ChatGPT can provide the same answers SO can because it mined SO as part of its training data.
From my experience, the majority of answers (accepted or highest voted) are wrong/outdated, with the second or third being the correct answer.
Yes. Giant disorganized piles of atomized information are less useful than optimized integrated data representations queryable by LLMs.
From my experience, the majority of answers (accepted or highest voted) are wrong/outdated, with the second or third being the correct answer.