[pmlnr]

Answers, sure. Bit you'll never know if those are answers, bullshit, trolling, or straight up hallucinations. So no, it's not an option at all.

[adolph]

You write this as if that is substantially different from common epistemology of the first page of links.

I’m not an llm advocate but responses like this bring to mind the “no wireless. Less storage than a nomad. Lame” reaction to the iPod.

[dmje]

No idea what this means

[VancouverMan]

> you'll never know if those are answers, bullshit, trolling, or straight up hallucinations

That's how it has always been for purely human-generated content, too.

[dmje]

Exactly. This seems to be one of the skills we need to learn. In the same way we have to use our own intelligence and skills when searching google (Is this source website reputable? What’s the context I’m searching in? Do I know this person / company? Does that result sound likely? Can I compare to other answers to see if it makes sense?) - it seems to me that using ChatGPT to do anything requires a similar set of (slightly cynical / take what you read with a pinch of salt / check your facts - but nonetheless understand that this is a useful tool) type response to it.