In terms of the impact thus far of procedurally-generated graphics, so far it is only supporting specific kinds of games, namely visual novels. That genre specifically is both art-intensive and programming-lite. As a result, “AI” art is encouraging for individual developers with minimal experience but lacking the ability of confidence to produce their own art assets. However, the technology is deceptively ‘easy’; turning algo outputs into non-horrific objects is time-consuming, the cost of hardware and power is non-trivial for the average person and the data they’re trained on is biased towards specific kinds of images.
It might also see use in RPGs but the utility outside specific genres is limited. Yes, we’ll see more games, but it will not result in more variety - nor will there necessarily be an increase in quality. I think it will make it even less likely for group projects to happen, since there’s one less reason for people to team up with a writer or an artist, when they think an algo can do both jobs for them.
Human art is not going away. What we are going to see is a combination of the cost of procedural generation going up as venture capital behind it starts looking for an ROI, with a concurrent push by art spaces and associated business to seek damages for unlicensed use of art. I highly doubt it will go fully uncontested, albeit only from those with the ability to react. A collective of artists can have an impact, but without union levels of organisation, there’s a risk of spaces like deviantart simply settling any disputes for their own benefit, such as using algo access to uploaded content as a source of revenue.
In addition to the above, all art generation tech around today is heavily flawed and in most cases, isn’t reliable enough for high-end commercial applications. These are just some of the things that will stop it being used everywhere. Even as it looms large, I think we’ll ironically see artists paid to clean up its mess of black voids, extra arms and wavy buildings. Giving control over this tech to artists would be super cool, honestly but lbr artists will get priced-out of the tech the moment it starts getting good.
Despite the above, automation has already devastated other jobs and I don’t see any way to fully close the pandora’s box of algos and techbros and people on the edges of the market are gonna get screwed. Specifically, artists reliant on small-scale commissions eg; fanart, personal avatars, OCs, with their commission pages on twitter. It’s not a problem of technology so much as a problem of technology under capitalism, where the worker gets the short end of the stick via whatever means available and corpoate interests will block any attempt to protect those actually doing the work that algos feed upon.
Government support for the arts is the sort of thing that can help short of legislation restraining unethical use of algorithms (good luck getting any good laws passed in the US). All of the (non-fetish) art spaces I’m in are also alarmed but most are already talented enough that their labour can’t be reproduced (yet). It will hit hardest on those with less experience (especially those still studying) and/or in niche roles (you mentioned colourists, but lineart, shading, typesetting are some other jobs non-artists here might not know about) that are easier for an algo to replicate.