# OpenAI Develops AI That Learns to Cooperate by Modeling Other Minds

OpenAI has announced a new algorithm called Learning with Opponent-Learning Awareness (LOLA) that represents a breakthrough in how artificial intelligence agents interact with each other.

Unlike traditional AI systems that treat other agents as static obstacles, LOLA recognizes that other agents are also learning and adapting. This awareness allows it to develop sophisticated strategies that balance self-interest with cooperation—similar to how humans navigate social situations.

The algorithm successfully discovered "tit-for-tat" strategies in the classic iterated prisoner's dilemma, a game theory scenario where cooperation can benefit both parties but requires trust. Tit-for-tat is a proven strategy where an agent cooperates initially and then mirrors the other agent's previous move, encouraging mutual cooperation over time.

**Why it matters:** This development moves AI closer to understanding and modeling other minds