@nextcloud Every time I use it for things I know a lot about, I can tell immediately that it produces complete BS, even if it sounds very confident and even competent - IT IS NOT.

So when I use it for things I know less about, why should I trust it? It may be equally confidently wrong, I just can't tell.

Why anyone chooses to use this is beyond me. On top of concerns like energy use, ethics of training data, legality of evidence ("enhancing" of phone cameras/generative AI) …