Artificial Empathy: Towards a Framework for Unsupervised Agency Detection and Policy Reconstruction
2026-08-24 • Artificial Intelligence
Artificial Intelligence
AI summaryⓘ
The authors explore how an AI can recognize and understand other agents it sees, just by watching them, which helps it work together better. They point out that this problem is more open-ended than a related area called inverse reinforcement learning. To tackle this, they use a reinforcement learning agent trained on a separate task to guess how other agents behave and what their goals are. This approach helps the AI identify agents and predict their actions without extra information.
AI agentsagency detectionpolicy reconstructionreinforcement learninginverse reinforcement learningagentic dynamicscooperative behaviormulti-agent systems
Authors
Peter Kuhn, Chris Pang, Sonakshi Chauhan
Abstract
We study how an AI system can identify and model other agents in its environment from observation alone, which is a capability necessary for cooperative behaviour in the real world. This problem is less constrained than inverse reinforcement learning and remains largely unexplored. We propose a framework that uses a reinforcement learning agent, trained on an independent task as a prior about agentic dynamics, to perform agency detection and policy reconstruction.