On the Utility of Learning about Humans for Human-AI Coordination

TL;DR

This paper investigates how agents trained with self-play and population-based methods perform poorly when coordinating with humans, emphasizing the importance of learning about human behavior for effective human-AI collaboration.

Contribution

The paper introduces a simple environment based on Overcooked to evaluate human-AI coordination and demonstrates that agents trained without human models struggle to coordinate effectively with humans.

Findings

Agents trained via self-play perform poorly with humans.

Incorporating human models improves coordination with humans.

Human-aware planning enhances human-AI collaboration.

Abstract

While we would like agents that can coordinate with humans, current algorithms such as self-play and population-based training create agents that can coordinate with themselves. Agents that assume their partner to be optimal or similar to them can converge to coordination protocols that fail to understand and be understood by humans. To demonstrate this, we introduce a simple environment that requires challenging coordination, based on the popular game Overcooked, and learn a simple model that mimics human play. We evaluate the performance of agents trained via self-play and population-based training. These agents perform very well when paired with themselves, but when paired with our human model, they are significantly worse than agents designed to play with the human model. An experiment with a planning algorithm yields the same conclusion, though only when the human-aware planner is given the exact human model that it is playing with. A user study with real humans shows this pattern as well, though less strongly. Qualitatively, we find that the gains come from having the agent adapt to the human's gameplay. Given this result, we suggest several approaches for designing agents that learn about humans in order to better coordinate with them. Code is available at https://github.com/HumanCompatibleAI/overcooked_ai.