I found this interesting. But the best approach is start with LLM, then check every point yourself, and summarize with real links. The moment we are ok with LLM output just once, it won't be just once, and things get too murky.
replyThe purpose of the exercise was to see what OpenAI thinks of itself to a large degree. I hope nobody takes the answers at face value considering they clearly have a conflict of interest at their very core. It has turned into an interesting social experiment though. There is a very real instant negative reaction to saying 'an LLM generated this' no matter the context or intent.
replyAnd the powerful win even more
reply