One of the problems I see with AI discussions, and many unproductive discussions, is that discussions are total, that we can never discuss what works and what doesn't.

> it has had a large corpus of human creations to steal.

I certainly have to roll my eyes at companies trying to justify torrenting Anna's Archive, and reprocessing of current material into functionally equivalent material is problematic. But reading a bunch of stuff written on a topic and producing a new synthesis is what humans do all the time.

> Once humans are priced out of the market and stop producing works, what's AI going to be trained on?

An existing AI doesn't need training; a translation AI or image-producing AI could go years without updating without issues. I'm sure that work is being done right now on producing results from smaller corpuses and producing more refined corpuses.

> Nobody's expecting everyone to spend hours to learn to draw.

>> "Everybody can do it; but not everybody wants to do put in the effort, the practicing time and the soul."

is the quote I was responding to.

> We have divisions of labor; some people like making art, some people like making software

Note that we've usurped these divisions before, repeatedly. The camera stole jobs from portrait artists. The history of making software, particularly the first half-century, is a history going from people who rearrange physical computing units, to those writing machine language, to assembly, to FORTRAN and COBOL, to BASIC and Java, for many users to RPGMaker and Excel. They're letting people who have never heard of de Morgan's laws program nowadays.

> But society is not prepared, IMO, for the massive social upheaval caused by creators' livelihoods being decimated by AI.

As I said, we've been here since the Jacquard loom. I'm not blasé about the future, but we're rolling forward. To most adequately respond to it, prepare for it, or direct it, it's important to be honest about what AI can and can't do, and what it will and won't be able to do in the future.