[sddhrthrt]

I still don't understand why these models can't be more "trustworthy", but I also don't understand the theory, I'd love to hear what you all think about this.

I asked it a question that's pretty subjective and culturally specific, and I appreciate that I got a reasonable answer back. The question was "should I?" and the answer was "definitely, don't miss it" in three different ways. However, I found that the literal sources it quoted didn't have the same opinion it expressed to me so convincingly. I asked a clarifying question and it goes "okay so I read the material, and it actually says it's optional".

So why not read the material? I wonder if it could even embed the website in the results, giving the website the traffic and ad space. I wonder if a meta browser is a better product for these tools.

https://screenshot.click/13-56-232ze-p3nzf.png

[brap]

I think one factor is that all these LLMs are tuned to be ridiculously agreeable, almost everything you say will be met with some variation of “you’re absolutely right!”.

It’s like, look, I’m definitely not “absolutely right” 90% of the time, so how the hell am I supposed to trust what you’re saying?

I would prefer a model that’s tuned to prefix answers with “no, dumbass. Here’s why you’re an idiot:”. And yes you can promot them to answer this way, but they’re simply not wired to challenge you except for very trivial things.