Well said and I fully agree, as a certified AI hater.
I think your post alludes to this, but to say it a different way - I know I am personally vulnerable to excessively offloading my decision making. I haven't ever used a chatbot aside from customer service ones, but I know it would be very easy to just turn my brain off and reflexively ask the bot what to do for everything. I know this because at times in the past I would copy paste stack overflow code that would have been the same level of effort for me to just type. And I see people near to me going through mild psychosis.
If this is fixable in LLM assistants along with what you mention, then I might start using them. Until that happens, it's dangerous.