AI and Market Power

Question A:

Use of artificial intelligence is likely to lead to a substantial increase in problems associated with market power in digital markets.

Responses weighted by each expert's confidence

Question B:

Artificial intelligence offers substantial opportunities for new entrants into digital markets that have previously been concentrated.

Responses weighted by each expert's confidence

Question C:

Artificial intelligence is likely to be a highly concentrated industry, dominated by a handful of players.

Responses weighted by each expert's confidence

Question A Participant Responses

Participant University Vote Confidence Bio/Vote History
Allen
Franklin Allen
Imperial College London Did Not Answer Bio/Vote History
Antras
Pol Antras
Harvard
Agree
7
Bio/Vote History
Blanchard
Olivier Blanchard
Peterson Institute
Uncertain
3
Bio/Vote History
I feel incompetent, but the proposition seems far from obvious. Why should the largest players profit proportionally more from AI?
Bloom
Nicholas Bloom
Stanford Did Not Answer Bio/Vote History
Blundell
Richard William Blundell
University College London
Uncertain
5
Bio/Vote History
Botticini
Maristella Botticini
Bocconi
Uncertain
7
Bio/Vote History
Bénassy-Quéré
Agnès Bénassy-Quéré
Paris School of Economics
Agree
2
Bio/Vote History
Carletti
Elena Carletti
Bocconi Did Not Answer Bio/Vote History
Danthine
Jean-Pierre Danthine
Paris School of Economics Did Not Answer Bio/Vote History
De Grauwe
Paul De Grauwe
LSE
No Opinion
Bio/Vote History
Eeckhout
Jan Eeckhout
UPF Barcelona
Strongly Agree
9
Bio/Vote History
These technologies exhibit returns to scale and network externalities at a global level. Moreover, data is non-rival and excludable. These are all ingredients for failure of a competitive equilibrium allocation.
Fehr
Ernst Fehr
Universität Zurich Did Not Answer Bio/Vote History
Freixas
Xavier Freixas
Barcelona GSE Did Not Answer Bio/Vote History
Fuchs-Schündeln
Nicola Fuchs-Schündeln
Goethe-Universität Frankfurt
Agree
5
Bio/Vote History
Galí
Jordi Galí
Barcelona GSE
No Opinion
Bio/Vote History
Garicano
Luis Garicano
LSE
Uncertain
5
Bio/Vote History
The IP itself is hard to protect- none of the innovations are proprietary; as to scale advantages, it appears that fine tuning a model with cheap hardware dominates training from scratch a gigantic model. So the jury is out, but this is potentially a plus for competition.
Gorodnichenko
Yuriy Gorodnichenko
Berkeley Did Not Answer Bio/Vote History
Griffith
Rachel Griffith
University of Manchester Did Not Answer Bio/Vote History
Guerrieri
Veronica Guerrieri
Chicago Booth
Uncertain
4
Bio/Vote History
Guiso
Luigi Guiso
Einaudi Institute for Economics and Finance
Uncertain
5
Bio/Vote History
Guriev
Sergei Guriev
Sciences Po
Strongly Agree
8
Bio/Vote History
Honohan
Patrick Honohan
Trinity College Dublin
No Opinion
Bio/Vote History
Javorcik
Beata Javorcik
University of Oxford
Strongly Agree
8
Bio/Vote History
Krahnen
Jan Pieter Krahnen
Goethe University Frankfurt Did Not Answer Bio/Vote History
Kőszegi
Botond Kőszegi
Central European University
Agree
4
Bio/Vote History
La Ferrara
Eliana La Ferrara
Harvard Kennedy Did Not Answer Bio/Vote History
Leuz
Christian Leuz
Chicago Booth
Uncertain
2
Bio/Vote History
Mayer
Thierry Mayer
Sciences-Po Did Not Answer Bio/Vote History
Meghir
Costas Meghir
Yale Did Not Answer Bio/Vote History
Pagano
Marco Pagano
Università di Napoli Federico II
Agree
4
Bio/Vote History
Pastor
Lubos Pastor
Chicago Booth
Uncertain
1
Bio/Vote History
Persson
Torsten Persson
Stockholm University Did Not Answer Bio/Vote History
Pissarides
Christopher Pissarides
London School of Economics and Political Science Did Not Answer Bio/Vote History
Portes
Richard Portes
London Business School
Uncertain
5
Bio/Vote History
Prendergast
Canice Prendergast
Chicago Booth
Agree
8
Bio/Vote History
Propper
Carol Propper
Imperial College London
Strongly Agree
5
Bio/Vote History
Rasul
Imran Rasul
University College London
Agree
8
Bio/Vote History
Reichlin
Lucrezia Reichlin
London Business School Did Not Answer Bio/Vote History
Reis
Ricardo Reis
London School of Economics
Agree
5
Bio/Vote History
Repullo
Rafael Repullo
CEMFI
Agree
4
Bio/Vote History
Rey
Hélène Rey
London Business School Did Not Answer Bio/Vote History
Schoar
Antoinette Schoar
MIT
Agree
7
Bio/Vote History
Storesletten
Kjetil Storesletten
University of Minnesota
Agree
3
Bio/Vote History
Sturm
Daniel Sturm
London School of Economics
Agree
5
Bio/Vote History
Van Reenen
John Van Reenen
LSE
Agree
6
Bio/Vote History
Van der Ploeg
Rick Van der Ploeg
Oxford
Strongly Agree
6
Bio/Vote History
Most of the AI technologies are based on machine learning, which get better the more data they can learn from. This leads naturally to monopolies.
Vickers
John Vickers
Oxford
Uncertain
4
Bio/Vote History
Voth
Hans-Joachim Voth
University of Zurich
Disagree
7
Bio/Vote History
Whelan
Karl Whelan
University College Dublin Did Not Answer Bio/Vote History
Wyplosz
Charles Wyplosz
The Graduate Institute Geneva
Disagree
2
Bio/Vote History
Isn't it about disruption?

Question B Participant Responses

Participant University Vote Confidence Bio/Vote History
Allen
Franklin Allen
Imperial College London Did Not Answer Bio/Vote History
Antras
Pol Antras
Harvard
Uncertain
6
Bio/Vote History
Perhaps, but that doesn't mean these entrants will remain stand-alone competitors, rather than being bought out by tech giants.
Blanchard
Olivier Blanchard
Peterson Institute
Uncertain
3
Bio/Vote History
same reaction. It may well happen, but it does not seem pre ordained.
Bloom
Nicholas Bloom
Stanford Did Not Answer Bio/Vote History
Blundell
Richard William Blundell
University College London
Uncertain
5
Bio/Vote History
Botticini
Maristella Botticini
Bocconi
Uncertain
7
Bio/Vote History
Bénassy-Quéré
Agnès Bénassy-Quéré
Paris School of Economics
Agree
2
Bio/Vote History
Carletti
Elena Carletti
Bocconi Did Not Answer Bio/Vote History
Danthine
Jean-Pierre Danthine
Paris School of Economics Did Not Answer Bio/Vote History
De Grauwe
Paul De Grauwe
LSE
Agree
6
Bio/Vote History
Eeckhout
Jan Eeckhout
UPF Barcelona
Strongly Disagree
9
Bio/Vote History
If there are any new entrants, they are likely to replace existing incumbents and achieve more dominant positions than before. New entrants are unlikely to foster competition.
Fehr
Ernst Fehr
Universität Zurich Did Not Answer Bio/Vote History
Freixas
Xavier Freixas
Barcelona GSE Did Not Answer Bio/Vote History
Fuchs-Schündeln
Nicola Fuchs-Schündeln
Goethe-Universität Frankfurt
Uncertain
3
Bio/Vote History
Galí
Jordi Galí
Barcelona GSE
No Opinion
Bio/Vote History
Garicano
Luis Garicano
LSE
Agree
5
Bio/Vote History
Same explanation. Barriers to entry appear right now low - available open source models (Meta), no IP barriers, cheap finetuning
Gorodnichenko
Yuriy Gorodnichenko
Berkeley Did Not Answer Bio/Vote History
Griffith
Rachel Griffith
University of Manchester Did Not Answer Bio/Vote History
Guerrieri
Veronica Guerrieri
Chicago Booth
Uncertain
4
Bio/Vote History
Guiso
Luigi Guiso
Einaudi Institute for Economics and Finance
Agree
6
Bio/Vote History
Guriev
Sergei Guriev
Sciences Po
Agree
8
Bio/Vote History
Honohan
Patrick Honohan
Trinity College Dublin
Uncertain
1
Bio/Vote History
Javorcik
Beata Javorcik
University of Oxford
Uncertain
3
Bio/Vote History
Krahnen
Jan Pieter Krahnen
Goethe University Frankfurt Did Not Answer Bio/Vote History
Kőszegi
Botond Kőszegi
Central European University
Uncertain
10
Bio/Vote History
La Ferrara
Eliana La Ferrara
Harvard Kennedy Did Not Answer Bio/Vote History
Leuz
Christian Leuz
Chicago Booth
Uncertain
2
Bio/Vote History
I agree directionally, meaning there will be new opportunities and AI could help smaller players to compete. But unclear if effect will be substantial. And bigger players will also use AI.
Mayer
Thierry Mayer
Sciences-Po Did Not Answer Bio/Vote History
Meghir
Costas Meghir
Yale Did Not Answer Bio/Vote History
Pagano
Marco Pagano
Università di Napoli Federico II
Disagree
4
Bio/Vote History
Pastor
Lubos Pastor
Chicago Booth
Agree
1
Bio/Vote History
Persson
Torsten Persson
Stockholm University Did Not Answer Bio/Vote History
Pissarides
Christopher Pissarides
London School of Economics and Political Science Did Not Answer Bio/Vote History
Portes
Richard Portes
London Business School
Uncertain
5
Bio/Vote History
Prendergast
Canice Prendergast
Chicago Booth
Uncertain
8
Bio/Vote History
Propper
Carol Propper
Imperial College London
No Opinion
Bio/Vote History
Rasul
Imran Rasul
University College London
Uncertain
5
Bio/Vote History
Reichlin
Lucrezia Reichlin
London Business School Did Not Answer Bio/Vote History
Reis
Ricardo Reis
London School of Economics
Disagree
4
Bio/Vote History
Repullo
Rafael Repullo
CEMFI
Uncertain
4
Bio/Vote History
Rey
Hélène Rey
London Business School Did Not Answer Bio/Vote History
Schoar
Antoinette Schoar
MIT
Uncertain
7
Bio/Vote History
Storesletten
Kjetil Storesletten
University of Minnesota
Disagree
4
Bio/Vote History
Access to big data is what gives AI power. Actors with large market share possess precisely such data. AI will therefore increase market power for the already dominant producers
Sturm
Daniel Sturm
London School of Economics
Disagree
5
Bio/Vote History
Van Reenen
John Van Reenen
LSE
Uncertain
5
Bio/Vote History
Van der Ploeg
Rick Van der Ploeg
Oxford
Strongly Agree
6
Bio/Vote History
Lots of new small entrants can be successful with disruptive technologies.
Vickers
John Vickers
Oxford
Uncertain
4
Bio/Vote History
Voth
Hans-Joachim Voth
University of Zurich
Agree
8
Bio/Vote History
in search in particular, disruption seems likely
Whelan
Karl Whelan
University College Dublin Did Not Answer Bio/Vote History
Wyplosz
Charles Wyplosz
The Graduate Institute Geneva
Agree
3
Bio/Vote History

Question C Participant Responses

Participant University Vote Confidence Bio/Vote History
Allen
Franklin Allen
Imperial College London Did Not Answer Bio/Vote History
Antras
Pol Antras
Harvard
Agree
5
Bio/Vote History
Blanchard
Olivier Blanchard
Peterson Institute
Disagree
4
Bio/Vote History
AI is not an industry. it is a set of techniques, which can benefit potentially all firms.
Bloom
Nicholas Bloom
Stanford Did Not Answer Bio/Vote History
Blundell
Richard William Blundell
University College London
Uncertain
5
Bio/Vote History
Botticini
Maristella Botticini
Bocconi
Agree
7
Bio/Vote History
Bénassy-Quéré
Agnès Bénassy-Quéré
Paris School of Economics
Agree
2
Bio/Vote History
Carletti
Elena Carletti
Bocconi Did Not Answer Bio/Vote History
Danthine
Jean-Pierre Danthine
Paris School of Economics Did Not Answer Bio/Vote History
De Grauwe
Paul De Grauwe
LSE
Agree
6
Bio/Vote History
Eeckhout
Jan Eeckhout
UPF Barcelona
Strongly Agree
9
Bio/Vote History
Fehr
Ernst Fehr
Universität Zurich Did Not Answer Bio/Vote History
Freixas
Xavier Freixas
Barcelona GSE Did Not Answer Bio/Vote History
Fuchs-Schündeln
Nicola Fuchs-Schündeln
Goethe-Universität Frankfurt
Uncertain
3
Bio/Vote History
Galí
Jordi Galí
Barcelona GSE
Agree
5
Bio/Vote History
Garicano
Luis Garicano
LSE
Disagree
5
Bio/Vote History
Response as explained in two previous answers
Gorodnichenko
Yuriy Gorodnichenko
Berkeley Did Not Answer Bio/Vote History
Griffith
Rachel Griffith
University of Manchester Did Not Answer Bio/Vote History
Guerrieri
Veronica Guerrieri
Chicago Booth
Uncertain
4
Bio/Vote History
Guiso
Luigi Guiso
Einaudi Institute for Economics and Finance
Agree
6
Bio/Vote History
Guriev
Sergei Guriev
Sciences Po
Agree
8
Bio/Vote History
Honohan
Patrick Honohan
Trinity College Dublin
Agree
1
Bio/Vote History
Javorcik
Beata Javorcik
University of Oxford
Agree
7
Bio/Vote History
Krahnen
Jan Pieter Krahnen
Goethe University Frankfurt Did Not Answer Bio/Vote History
Kőszegi
Botond Kőszegi
Central European University
Agree
8
Bio/Vote History
La Ferrara
Eliana La Ferrara
Harvard Kennedy Did Not Answer Bio/Vote History
Leuz
Christian Leuz
Chicago Booth
Uncertain
2
Bio/Vote History
This question is very similar to a)
Mayer
Thierry Mayer
Sciences-Po Did Not Answer Bio/Vote History
Meghir
Costas Meghir
Yale Did Not Answer Bio/Vote History
Pagano
Marco Pagano
Università di Napoli Federico II
Agree
4
Bio/Vote History
Pastor
Lubos Pastor
Chicago Booth
Uncertain
1
Bio/Vote History
Persson
Torsten Persson
Stockholm University Did Not Answer Bio/Vote History
Pissarides
Christopher Pissarides
London School of Economics and Political Science Did Not Answer Bio/Vote History
Portes
Richard Portes
London Business School
Agree
7
Bio/Vote History
Prendergast
Canice Prendergast
Chicago Booth
Agree
9
Bio/Vote History
Propper
Carol Propper
Imperial College London
Agree
5
Bio/Vote History
Rasul
Imran Rasul
University College London
Agree
7
Bio/Vote History
Reichlin
Lucrezia Reichlin
London Business School Did Not Answer Bio/Vote History
Reis
Ricardo Reis
London School of Economics
Agree
4
Bio/Vote History
Repullo
Rafael Repullo
CEMFI
Agree
4
Bio/Vote History
Rey
Hélène Rey
London Business School Did Not Answer Bio/Vote History
Schoar
Antoinette Schoar
MIT
Uncertain
7
Bio/Vote History
It depends on what is meant by AI. If we are talking about ChatGTP, that is mostly likely true. If this includes applications to many different industry functionalities there might be some specialized providers as well.
Storesletten
Kjetil Storesletten
University of Minnesota
Uncertain
2
Bio/Vote History
Sturm
Daniel Sturm
London School of Economics
Agree
5
Bio/Vote History
Van Reenen
John Van Reenen
LSE
Agree
6
Bio/Vote History
Van der Ploeg
Rick Van der Ploeg
Oxford
Strongly Agree
6
Bio/Vote History
Vickers
John Vickers
Oxford
Uncertain
4
Bio/Vote History
Voth
Hans-Joachim Voth
University of Zurich
Uncertain
7
Bio/Vote History
Whelan
Karl Whelan
University College Dublin Did Not Answer Bio/Vote History
Wyplosz
Charles Wyplosz
The Graduate Institute Geneva
Agree
2
Bio/Vote History