Optimising for Trouble – Game Theory and AI Safety | with Jobst Heitzig

In this episode, we explore how game theory and economics can inform the use of AI. Our guest, Annie Liang, introduces the concept of AI clones, systems designed to imitate specific humans, and discusses their potential applications, for example in recruiting. We examine the trade-off between the efficiency gains the clones can provide and the costs that arise when AI cannot perfectly create such clones without error. The conversation then turns to creative ownership, focusing not only on content but also on the ownership of creative style as AI becomes increasingly capable of reproducing it. Finally, we discuss how insights from information design can contribute to the broader debate on AI. Annie Liang is an associate professor of economics and (by courtesy) of computer science at Northwestern University.

Can we trust lab experiments to predict real-world behaviour? In this episode, we explore the question of external validity together with Jan Stoop. From classic dictator games in the lab to misdelivered envelopes with real cash in private homes, we discuss whether people act differently when they are being observed - and what happens when established lab findings meet reality. We also talk about rich versus poor behaviour, financial stress, and why small "hassle costs" can create large inequalities in outcomes. Jan Stoop is Associate Professor at Tilburg University and Erasmus University Rotterdam. His research focuses on behavioural and experimental economics, with a particular emphasis on pro-social and unethical behaviour, financial stress, and inequality. He is known for creative field experiments that rigorously test whether laboratory results hold in real-world settings.

What happens when an AI system faithfully follows a flawed goal? In this episode, we explore how even well-designed algorithms can produce dangerous outcomes — from amplifying hate speech to mismanaging infrastructure — simply by optimising a reward function which, like all reward functions, fails to encode all that matters. We discuss the hidden risks of reinforcement learning, why over-optimisation can backfire, and how game theory helps us rethink what it means for AI to act "rationally" in complex, real-world environments.   Jobst Heitzig is a mathematician at the Potsdam Institute for Climate Impact Research and an expert in AI safety and decision design.

In this episode, we discuss how game theory sheds light on modern electoral competition. Our guest, Adam Meirowitz, introduces the classical model of candidate convergence and explains why real elections often depart from this prediction. We then explore the role of echo chambers and selective exposure, and why these surprisingly do not necessarily lead to greater polarisation. We also touch on how bounded rationality among voters can further impact these dynamics.   Adam Meirowitz is the Damon Wells Professor of Political Science with a courtesy appointment in Economics at Yale University. His referenced research paper on echo chambers "Selective Exposure and Electoral Competition" is published in the Journal of Politics (2025) and is joint work with Avidit Acharya, Peter Buisseret, and Floyd Zhang. You can find it here.

Talk is cheap – literally. But if words cost nothing, why do we ever trust them? In this episode, Yale Professor Navin Kartik breaks down the game theory of communication. We kick things off with the basics: what is the academic definition of "cheap talk", and why does cheap talk sometimes work but other times break down? From there, we transition to Navin's own research, which explores deeper facets of communication. We learn about lying costs and "inflated language": why does a reference letter need to say "excellent" to just mean "competent"? Finally, we dive into topics like pandering (why experts lie to agree with you) and muddled information (why test scores tell you more about gaming skills than natural ability).

In this episode, we explore the foundations and evolution of decision theory. Our guest, Itzhak Gilboa, begins with a brief historical overview of how the field has developed over time. We naturally discuss maximising expected utility, Bayesian decision theory, and Savage's representation theorem. Itzhak then delves into critiques of the Bayesian approach, especially concerning its interpretation of what constitutes a "rational decision maker." He presents a range of alternative decision frameworks, including approaches that do not require individuals to specify a full subjective probability distribution.   Itzhak Gilboa is Professor of Economics and Decision Sciences at HEC Paris.

In this episode, we explore how game theory and network effects intersect to inform optimal incentive structures in organizations. Pau Milán walks us through his recent research, co-authored with Nicolás Oviedo-Davila and Marc Claveria, which uses a novel game-theoretic model to show how the productivity of one employee can amplify the output of others in their network—and how firms can leverage this to design smarter compensation schemes. Drawing on real-world examples and his theoretical work, Pau offers practical insights for employers and economists alike.   Pau Milán is an Associate Professor of Economics at Universitat Autònoma de Barcelona (UAB) and an Affiliated Professor at the Barcelona School of Economics (BSE). His research interests include Social and Economic Networks, Information Economics, Development Economics, Organisations, and Applied Game Theory. You can find the working paper "Incentive Contracts and Peer Effects in the Workplace" here.

In this episode we explore the most classical topic from Game Theory – equilibrium analysis. Our guest Colin Camerer shares insights on the game LUPI ('lowest unique positive integer'). We first discuss the actual mathematical equilibrium analysis and then dive into real life: results from a Swedish game show in which the game was played in a lottery format. Colin also compares the results from the game show with a study of the same game in a controlled lab environment. We finish the episode by deep-diving into level-k reasoning and Colin shares an example based on the role of published movie reviews.   Colin Camerer is the Robert Kirby Professor of Behavioral Economics at Caltech. His research interests include decisions, games, and markets.

In this episode we explore the economic and strategic foundations of networks – from personal connections in the labor market to complex webs of international trade and conflict together with our guest Matthew Jackson. He unpacks how networks shape opportunities, inequalities, and global dynamics, and how game theory helps us understand the power and fragility of these structures. Along the way, we discuss surprising applications, including the role of tariffs in supply chains and how network insights can explain patterns of cooperation and rivalry.   Matthew Jackson is the William D. Eberle Professor of Economics at Stanford University. His research interests include game theory, microeconomic theory, and the study of social and economic networks.

In this episode we are talking about corporate scandals, selfishness, and how game theory can help us understand when people give in to temptation. Carlos Alós-Ferrer explains how his "Big Robber" experiment shows that while people often act pro-socially in classical economic games, things change when individuals are given the chance to earn large sums of money at the expense of harming many others. We also discuss how insights from psychology and neuroscience enrich economics by highlighting the importance of process data such as response times.   Carlos Alós-Ferrer is Chair Professor of Economics at Lancaster University (UK). His research spans economics, psychology, and neuroscience, with a focus on decision-making processes and the interdisciplinary field of psycho-economics. He is Editor-in-Chief of the Journal of Economic Psychology and also writes the blog Decisions and the Brain on Psychology Today.