https://daringfireball.net/linked/2025/10/16/oatmeal-ai-art
@daringfireball AI art therefore fails for me in exactly the same ways AI writing does: a lack of intention. Mimicking creative choices, like the structure of a third act twist, without intending to say anything by it. It’s genuinely just sad!
Do I think it’s doomed forever? Kinda, yeah! I don’t see how something derived from the current tech is ever going to get there.
@daringfireball I felt by saying that finding out how a certain work was created, like Matthew states, can evoke an involuntary reaction. It is his own personal reaction that he cannot control, thus it is not a choice.
It's not different from finding out that seemingly inoffensive painting was done by Hitler. Or some "cheap" painting is an original Rembrandt.
The source has a reaction independent of the work itself.
"I think most 'AI art' today completely sucks. But not because it was made using AI generation tools."
The tools are built on trillions of dollars of theft, so for me AI art is fundamentally bad as their training method is illegal and immoral.
@daringfireball "If your opinion of a work art changes..."
This ignores the fact all generative AI content is a remix. Sometimes a remix is valuable.
I don't have access in my head to all the art that goes into training AI data, so I might get a moment of joy when I see it, thinking it's original. But when I learn it's AI, the joy is gone because I know the original pieces - the layout, the colours, etc are all ripped off from existing things and mashed together by a computer.
@daringfireball For me, the flood of AI photos and videos have taken the joy out of online media.
I used to enjoy the amazing photos or videos people posted because they were real, rare, and amazing, like an animal playing on a trampoline.
Now, there are so many fake versions, my default position is everything is fake. Kind of sad.
Art may become the same. Part of what I enjoy about art is seeing the artist's skill being displayed.
@daringfireball There’s a great moment in The Truman Show where we see Truman sleeping, there’s emotionally evocative music playing over it, and we, the members of the audience, are having an emotional response. Then the camera pulls back and we see that Christof is directing the music, the camera angle, whatever, and we realize how easily we can be played. In the movie, it’s making a point.
It’s the same thing with AI art. Nobody wants to be told that their emotions can be manipulated so easily.
@daringfireball “…you’re no longer judging the art. You’re making a choice not to form your opinion based on the work itself, but rather on something else.”
If I read something, like it, and then find out later it was plagiarized and think less of it as a work, that’s pretty valid and I’m judging the work and not the writer, right?
I think that’s a better analogy than the bizarre Woody Allen one you jumped to. We’re never not talking about the work when dissecting how it was made.
@billyok @daringfireball Take Dazed and Confused - compare to the original Jimmy Page copied - https://music.apple.com/us/album/dazed-and-confused/1100453959?i=1100454021 a song from Led Zeppelin 1 which, couldn't be more blatant plagiarism from the Jake Holmes original. "Inspired" it ain't. Plant didn't even bother changing the name of the song. He didn't add Holmes' name to the songs credits until very recently.
Nonetheless this knowledge in no way diminishes enjoying the Zeppelin version.
For those who haven’t heard the announcement I posted, songs from this point on will sometimes be split among multiple episodes, so this is the first part of a two-episode look at the song “Dazed a…
@daringfireball Be an adult and just accept that you can like Sora and it’s also shitty
This thing you’re doing where you bend over backwards to come up with a defense of this viscerally disgusting and humiliating slop doesn’t work
@daringfireball I really enjoyed this edition of The Oatmeal! I didn’t like this take on it.
The worst part to me is the belief that AI will just keep getting better. Uh… it’s getting WORSE. Look at it. And now it’s gonna be “training” on its own slop.
I thought for sure Gruber would be skeptical of the AI hype. But maybe it’s mandatory for tech journalists to buy into the scam now?
@daringfireball Took me a day or two to compose a reply.
Short take:
"Once we learn that a work has been generated by AI, this initial interest — this potential beginning of a human relationship with the work and its author — is suddenly replaced by a hideous void, as we realize we have been duped, have been presented with what purports to be a human communication but is now revealed as having come from a source that is profoundly inhuman and, perhaps, even anti-human."
Find the whole thing (5-minute read) here:
@daringfireball Curious how you consider, say, finding out a work of art you were appreciating was plagiarized. I suspect your opinion of it would indeed change pretty immediately, and that it would not be a choice. Say, looking at Roy Lichtenstein’s worst examples, which were effectively direct tracings.
Art is not devoid of the circumstance of its creation. This would come a lot closer to how Inman and others consider generative AI.
@daringfireball You really don’t understand LLMs, huh? Nor art, to be honest, but the latter’s never been your strong point.
The fact that your LLM takes are this uncritical is making me question my long-held assumption that you understand software better than I do.