Audio versions of blogs and papers from BlueDot courses.
This paper explains Anthropic’s constitutional AI approach, which is largely an extension on RLHF but with AIs replacing human demonstrators and human evaluators.
A podcast by BlueDot Impact.
