People will ask questions on the edge of the domains they already know. If they can’t comprehend AI’s answer they wouldn’t be able to come up with the answer themselves anyway.
BTW, have you noticed it’s always “other people will be too stupid for chatgpt” not “I will be confused by chatgpt”? Don’t worry about the hypothetical others.
It is not hypothetical, some of us do have to worry about "others". I( so do many here) have to take a decision for all my teams whether to approve this or not.
It can be become a long-term handicap to developers learning, or also hell to staff engineers and architects who are reviewing and designing. The power of LLMs based code assistance to be transformative is significant and cannot be ignored either, so yes we need to worry
One one hand, I have experimented with co-pilot and this was my experiencerience - great when it worked, easy to fix when it didn't.
On the other hand, I worry people are not ready for this - get these magical answers and go double check them. Most people don't read the Wikipedia referenced they just trust it - are they going to double check LLMs?
