Not because you can't distinguish between _one_ bad piece and _one_ good piece, but because there is so much production capacity that no human will ever be able to look at most of it.
And it's not just the AI stuff that will suffer here, all of it goes into the same pool, and humans sample from that pool (using various methodologies). At some point the pool becomes mostly urine.
If nothing else, it’s now much more feasible for them to be successful by sheer force of drowning out any “worthwhile” material.
Why does any human have to look at any art in your problem statement?
Random illustration on a random blog post sure could.
Art as an evocative expression of the artist shouldn't be cheapened. But those freelancers churning content on Fiverr aren't pouring their soul into it.
I sometimes put up the hero image on my blog posts if I feel it makes sense, for example: https://blog.senko.net/learn-ai (stock photo, ai-generated or none if I don't have an idea for a visualization that adds to the content)
Ghost writing rich people’s vanity/self-marketing trash business or self-help books (you would not believe how many of these are written every year). Images (and prose) for internal company department newsletters that almost nobody reads.
Great at that crap—because it doesn’t matter anyway.
Whether making it far cheaper to produce things with no or negative value (spam, astroturf, scams) is a good idea… well no, it’s obviously terrible. It’d (kinda) be good if demand for such things remained the same, but it won’t, so it’s really, really bad.
On the contrary. Slop makes the good stuff stand out.
Now, we have AI. Yours is still the only good one, but there are now 100,000 paintings.
How likely is it that your painting is still recognised as good, by someone who looks at 100 out of 100,000 of those paintings at random?
It doesn't matter that your painting is good, the discovery mechanism is shot to bits.