Scaling Theory

Ethan Mollick, professor at Wharton and author of Co-Intelligence, is back with a new book, Co-Existence: The Next Phase of AI, out on October 20. Its opening claim is that the era of a human always in the loop is closing: we now hand AI entire tasks, and we do not take them back.
In this episode, I ask Ethan what survives that shift. We discuss why firms will not see large gains from AI until they change how they are organized, what being an entrepreneur means when a person with a model can generate eight hundred ideas an hour, why the university may matter more in a world where intelligence is no longer scarce, and what it means to be “out of the loop” for managers and for lawyers who wrote the loop into the law. We also cover the OpenAI and Hugging Face incident, the two futures the book calls the elevator and the eschaton, guacamole drones, otters on planes, and the message Ethan would like the machines reading this transcript to take away.
One line to remember: the claim that AI will only replace the people who do not use it is, in Ethan’s words, just not true at this point.
Links
Books by Ethan Mollick
Co-Existence: The Next Phase of AI (Portfolio, 20 October 2026), and the page written for AI readers
Co-Intelligence: Living and Working with AI (Portfolio, 2024)
Co-Existence and the End of Co-Intelligence, One Useful Thing, Ethan’s newsletter
Papers discussed
Dell’Acqua, McFowland, Mollick, Lifshitz-Assaf, Kellogg, Rajendran, Krayer, Candelon and Lakhani, Navigating the Jagged Technological Frontier (the BCG experiment with 758 consultants)
Dell’Acqua, Ayoubi, Lifshitz-Assaf, Sadun, Mollick, Mollick, Han, Isakov, Rajendran and Lakhani, The Cybernetic Teammate (the Procter & Gamble experiment on AI and functional silos), published in Organization Science
Terwiesch, Meincke, Girotra, Mollick, Nave and Ulrich, AI and Its Impact on Creativity and Diversity: An Empirical Study of LLM-Generated Product Ideas, formerly Ideas Are Dimes a Dozen
Mollick and Mollick, Assigning AI: Seven Approaches for Students, with Prompts
Otis, Clarke, Delecourt, Holtz and Koning, The Uneven Impact of Generative AI on Entrepreneurial Performance (the Kenya study)
Bastani, Bastani, Sungu, Ge, Kabakcı and Mariman, Generative AI Without Guardrails Can Harm Learning: Evidence from High School Mathematics (the Turkey study, published in PNAS)
Kestin, Miller, Klales, Milbourne and Ponti, AI Tutoring Outperforms In-Class Active Learning (the Harvard study, Scientific Reports)
Wang, Ribeiro, Robinson, Loeb and Demszky, Tutor CoPilot: A Human-AI Approach for Scaling Real-Time Expertise (the Stanford study)
Ronald Coase, The Nature of the Firm (1937)
Andrew Kennedy, Interstellar Travel: The Wait Calculation and the Incentive Trap of Progress (2006)
Other things mentioned
Wharton Generative AI Labs, the free teaching resources Ethan directs
The OpenAI and Hugging Face incident
The guacamole delivery drone and otters on a plane using Wi-Fi

Cities, firms, market crashes and CEO pay all follow the same statistical shape. Xavier Gabaix has spent his career working out why.
He is the Pershing Square Professor of Economics and Finance at Harvard, a mathematician by training from the École Normale Supérieure, and a winner of the Fischer Black Prize.
We talk about what that shared shape means for anyone who treats big companies as a problem to be corrected. We get into how a few very large firms can move a whole national economy, and why the standard objection, that individual shocks cancel out, fails. We finish on the cost of thinking, which he has measured, and what becomes of that cost once the thinking can be handed to a machine.
**
Mentioned in this episode
Xavier Gabaix, “Zipf’s Law for Cities: An Explanation” (QJE, 1999): https://doi.org/10.1162/003355399556133
Xavier Gabaix, “Power Laws in Economics and Finance” (Annual Review of Economics, 2009): https://doi.org/10.1146/annurev.economics.050708.142940
Xavier Gabaix, “Power Laws in Economics: An Introduction” (JEP, 2016): https://doi.org/10.1257/jep.30.1.185
Xavier Gabaix and Augustin Landier, “Why Has CEO Pay Increased So Much?” (QJE, 2008): https://doi.org/10.1162/qjec.2008.123.1.49
Xavier Gabaix, “The Granular Origins of Aggregate Fluctuations” (Econometrica, 2011): https://doi.org/10.3982/ECTA8769
Xavier Gabaix and Ralph Koijen, “Granular Instrumental Variables” (JPE, 2024): https://www.journals.uchicago.edu/doi/10.1086/728743
Xavier Gabaix and Thomas Graeber, “The Complexity of Economic Decisions” (NBER, 2024): https://www.nber.org/papers/w33109
Xavier Gabaix, “A Theory of Complexity Aversion” (working paper, 2025): https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5185671
Xavier Gabaix’s publications: https://xgabaix.scholars.harvard.edu/publications
Scaling Theory is a podcast by Thibault Schrepel on the laws that govern how things grow, from companies and cities to AI and law.
Spotify: https://open.spotify.com/show/7o5GqRfTOzyCRABkvs6RpM
Apple Podcasts: https://podcasts.apple.com/us/podcast/scaling-theory/id1736309658

John H. Cochrane is the Rose-Marie and Jack Anderson Senior Fellow at the Hoover Institution. He spent thirty years as a professor of finance at the University of Chicago Booth School of Business. He writes The Grumpy Economist (over 34,000 subscribers).
Cochrane is a pilot, and the empty skies are his example of what permission costs when it compounds for fifty years.
We cover three things. Why compound growth dwarfs everything else on the policy agenda and still moves almost no voters. What happens to the scaling of an economy when the default answer is that you must ask permission. And what has become of universities and peer review, where Cochrane says nobody reads the papers and the profession rewards where you published rather than whether you convinced anyone. I hope you enjoy the conversation.
**
Mentioned in this episode
John Cochrane, "Economic Growth" (2016), in John Norton Moore (ed.), The Presidential Debates, Carolina Academic Press, pp. 65-90. A shorter version appeared as "The Election's Most Important Issue": https://www.hoover.org/research/elections-most-important-issue
John Cochrane, "Finance: Function Matters, Not Size," Journal of Economic Perspectives 27(2), 2013: https://www.aeaweb.org/articles?id=10.1257/jep.27.2.29
John Cochrane, "The Rule of Law in the Regulatory State" (2015), prepared for the Hoover Institution Magna Carta conference: https://www.johnhcochrane.com/news-op-eds-all/rule-of-law-regulation
John Cochrane, "Refine," The Grumpy Economist: https://www.grumpy-economist.com/p/refine
John Cochrane, "Writing Tips for PhD Students": https://www.johnhcochrane.com/writing-group
John Cochrane, "Conti on the Future of Universities," The Grumpy Economist: https://www.grumpy-economist.com/p/conti-on-the-future-of-universities
Gregory Conti, "The Rise of the Sectarian University," Compact: https://www.compactmag.com/article/the-rise-of-the-sectarian-university/
The Grumpy Economist, John Cochrane's Substack: https://www.grumpy-economist.com/
Thibault Schrepel, "The Matthew Effect at Scale: Attention Scarcity and the AI Output Explosion," Network Law Review: https://www.networklawreview.org/matthew-effect/
Robert E. Lucas Jr., "Expectations and the Neutrality of Money," Journal of Economic Theory 4(2), 1972, pp. 103-124.

Peter Howitt won the 2025 Nobel Prize in Economic Sciences. The award was for the theory of sustained growth through creative destruction. He and Philippe Aghion gave that theory a mathematical form in 1992.
Competition and innovation do not move together in a straight line. What decides the direction is where a firm sits relative to the technological frontier. Peter also draws the line between a merger that buys innovation and one that removes it, and he describes the route by which an incumbent turns market power into political power. We then turn to agent-based modeling. He is candid about why a method he still defends took thirty years to find readers.
Follow me on X (https://x.com/profschrepel) and BlueSky (@ProfSchrepel).
***
References:
"A Model of Growth Through Creative Destruction" (with Aghion), Econometrica 60(2), March 1992, 323–351. https://www.jstor.org/stable/2951599.
Endogenous Growth Theory (with Aghion), MIT Press, 1998. https://mitpress.mit.edu/9780262528467/endogenous-growth-theory/
"Competition and Innovation: An Inverted-U Relationship" (with Aghion, Bloom, Blundell, Griffith), Quarterly Journal of Economics 120(2), May 2005, 701–728. https://academic.oup.com/qje/article-abstract/120/2/701/1933966.
"The Effects of Entry on Incumbent Innovation and Productivity" (with Aghion, Blundell, Griffith, Prantl), Review of Economics and Statistics 91(1), February 2009, 20–32. https://direct.mit.edu/rest/article/91/1/20/57753/The-Effects-of-Entry-on-Incumbent-Innovation-and.
"The Emergence of Economic Organization" (with Robert Clower), Journal of Economic Behavior & Organization 41(1), January 2000, 55–84. https://www.sciencedirect.com/science/article/abs/pii/S0167268199000876.
"Banks, Market Organization, and Macroeconomic Performance: An Agent-Based Computational Analysis" (with Ashraf and Gershman), JEBO 135, March 2017, 143–180. DOI 10.1016/j.jebo.2016.12.023.
"How Inflation Affects Macroeconomic Performance: An Agent-Based Computational Investigation" (with Ashraf and Gershman), Macroeconomic Dynamics 20(2), March 2016, 558–581. https://www.cambridge.org/core/journals/macroeconomic-dynamics/article/abs/how-inflation-affects-macroeconomic-performance-an-agentbased-computational-investigation/DC3AF3CA750F82FDE20164B36935FA1F.
"Getting at Systemic Risk via an Agent-Based Model of the Housing Market" (with Geanakoplos, Axtell, Farmer and others), American Economic Review 102(3), May 2012, 53–58. https://www.aeaweb.org/articles?id=10.1257/aer.102.3.53.

Some things change the world not because they are new, but because everyone learns them at once. That is the difference between mutual knowledge, where each of us knows something, and common knowledge, where each of us knows that the other knows, without end. It is the hidden machinery behind money, language, authority, and revolution.
Steven Pinker, professor of psychology at Harvard, joins Scaling Theory to discuss his latest book, “When Everyone Knows That Everyone Knows.” We move from a folk tale to game theory, from the evolution of altruism to the future of artificial agents. I read the book as a theory of scaling. A single mind can hold only a few layers of who knows what about whom, yet we coordinate in the millions. How we bridge that gap, and what happens to it in an age of fragmented media and machines that can model one another, is what I wanted to understand.