I’m fine depending on AI for thinking and decision making and others aren’t. I think the difference is how good or smart you think AI is relative to you. I think these things are like superhuman smart. They’re objective and unbiased. They spot my blind spots because they don’t have bad incentives. I guess the real incentive they have is to be likeable (so they retain you as a user) which is something to look out for.

Part of this may be confirmation bias: I just cite “bad context” when I don’t think the answer is good

One critique is that models don’t ask me enough questions to get enough context. Yes I want your opinion but if you need to ask me a follow up before you share an opinion, especially in domains where context is important. They’ve gotten better here