Previously: Store-Bought Glaze: Learning Craft With AI Text.
You’ve heard it without knowing you were hearing it. The lo-fi piano loop playing under the study-session playlist. The ambient swell behind the explainer video. The background track on the brand reel that sounds like music without quite being music. It fits. It functions. It asks nothing of you, and you give it nothing back.
That’s not a quality problem. The sound is often technically fine. What’s missing is harder to name, and naming it is the whole point of this post.
The series has been building toward this question. We’ve defined craft as skill: the technique required to execute something well. Aura, by contrast, is the perceived presence around a piece — the story, distance, wound, or intention that makes you care about the thing beyond its surface. Those two categories look like they should add up. Good craft plus aura equals art. But they don’t always compound. You can have one without the other. And the most common failure mode for AI-generated work isn’t what critics usually say it is.
The tool isn’t lazy. Nobody showed up to want anything.
The two places aura lives
Before we get to AI specifically, the framework needs one distinction it hasn’t made explicit yet: aura can live in two different places, and they behave differently.
Piece-aura is the aura that belongs to a specific work. It’s what you feel when you stand in front of a Rembrandt and understand, even without being told, that this object has a history and a presence you can’t fully account for. It’s the way a particular song feels like it knows something about your life. Piece-aura attaches to the individual artifact. It makes you want to know where it came from.
Creator-aura is the aura that belongs to the person making the work. Follow enough of someone’s posts, videos, or albums, and they accumulate a persona you track over time. The individual pieces may be thin (a fifteen-second clip, a quick sketch), but you keep coming back because you’re oriented to the maker, not the specific output. A TikTok artist or Substack essayist with strong creator-aura can hold an audience across thin material. The creator-as-performer is a real and legitimate model. It has its own integrity. But it’s a different bargain than piece-aura. The work is secondary to the performance of a self.
Both kinds of aura are real. Both can be art. But they’re not interchangeable, and conflating them is exactly where AI content tends to fail without anyone quite understanding why.
What AI floods tend to produce
The ambient tracks have no piece-aura because nobody built any. There’s no story attached to the piano loop: no originating moment, no wound it came from, no choice that felt like a risk. The lo-fi beat is optimized for non-intrusiveness. It was generated, filtered, uploaded, and it’ll sit on a playlist until it’s algorithmically replaced by another like it.
That’s not the machine’s fault. The machine does exactly what it’s pointed at. The failure is upstream, in the decision not to bring anything to the transaction beyond the desire to produce output.
There’s a lot of that out there. It’s not that AI attracts bad artists — bad artists exist in every medium. There are bad watercolors and bad novels. The tool isn’t the problem. The problem is that AI lowers the friction on the output side so drastically that people who would never have spent three months throwing pots can produce fifty tracks before lunch, and many of them genuinely believe that volume is equivalent to presence. It isn’t.
To make things worse, AI output often has the surface texture of craft. The chords resolve correctly. The sentences are grammatical. The image is composed. A technically deficient work at least announces its own struggle. Slick, competent, empty output announces nothing. It just sits there, occupying space that could be held by something that mattered.
Why creator-aura doesn’t fill the gap
The usual escape hatch, when people press on the hollowness of AI output, is to point to the creator’s persona. The artist builds a following. They show their process. Their story is interesting even if individual pieces aren’t.
Sometimes this works. The most successful AI-using artists, the ones genuinely moving audiences, tend to treat the production of work as a performance, not just an output loop. They show the iteration, the refusals, the dead ends. They connect their work to a coherent set of intentions you can track over time. In doing that, they’re doing something real: they’re manufacturing creator-aura with discipline and transparency, and the individual pieces become evidence of a larger project you’re following.
But creator-aura only carries you so far. It can’t permanently substitute for piece-aura in work that aspires to last. A musician with strong creator-aura who releases songs with no piece-aura will eventually feel thin. The fans who love them will say things like “I love what she’s doing, but nothing since the early albums quite got me the way that one song did.” That one song had piece-aura. The others coasted on borrowed presence.
This is the real stakes of what’s flooding the zone right now. A lot of it is creator-aura-free and piece-aura-free simultaneously: anonymous channels producing ambient tracks, stock image accounts posting AI illustrations without attribution to any human project, text generators filling blog posts with no point of view. Nobody is performing anything. Nobody wants anything from the audience except a click. The loop of aura — making something that earns a response from a person and carries that response forward — never closes.
Craft without aura is still craft
Here’s where the argument can go wrong, and I want to be precise: none of this means the work is worthless.
Craft without aura is still craft. A technically correct piece of ambient music might serve a real function — it fills a waiting room, it provides a study environment, it does the job it was hired to do. That’s fine. Commercial function isn’t an insult. A lot of human creative labor throughout history has been craft-for-hire with no pretension to aura, and it served its purpose without embarrassing anyone.
The problem isn’t craft without aura. The problem is craft without aura being mistaken for art: pushed into contexts where aura is what matters, where audiences are expected to care, where the implicit contract is that something was made for them specifically to feel. When the ambient track shows up in that context, when the AI illustration appears on the cover of a book that’s supposed to mean something, when the generated essay tries to build a relationship with a reader, the absence lands hard, even if no one can immediately articulate what’s wrong.
Chiang’s argument, as I’ve read it summarized, is roughly that current generative systems are designed to spare users from making choices — that the statistical averaging built into the model fills in most of the decisions that, in more demanding creative work, would be the actual craft. There’s something to this as a diagnosis. The default output is smooth because it’s the average of everything the model has seen. Aura, almost by definition, isn’t average. It comes from specificity, from choices that didn’t have to be made that way but were, from the willingness to be recognizable and therefore vulnerable.
