Update: Apparently there’s an AI Poetry Turing test too, which most people fail except poetry experts.
You think AI art is pathetic? I used to because so many illustrations in articles these days look like the same ChatGPT-generated slop. But…
One of these two pretty hillsides is by one of history’s greatest artists. The other is soulless AI slop. Can you tell which is which? (Caption and images both stolen from AstralCodexTen.)
Before we go further, I invite you to take the AI Art Turing Test. Let’s find out whether you can really tell the difference between real human artists (good artists, no cheating). Most people did not do well on this Turing test.
Since there were two choices (human or AI), blind chance would produce a score of 50%, and perfect skill a score of 100%.
The median score on the test was 60%, only a little above chance. The mean was 60.6%. Participants said the task was harder than expected (median difficulty 4 on a 1-5 scale)
I have seen a lot of AI art on the web recently. And 90% of it is terrible. So what explains the above result?
Sturgeon’s Law. 90% of AI art is crap because 90% of everything is crap. The average schoolteacher is terrible but the top 10% will change your child’s life; so just as you shouldn’t dismiss the entire teaching profession based on the average teacher, you shouldn’t dismiss AI art and illustrations based on the average AI art you see.
I know this is bad news for some artists. I am not insensitive to the plight of artists, but I think ignoring the reality is not a good way of dealing with it.
I think a lot of mediocre artists will have to find alternative ways of earning their living. But the best artists are on a different plane.
Here is an interesting finding from Scott:
I asked a friend (who does digital art under the handle “Ilzo”) to beta-test an early version of the challenge. She wowed me with her ability to correctly identify AI pictures that I considered well-camouflaged. When we got to Piotr Binkowski’s ruined gateway - an AI picture I especially liked, but which she found especially slop-ish, I demanded she explain herself.
She said:
When real pictures have details, the details have logic to them. I think of Ancient Gate being in the genre "superficially detailed, but all the details are bad and incoherent".
…there are many more problems with the AI art that Ilzo pointed out. Read the original article for full details. Here are the other interesting bits from the article:
Most people couldn’t help judging art by its style: for example, everyone thought impressionistic paintings were human, and the painting below was AI (it wasn’t):
Mitchell Stuart’s “Victorian Megaship”, which 84% of people thought was AI generated. (Caption and image taken from the same article.)
Most people slightly Prefer AI art to human art. When people were asked to pick their favourites, the top 2 were AI, and 6 of the top 10 were AI art.
Even many people who thought they hated AI art preferred it to human art
Anyway, the important lesson of this article is this: many of you are dismissing AI because you haven’t been able to make it do interesting things and a lot of the uses of AI that you’ve seen in the wild are mediocre slop. But that means nothing—a flute in the hands of most kids is torture for the parents, but that doesn’t mean that a flute is useless as a musical instrument. AI is a tool and like any tool, it is incredible if you learn how to use it well.
