In terms of making money? Yeah probably, VCs have no idea what’s going on.

In terms of creating content that’s increasingly hard to distinguish from human effort? Holy shit, we’re just getting started. A tiny fraction of people are realistically able to use primitive models. Right now is the hardest it will ever be, again, and right now is the worst the results will ever be, again. And it’s already good enough to make still images kind of a crapshoot even for people who’ve plowed through a ton of examples.

And hey - some people won’t care. Art sites were always awash in content users had to sift through themselves, and now they’re flooded with ten thousand variations on any subject imaginable. If you see a gap… you can fill it, just by having a nice GPU and the patience to Google stuff. Even those hurdles will slowly disappear. Hell, the sites themselves might just make shit up whenever you search. Beats saying ‘lol nope sorry,’ right?

Video’s coming next. It should be a thing already, since Stable Diffusion’s whole deal is denoising, and it seems pretty tractable to guess what’s valid from one frame to the next. Every movie is a training set of one hundred thousand examples.

Audio’s already half-decent, and that’s going to have immense economic impact - at least in video games. Actors are worried about being cloned as if the only possible use of a voice generator is replicating an existing person. Nah: some writer is going to tune a character to sound exactly like the voice in their head, and it’s going to have a thousand unique lines, before asking an LLM to vamp and provide ten rephrased versions of all the boring ones.