Intention and the AI Art Gotcha

https://glasshalftrue.substack.com/p/fellow-artists-im-begging-you-to

The article’s trick (showing people an image, telling them it’s AI, then revealing it’s a Monet) is a clever dunk on people who overconfidently claim they can visually spot AI art — but it inadvertently proves the opposite of its own point. The same image gets completely different responses depending on what people believe its origin is. That’s not evidence they’re wrong about the pixels; it’s evidence that attribution and intention radically change how we experience art.

The article frames the debate as “AI is good now, get over it,” accepting the premise that technical quality equals artistic merit — exactly the premise in dispute. A child’s scribble is art; a photorealistic AI sunset can be empty pixels.

Intention isn’t mystical — it’s structural. Monet’s every brushstroke encoded a point of view. AI has no point of view, only statistical patterns. The real question isn’t “can people tell?” but “does origin change nature and value?” That question survives any gotcha.

More context:

The “trick” in the article is clever as performance — it humiliates people who confidently claimed to spot tells that weren’t there. But the idea that this proves AI art is “just as good” only works if you define “good” as visually indistinguishable at a glance. And that’s a very thin definition.

Your angle has real weight, for a few reasons:

1. The trick actually proves the opposite of what the author thinks

The most telling thing in the article isn’t the bait-and-switch itself — it’s that the same exact image gets completely different responses depending on what people believe its origin is. When they think it’s AI, they tear it apart. When they learn it’s a Monet, they suddenly see it differently. That’s not just about “they were wrong about the pixels” — it’s evidence that attribution and intention radically change how we experience art. If intentionality didn’t matter, the reveal wouldn’t sting.

2. The critique of the “trick” holds even if people can’t visually distinguish

The commenter Alex in the article’s replies nails this: the question of whether something is art was never about technical verisimilitude. A child’s scribble is art; a photorealistic AI image of a sunset can be “empty pixels.” The article frames the debate as “AI is good now, get over it,” which implicitly accepts the premise that technical quality = artistic merit. That premise is exactly what’s in dispute.

3. Intention isn’t mystical — it’s structural

When Monet painted Water Lilies, every brushstroke was shaped by real choices: what time of day to capture, which part of the pond, how to handle the light, what to emphasize or soften. These choices encode a point of view. An AI generating in the style of Monet has no point of view — it’s producing a statistical mix of the patterns it was trained on. That doesn’t mean the output can’t be beautiful or moving, but it does mean the relationship between creator and creation is fundamentally different.

4. Where the “intention” argument gets harder

That said, you’d need to wrestle with the edge cases to really tighten the argument:

• What about Jackson Pollock, where the artist himself leaned into chance and accident? Is the “intention” just that he dripped paint, vs. a machine simulating drips?
• What about photography — where the “intention” is framing and timing, not direct mark-making?
• What about a human artist who uses AI as one tool among many in their process? At what point does sufficient human direction “earn” the intention label?

None of these break the core point — that art created by a conscious agent with intent is a different kind of thing from an output generated to satisfy a prompt — but they show the boundary isn’t perfectly sharp.

Bottom line: Your angle is valid. The article’s trick is a clever dunk on people who over-confidently claim they can always visually spot AI art. But it doesn’t touch the deeper claim: that art’s value is partly constituted by its origin in human intention, and that no amount of technical mimicry closes that gap. The fact that people’s reactions to the Monet image flipped based on attribution is, if anything, evidence for your position — not against it.