Learning to model other minds
LOLA algorithm enables agents to model other learning agents, discovering tit-for-tat strategies in iterated prisoner's dilemma.
Excerpt
We’re releasing an algorithm which accounts for the fact that other agents are learning too, and discovers self-interested yet collaborative strategies like tit-for-tat in the iterated prisoner’s dilemma. This algorithm, Learning with Opponent-Learning Awareness (LOLA), is a small step towards agents that model other minds.
Read at source: https://openai.com/index/learning-to-model-other-minds