site banner

Small-Scale Question Sunday for August 16, 2026

Do you have a dumb question that you're kind of embarrassed to ask in the main thread? Is there something you're just not sure about?

This is your opportunity to ask questions. No question too simple or too silly.

Culture war topics are accepted, and proposals for a better intro post are appreciated.

1
Jump in the discussion.

No email address required.

It's really weird how different experiences everyone seems to have. The 2000 words when 200 would do seems to me like a general AI issue and if anything it's worse with Claude when I'm using it. Claude is less sycophantic, but I've had the opposite problem with it: It told me an approach wouldn't work because of a problem, that later Chatgpt found out was wrong - the approach did work as intended, and Claude was just being faux-critical for no reason. Admittedly it's partially bc of advice from my wife who is a very heavy AI user, I've started to remind every AI to be critical ("please look over this proposal, tell me potential problem fields, and critically evaluate it from the perspective of a domain-expert evaluator") to get around the sycophancy, which seems to work better with Chatgpt. It's nowhere near perfect and still operates at the level of an early PhD who still makes plenty of mistakes, but it's still amazing to get instant feedback with a decent accuracy rate for only something like 20/month.