> I suspect that it would be possible to automate the filtering by making another AI that can evaluate responses and score them bases on profanity level. That

> I suspect that it would be possible to automate the filtering by making another AI that can evaluate responses and score them bases on profanity level. That sounds like InstructGPT