Monday, February 10, 2020

Artificial Intelligence, Turing Tests, Art, and Reframing

Art is any human creation devised to induce a reframing of perspective. (more definitions)

Computers can't freely reframe perspective (though they can be programmed to reframe in canned, finite ways). They're essentially stuck. That's what conveys the sense that computers are basically "dumb" - which we all recognize despite their prodigious calculative prowess. They're not hip. They can't get the joke. They're married to the page. They are, in other words, machine-like, and there's no higher threshold of yet-more-awesome-calculative-prowess that will allow them to transcend that (Kurzweil's "Singularity" be damned). Your bookworm friend who can't get a date won't improve his results by reading another 200 books. 

Maslow's hammer can only do so much, even if it's an outstandingly great hammer. The problem is that framing is a whole other faculty, completely unrelated to calculation. So while computers can certainly simulate artwork (someone has even built a writing bot), and even lousy art may spur a person to reframe, reframing is not something a computer can register, much less devise and anticipate. That's the part a computer is missing. 

(Who, exactly, frames?)

AI will only pass Turing tests via clever pre-programmed chicanery. The secret to effective Turing testing is to ply agile reframing and see whether the AI keeps up. But it won't. This will always be its limitation.

No comments:

Blog Archive