I don’t know what the developers of AI are doing to prevent it from diverging from reality, rather than converging on the truth. Every single piece of AI generated content I have seen contains inaccuracies. It draws on existing information already on the internet, seemingly without having the capacity to judge if that information is correct or not. The lazy or uninformed use the AI content, without knowing enough to spot the inaccuracies, and it winds up again in a document on the internet somewhere, to be drawn upon by AI in future queries. It seems like the inaccuracies would get perpetuated and compounded until all you get out of AI is unreliable junk. The problem is that unless the user has a thorough understanding of the subject matter, they can’t spot the inaccuracies.
I have to assume they are trying to do something about it, but based on my attempts to use AI for what I do, anyway (also technical), it doesn’t seem to be working.
Of course I may also be completely misunderstanding how AI actually works.