International Joint Conference On Theoretical Computer Science – Frontier of Algorithmic Wisdom

August 15-19, 2022, City University of Hong Kong, Hong Kong

 

Invited Speakers

Track C


Reverse Engineering Human Cooperation and Cultural Evolution

Joel Z Leibo

DeepMind

Abstract:
Humanity's collective intelligence and adaptability critically depend on both our cooperativeness and our accumulating culture. Both rely on a set of social-cognitive capacities, representations, and motivations like theory of mind, reciprocity, and language. Thus I propose that we approach our goal of building cooperative artificial general intelligence through the lens of reverse engineering human cultural evolution. In this talk I will discuss how to use multi-agent reinforcement learning models of group behavior in spatio-temporally complex cooperation challenges as a testbed for these ideas. I'll describe some of the specific ways in which artificial agents often fail to cooperate, as well as some of the algorithms we have developed to overcome them and thereby cooperate in diverse social and physical environments involving both familiar and unfamiliar other agents.

Bio:
Joel Leibo is a research scientist at DeepMind. His research is aimed at the following questions: How can we get deep reinforcement learning agents to perform complex cognitive behaviors like cooperating with one another in groups? How should we evaluate the performance of deep reinforcement learning agents? How can we model processes like cumulative culture that gave rise to unique aspects of human intelligence?