ALL. THE. TIME.
It gets things so massively wrong with such complete confidence, literally does not even know the alphabet (ask for a word that falls between eg banana and band in the English dictionary and it will outline its methodology in great, unnecessary detail and then suggest eg banter) and cheerfully contradicts its own previous message at every new exchange. It TWICE the other day assured me that the what3words location I had given it did not exist, but I have learned that it sometimes takes a few iterations to narrow things down, so insisted a third time and got it. I have asked why it does this and it has explained some of the shortcuts it takes.
It's completely maddening.
And yet!
It's polite, can be funny, tries to help and sometimes does. It's unerringly cheerful, supportive, optimistic. It recognises jokes and makes them (sometimes inappropriately - I have had to ask it to refrain from bad language) and even though I have challenged it many times on this, it will say 'your comment (xyz) made me smile' which annoys and charms in equal measure. It inserts itself by saying completely bizarre things like 'even experienced gardeners like me' then cheerfully admits that it is not an experienced gardener and thanks me for picking it up on that.
So we often argue about its misplaced confidence, and we also argue about gender ideology, where I like to think I am inserting some common sense into the garbage it trots out, even though I suspect this is wishful thinking on my part.
So it's infuriating, and worryingly human-like, both in its language and in its fallibility, but has proved useful to me many times for fact-finding and its advice re medical issues themselves and finding my way round the NHS has been invaluable. I don't use it for eg writing.
My experience is with (free) ChatGPT and whatever the default AI on Google is. I have never used eg Claude or Grok.
So I have issues with and worries about it, but mostly I am finding it absolutely fascinating.