Comment on Pro-AI Subreddit Bans 'Uptick' of Users Who Suffer from AI Delusions
whaleross@lemmy.world 10 months ago
I’ve been trying to configure ChatGPT tell me if I’m wrong in a question or statement but damn it never does unless I keep probing for support or links. I’ve been having the feeling that it has become worse with latter models. Glad but also sad to see I was right.
Anybody know other LLM that are more trustworthy and capable of searching online for more information?
Claude definitely has its impressive moments where it calls out something inaccurate.
It’s also way less sycophantic, mature and better for light coding.
My only issue is that the servers are sometimes slow and so is the ios app which frequently trows an error after 2 minutes if waiting.
There are no trustworthy LLMs. They don’t know it understand what they’re saying - they’re literally just predicting words that sound like they match what it was taught. It’s a only barely smarter than a parrot, and it has no idea how to research anything or tell facts from made-up bullshit. You’re wasting your time by trying to force it to do something it’s literally incapable of doing.
You’re better off researching them the hard way; check primary sources and then check the credibility of those sources.
Considering that parrots can have actual thoughts, I’d say LLMs are even less smart than that.
Claude 3.7 told me i’m wrong a couple of times. It knows how to search. I don’t have an opinion on 4 yet but it can search too
Lemminary@lemmy.world 10 months ago
On my Lemmy?? Impossible. I was just told by three different people rather condescendingly and assuming just as much about me that I was suffering from AI brain rot for daring to compare something to AI. The horror.