The problem you have with AI is that there are some very powerful forces at work to convince people its answers are always reliable and complete..
I'm sorry, please stop that bullshit with conspiratorial spice on top if you want to be taken seriously. You could replace AI with Wikipedia and make as much sense, or "The Media". What example would you even go with here. Grokipedia? Elon Musk's attempt to convince people that Elon Musk's answers are always reliable and complete?
The problem with AI is NOT that there aren't enough disclaimers that the answers may be incomplete and unreliable. That is the standard disclaimer on every god damned LLM service, and we all know it, and we all talk about it. It's the reason "Grokipedia" makes us all chuckle just thinking about the idea of it and knowing nothing else.
LLMs are a tool, and when they mess up, YES we attempt to explain why and prod around at how to avoid pitfalls. But reading that as there being a shadowy cabal of AI apologists convincing you up is down and AI is always correct, that is straight up looney.
They're not a magic box, they fuck up, they're also really useful. Check their answers, they're not even stupid. The PROBLEM is most people are. Stupid. You will NOT fix that by telling them AI doesn't think. You won't fix it telling them it's a magic box not to be trusted. You won't fix it by telling them it's just a computer not to be trusted. You can't fix stupid, and as you pointed out, explaining how someone else's source of information is wrong goes over like a lead balloon. You have Joe Plumber over there thinks the loaded gun he's carrying for protection against "Them" shouldn't have a safety selector because it will only slow him down. If anyone thinks the one thing that guy, representative of our world of idiots, needs to know is that AI can't think and the PROBLEM isn't his overconfidence, it's Them convincing us AI is flawless... look this is stupid.