Comment by aabdi

You’re complicating things.

There’s no reward for prosocial in llm rl as compared to other targets.

Humans have it since prosocial and others have evolutionary reward signals that do.