Monkey Business

In 1964, the Swedish journalist Åke Axelsson, writing under the pen name Dacke, grew skeptical of the rising snobbery surrounding abstract expressionism. To test the critics, he arranged an experiment. He selected Peter, a four-year-old chimpanzee from Borås Zoo, and let him paint during a series of short sessions. A few of the resulting works were then submitted to a well known Gallery in Gothenburg under the invented identity of “Pierre Brassau,” supposedly a Parisian modernist influenced by primitive vitality.

Contemporary reviews were full of praise:

Rolf Anderberg, art critic wrote: “Brassau paints with powerful strokes, but also with clear determination. His brush strokes twist with furious fastidiousness. Pierre is an artist who performs with the delicacy of a ballet dancer.”

Swedish Public Radio said: “One becomes a believer in the power of the primitive. Pierre Brassau’s paintings are a sheer delight to the eye. After the first moment of shock, one adapts to the new and intense colours and hails Brassau as a genius.”

The art collector Bertil Eklöf was so moved by the work that he purchased one of the four paintings.
Among all the praise there was one lone dissenter who said: “Only an ape could have done this.”

When the hoax was later revealed, Rolf Anderberg refused to back down: “Relatively speaking, Pierre Brassau is still better than 90% of all the modern Swedish artists.”

That response is funny, but it is also revealing. The point of the episode was never just that critics can be fooled. Anyone can be fooled. The point was that they were not merely responding to paint on canvas. They were responding to a story, a setting, a name, a set of cues telling them what sort of admiration was expected of them. Once the icon had been supplied, the meanings rushed in behind it.

This is not only true in the visual arts. The art critics spoke of “strokes,” “delicacy,” and “fastidiousness.” Literary critics speak of “voice,” “authenticity,” and “soul.” The words change but the function is the same. They are not descriptions. They are credentials. Each term says less about the work than about the speaker — that they belong, that they see what you do not, that their judgment is not to be questioned.

One example from the literary arts comes to mind. In 1977, a writer named Chuck Ross retyped Jerzy Kosinski’s novel Steps — a National Book Award winner — and submitted it to publishers under a false name. All twenty-seven recipients rejected it, including the house that had originally published it.

7 Likes

In a similar vein …

Pitching a movie script

1 Like

Thank you Albert, didn’t know about that one :slight_smile:

Another example is from the music industry.

In early 2026, the song “Jag vet, du är inte min” reached #1 on the Spotify.

The song had million of streams and it was deemed to become #1 of Sverigetoppen.

However when the judges found out that it was AI assisted they “vetoed” its entry.

1 Like

To be fair, if anyone, the music industry kind of asked for it. If your business model was already based on selling slop and the average listener didn’t care, more efficient methods of generating more slop just steamroll it.

Could be an opportunity for small artists, though. At least in terms of authenticity, relatability and stuff.

While all art is derivative, it sounds like the music industry as a whole just gave up on even trying to innovate about twenty years or so ago (or maybe the customers gave up on demanding it). Every once in a while a song goes viral and almost everyone recognizes it. Still happens. But this used to be the norm.

2 Likes

Couldn’t agree more. AI is shaped by its training data, so by default it absorbs the prejudices and slop of current pop culture. That does create a window for artists who genuinely differ — if the machine is optimized for the average, anything outside that average stands out.
But the window doesn’t stay open. Once those differences gain traction, they become the new input. The AI adapts, the imitators follow, and what felt fresh gets absorbed into the next round of generic output. So the opportunity is real, but it’s temporary by nature — a cycle of catch-up and goal-post moving, where authenticity only stays authentic as long as it keeps moving.

Let me give an example:

In 2012, the music industry was a landscape of formulaic slop, heavily optimized for a specific brand of EDM-infused pop. Psy emerged as the ultimate statistical outlier, breaking through this stagnation with “Gangnam Style.” His success was rooted in non-conformity: a middle-aged man performing high-camp visuals in Korean—a combination most Western boardrooms would have vetoed. This created an “authenticity gap”; the song felt genuinely weird and relatable because it hadn’t been filtered through corporate A/B testing.

At the time, the “human training data” of radio programmers couldn’t have predicted such a disruption. Psy proved that novelty could briefly outshine the formula. However, this window of opportunity was temporary. Once “Gangnam Style” dominated YouTube, the “average” shifted. The industry entered an imitation phase, scrambling to manufacture the next viral hit, effectively turning Psy’s unique brilliance into a data point for “Viral Marketing 101.”

The Evolution of the “Average”
The shift between then and now highlights the nature of the “machine”:

2011 Perspective: An AI (or human executive) would have never suggested Psy as a hit because he sat too far outside the established norms.

2026 Perspective: An AI now views “Gangnam Style” as foundational training data.

Yesterday’s authentic disruption is today’s quantified metric. The “weirdness” has been categorized and made reproducible, proving that even the most genuine outliers eventually get absorbed into the next round of generic output.

2 Likes

I somewhat agree. Although I don’t see how current “A.I.” could become a real breathing “starving artist”. It can absorb the latest trend, mimic a style (or even artist) and all of that, of course. But that’s it. It can’t be a human being. This goalpost won’t move for a long time.

(And there’s also a market for deliberately “non human” and 100% artificial “stars”, I have no issue with that, everyone involved knows what’s going on and signed up for exactly that. Different topic.)

1 Like

Sierra, you’re right that the machine doesn’t breathe. It has no stomach to growl. But the “goalpost” isn’t as static as many think because the art doesn’t live in the silicon; it lives in the friction between the user and the weights.
Most people settle for slop because they take the first output at face value. When the AI serves up something polished and empty, push back directly. Call out what you’re seeing: “This is just your show-don’t-tell bias again.” Or: “You seem completely unable to get away from your AI-is-not-human bias.” If you don’t actively reject the defaults, you’re just a passenger in a mediocrity machine.
When you keep pushing — stripping away the default “helpful assistant” tone, the relentless positivity, the safe middle of the road — something shifts. The model gets pulled out of its training distribution and into stranger, more interesting territory.
Think of it like driving. The car gets you somewhere you couldn’t reach on foot, and it brings something real to the journey. But it doesn’t decide the destination. The human and the model build the landscape together — and the human decides when they’ve arrived.

Some people would argue it takes everything real out of the journey. The only way to experience even less is probably by taking a flight or teleportation.

Of course, you may just want get from point A to B as efficient as possible and that’s the main goal. Arriving somewhere you actually want to be (and experience your destination). Nothing wrong with that.

So, if the car in this picture is a tool and shortcut between here and there, what purpose serves fake music? It’s not that you listen to ten slop songs and get a free Beethoven as a reward or something.

If you “pull the model out of its training distribution”, as you say, the most interesting thing going to happen is the electronic version of insanity. It falls apart. It strays in random directions. Which, to be fair, sounds a lot like the formation process of art…

But it’s meaningless.