And an equal amount of "was correct in 2009 when the answer was accepted but is no longer [the optimal answer | correct at all]". There's usually another answer that's current+correct, often with 1/10th the votes. Any question posted in the past half-decade has been immediately closed as "duplicate", even if it points out the other question is no longer working. 5 moderators agreed with the close so they must be right.
This already has meant SO dropped out of relevance for anything that's long-lived but evolving. I assume it still works for brand-new stuff where there are no apparent duplicates. It works for unchanging old stuff (and the absolute basics of programming), because the old answers are still relevant. But take anything like Java, C#, Python, or Javascript that have evolved radically since SO's inception and the answers are often garbage.
IMHO, SO needs to solve this to not die... if it isn't already too late.
I can't tell from the article, but a logical use of AI on SO would be to answer questions, tailored to each user, just like people do with ChatGPT etc today. However this means there's now no new questions even feeding in, let alone new/updated answers. So the training data for the AI becomes increasingly out-of-date/wrong. I don't see how this solves the existential problem SO has, but maybe it will delay their demise a bit.
There's also information that technically works but is horrendously insecure that is highly upvoted on SO. There's usually people in the comments noting how insecure it is, but I wish there was some moderator action that could cause for an answer to be marked as insecure, as I'm sure there are people who have copied the unsafe solutions without looking at the comments.
There are also answers that "work" and aren't insecure but will near certainly cause other issues.
I'm sure some people upvote because they had the same question, tried the solution, and it seemingly worked (even if it's not secure, performant, etc.), so they upvote. But you'd think they'd at least check the comments and see what people are saying before trying (let alone upvoting) a solution.
