If a tool like this is currently only suitable for specific and minor cases under human oversight, how does it prove any better than a human? Wouldn't one of the only novel and useful cases of "AI" be general intelligence that is able to parse events in real time, instead of from manually selected information that we are so quickly running out of? We are so far from that and being sold Siri / Cortana for the third time.

I admit I'm definitely biased in this - even if information presented by one of these "AI" was factual, I would still take it upon myself to check. I don't trust their works at all.

> If a tool like this is currently only suitable for specific and minor cases under human oversight, how does it prove any better than a human?

Not to defend Grok, and I agree with your point about checking, but you can also say this about a hammer.

Not unless you wanna punch a nail in with your fist.

"AI" only does things we can do, because to do otherwise would be evidence against the general, human level intelligence that the marketing behind these abominations are so desperate for. The catch is they do it quicker, sometimes much quicker, but always much worse.

> This is just plain user error.

How are users supposed to know that that's an incorrect use of grok?

The secret is that there is no correct use of grok. Or any of the rest of these things, for that matter.