, F., 241n4

  Baumol, W., 38n14, 238

  Anderson, R.,
406, 416

  Bayati, M., 529, 531, 532, 536

  Andrade, E. B., 592n3

  Belloni, A., 93, 522

  Andrews, D., 30

  Bengio, S., 401

  615

  616 Author Index

  Bengio, Y., 71, 74, 75, 75n6, 149, 168, 168n6,

  Cavusoglu, H., 450

  172

  Cette, G., 27

  Benzell, S. G., 90, 336

  Chandler, A. D., 206

  Berg, A., 373n24

  Chapelle, O., 529

  Beron, K. J., 483

  Chapman, J. P., 599

  Bertocchi, G., 430

  Chapman, L. J., 599

  Bessen, J. E., 23n1, 203, 300, 413, 555

  Chernozkhukov, V., 93, 522, 523, 527

  Bhatt, M. A., 592n4

  Chetty, R., 270

  Bickel, P. J., 522

  Chevalier, J., 577n5

  Binmore, K., 589, 589n1, 593

  Chiou, L., 428, 448

  Bjorkegren, D., 516, 519

  Chong, J.-K., 593, 595

  Blake, T., 582

  Christie, W. G., 414

  Blattberg, R. C., 598

  Chui, M., 331

  Blei, D. M., 507, 510, 515, 532, 533, 534

  Clark, C., 295

  Bloom, N., 27, 150, 223, 259n15, 268, 560

  Cockburn, I., 150

  Bolton, P., 92, 95, 95n5, 96n7, 100, 112

  Coey, D., 514

  Boppart, T., 241n4, 295

  Cohen, J., 555

  Borenstein, S., 413

  Cohen, L., 501

  Bornmann, L., 153

  Comin, D., 241n4, 295

  Bostrom, N., 3, 286, 381, 382, 382n31

  Corrigan, B., 597

  Bottou, L., 79

  Costa- Gomes, M. A., 593, 593n9, 594n9

  Bousquet, O., 79

  Courville, A., 71, 75n6

  Bowen, W., 38n14

  Cowen, T., 27, 150

  Brandeis, L. D., 431, 459

  Cowgill, B., 562

  Brander, J. A., 473

  Cranor, L. F., 424, 458

  Brandimarte, L., 448

  Crawford, V. P., 593, 593n9, 594n9

  Brandt, L., 471

  Criscuolo, C., 30

  Bresnahan, T. F., 4, 39, 42, 116, 119, 120,

  169, 176n2, 310

  Dana, J., 599

  Bridgman, B., 38

  Danaylov, N., 253

  Brooks, R., 124

  Dasgupta, P., 366n15

  Broseta, B., 593n9, 594n9

  Datta, A., 433

  Brown, J., 314

  Dauth, W., 556

  Brunskill, E., 528

  David, P. A., 4, 41, 42n19, 119

  Brynjolfsson, E., 23, 23n1, 28n7, 30, 39, 40,

  Davies, R. B., 483

  42, 43, 47, 50, 89, 119, 120, 150, 197,

  Dávila, E., 354

  201, 204, 309, 555, 557, 560, 563

  Dawes, R. M., 597, 598, 599, 599n15

  Brzeski, C., 556

  Dawsey, K., 29

  Buera, F. J., 293, 293n2

  Deaton, A., 26, 85

  Buffi

  e, E. F., 373n24

  Della Vigna, S., 411

  Bughin, J., 408

  Delli Gatti, D., 361, 362n9, 380

 

‹ Prev