Yeah, I've seen that a few times as well, but it's more like it has one reputable link it can show you but its conclusion is based on forum posts that reference it or something. But I personally treat it like any other AI, it isn't telling me facts, it's telling me it's opinionated/flawed suggestions based on search results. I'd say it's about 60% reliable at best, but that's 60% of the time where I didn't have to open multiple tabs and spend minutes on each. that 40% does waste time, but at least for my searches, it's usually things I would have tried or looked into anyways if that makes sense. It depends a lot on the topic too, like I'd never take medical advise from it, but it does give good advice, and when I do use actual search results after reviewing that, I look for anything that contradicts it to see if it's the AI that's wrong or the site.

I've had programming problems I was stuck on (APIs mostly) where it found a quirk and then a work around for it. The docs, forums, every single link I manually clicked on page after page didn't mention that. Like you said, even what it linked didn't say that. But the AI was right, I would have given up or shelved the thing if it wasn't for the AI.

That's my struggle with this thread, it is just a semi-reliable tool, but it's very useful. How can I pretend all those times it did help me a lot didn't happen?

It can be useful IF you can spot the problems in the generated answer. What is obvious to a domain expert can confuse a regular user just searching for help.