Warning: couldn't RTFA, because paywall, but think the comment has merit anyway as I use it for coding, math and physics issues alike. I think it is now clear

Warning: couldn't RTFA, because paywall, but think the comment has merit anyway as I use it for coding, math and physics issues alike. I think it is now clear that, for openminded people at least, this is the case in many fields. While you cannot assume that what AI (especially LLMs) are giving you is true or works, in fields where you can validate that, like math or software dev, AI can provide inspiration for where to look when stuck. Even if it's not perfect or even great, it does simply provide a basis or another view to tackle a problem. And a tireless one at that; no matter how crap the directions taken are and no matter how many times you tell it, it will come up with something else, equally bad or something better and in rare cases even very good. Sure this is just Prediction so this existed somewhere before, but I couldn't find it via Google and the experience of chatting and trying the result in mathematica / matlab / vscode is just not the same as finding some answer in google.