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