LLMs are far from the only thing you rely on that will confidently lie to you.
> Every time a new study comes around saying “actually, everyone should be drinking 3 glasses of wine a night”, do you take a trip to the liquor store?
No, and doing so would be an example of blindly trusting something that you haven't or can't verify, so that supports my argument?
> LLMs are far from the only thing you rely on that will confidently lie to you.
Sure, but it's a new class of thing that will and does and yet people are trusting or haven't yet learnt that they can't. I mentioned seeing people trust it via commit messages; I don't see SO the same way, people generally realise they need to verify it, or it at least has a voting mechanism as a proxy. With GPT so far there seems to be a lot more assuming it's correct going on.
But with ChatGPT we're not saying disregard what it says, we're saying only disregard some of what it says and don't disregard what it says in future. Which becomes a lot of work to check everything it says every time.