Updates in Cooperative AI

Mechanisms help humans to coordinate in society (e.g. auctions, elections, tax codes, pay scales, author orders) and good mechanisms can alleviate strategic conflict between humans. In this talk, Google DeepMind Research Scientist Raphael Köster will show examples of how to leverage AI, specifically a Deep Reinforcement Learning (RL) agent to serve as a resource allocation mechanism that can promote sustainable human behaviour in a common-pool resource problem. Importantly, this requires modelling humans in neural networks models, to create a rich and authentic simulation for the RL agent to learn in. The RL agents discover novel policies that lead to more equal outcomes and larger surpluses by skillfully dealing with free-riders and motivating all players to contribute.

We’re delighted to host this second seminar in our 'Updates in Cooperative AI' series. We're running these seminars monthly, and you're welcome to subscribe to our Google Calendar to stay up-to-date on all upcoming events. Please feel free to ask questions and provide input for the seminar here.
Speakers

Raphael Köster (Google DeepMind)

Discussants
Time

16:00 - 17:00 UTC 7 August 2025

Links
This seminar has now finished
Register Here

Raphael Köster is a Research Scientist at Google DeepMind. He holds a Ph.D. in Cognitive Neuroscience from University College London. His recent publications have focused on AI for human cooperation, particularly human-centred mechanism design.

Modelling humans with neural networks for behavioural mechanism-design