“Artificial intelligence” (#AI) is the wrong term for the many services for which the term is being used. These are not generating intelligence. They are generating text or images or music—that is, information.
But it's not real information. It's ersatz* information. So when you hear "artificial intelligence,” I suggest that you substitute “ersatz information.”
This is of a piece with the discussion between Melanie Mitchell and Alison Gopnik that I just posted, at almost the precise second you posted your comment 😂
@clive Yes, that’s a good conversation, and I agree with them that Ted Chiang had a brilliant perspective early on. Put aside the seductive questions about whether we should be scared or awed by the machines/algorithms. Stop worrying about them and worry about the stuff they generate.
@JamesGleick It's Fake; to borrow a line from Eric Drexler (nanotech guy) AI is statistical herding. Crude. But a good word for multitudes of gullibles is simply Fake. It is fake intelligence or FI. Or maybe GI for gullible info... Too bad the orange buffoon has overused fake.
@JamesGleick the funny thing that I have noticed, is that the generated productions appear to be information. But then, it turns out that they are not. Is wrong information still information? I don’t know 🍸😺 It’s like a long drawn out version, with a semantic component (albeit stochastic ), of the famous old Chomsky example of a senseless yet grammatical sentence: “colorless green ideas sleep furiously“. 🍸😺
@JamesGleick funny how Feynman was playing with this back in the 80s. I think if the LLMs incorporate a truth vector/chain of custody they can become [more] reliable stores of information. The tricky part is where do you get this 'truth'...