Supposedly they use "RLAIF", but honestly given that the first step is to "generate responses... using a helpful-only AI assistant" it kinda sounds like RLHF wi

Supposedly they use "RLAIF", but honestly given that the first step is to "generate responses... using a helpful-only AI assistant" it kinda sounds like RLHF with more steps. https://www.anthropic.com/research/constitutional-ai-harmlessness-from-ai-feedback https://www.anthropic.com/research/constitutional-ai-harmles...