Skip to content

Comment on Understanding Agent Cooperationparent

Comments

we simply gave software a specific set of rules to train it. It doesn't know how we perceive the actions it is performing.

I think the author is making the same point from another angle.

The AI learns what we would consider aggressive moves when conditions favor those moves.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.