LLMs can't be trusted. They aé like an overconfident idiot who is pretending quite impressively, butif you check on the result it's just a bit too much bullshit

LLMs can't be trusted. They aé like an overconfident idiot who is pretending quite impressively, butif you check on the result it's just a bit too much bullshit in the result. So there's practically zero gain in using LLMs except WHEN you actually need a text that's nice and eloquent bullshit. Almost everytime I've tried using LLMs I've fallen into thepattern on calling out, correcting and argueing with the LLMs which is of course in itself sillyto do, because they don't learn, they don't really "get it" when they are wrong. There's no benefit to talking to a human.