Art as a Casino

It is tempting to treat artists as R&D researchers. Most artists aren’t making products; they’re doing R&D. That’s why they make no money. Very few artists like Taylor Swift are even interested in the product side of things. By this view, Disney’s best bet is hiring hundreds of artists to just do whatever and greenlight the most commercially compelling work.

Groups like the Midnight Oil Collective are trying to do exactly this by branding themselves as a “YC for artists.” They treat art like a venture-backed startup, providing “pre-seed” funding for artists to develop prototypes. It’s a noble attempt to formalize the chaos, but it fundamentally misunderstands the scale of the waste involved in art. An incubator can help a software engineer build a database because databases have utility; an incubator cannot help an artist find “the vibe” because the vibe doesn’t exist until the market creates it. You can’t optimize a lottery ticket.

This strategy doesn’t work for art. The closest equivalent that actually works is A24, the “cool” indie studio that has become a global tastemaker. Recently valued at $3.5 billion, they’ve built a lifestyle brand around “prestige” cinema, but they aren’t like YC in the sense that they don’t build from scratch. They go to film festivals and acquire finished R&D projects like Everything Everywhere All at Once or The Brutalist. In this sense, they’re less of an incubator and more like a private equity firm. Moreover, it doesn’t scale. This boutique model works for A24 because they can thrive on “small bites” of profit from $20 million movies. A studio like Disney, however, has such massive corporate overhead that it requires “Four-Quadrant” hits—billion-dollar behemoths that appeal to everyone simultaneously. You cannot reliably find a global, toy-selling franchise at a niche film festival. Disney needs hegemony, and hegemony doesn’t happen through auteur discovery; it happens through industrial engineering.

You can’t treat the art industry like tech because unlike tech, the art industry is fundamentally broken. It relies on irrational actors and extreme outliers. The math is terrible no matter how you cut it. The cost of R&D is astronomically high, not just in money, but in the years of uncompensated time an artist spends mastering a craft that the market might never want. The vast majority of artists go broke, having spent their prime years chasing a statistical impossibility. However, the payoff of becoming an outlier like Lady Gaga is so massive—both financially and culturally—that this asymmetric upside keeps the engine running. It’s a lottery where the ticket costs ten years of your life, but the jackpot is immortality.

And who is paying this cost? If it was a rational market, you’d expect the VCs and companies to come in here. But in art, the Disneys of the world sleep. They let the artists subsidize their own R&D. Artists are expected to work forty hours a week at a “day job” to pay for the privilege of doing their R&D at night. They buy their own equipment and fund their own marketing. As a result, corporations get a massive pool of “pre-vetted” talent and ideas for free. They don’t have to bet on who might be talented; they just wait for someone to go viral on their own dime and then offer them a contract.

But why would artists do this to themselves? It is brutally inefficient and even cruel for an individual artist to work for years on a project that might never pay off. Economists and researchers have a name for this exploitation: the “passion tax.” This is the systemic reality where the more an employee loves their work, the more the market feels entitled to underpay them. Because artists are intrinsically motivated to create, the industry can suppress wages to the point of subsistence, knowing the “enthusiastic” worker will keep producing even without a fair financial return.

The only reason this isn’t considered pure exploitation is Psychic Income. Artists receive a “wage” in the form of identity, status, and emotional fulfillment. To the artist, the work is the reward; the money is just a bonus. To the corporation, the artist’s fulfillment is a subsidy that lowers their production costs.

This is why initiatives like Adobe’s Film & TV Fund or Ireland’s Basic Income for the Arts (BIA) are so misguided. Ireland’s BIA pilot literally hands out hundreds of euros a week to artists just for being artists. While boosters point to modest gains in artistic output, they ignore the quality dilution. By removing the evolutionary pressure of the “all-in” lifestyle, you stop producing outliers. Paying an artist a living wage to do R&D doesn’t make the art better; it just makes the corporation’s eventual acquisition of that art more expensive. A government-subsidized artist is no longer an R&D researcher; they are a Civil Servant. Real breakthroughs require the existential risk of failure.

This phenomenon creates a “Barbell Economy” where art becomes both cheaper and more expensive simultaneously, while the quality splits into two extremes. On one end of the barbell, you have $300 million blockbusters. On the other end, you have a sea of “content”—ultra-cheap media produced for pennies. The “middle class” of art—the $20 million mid-budget movie—is being squeezed out because it carries the risk of R&D without the safety of a massive franchise.

Enter AI

So how does AI art fit into this “business model”? One might assume this shifts things by making the R&D free. I suspect a lot of people in Silicon Valley have this sentiment. But what they miss is that the R&D was already free. As in, the artists were already eating the costs themselves. The problem for Disney, HBO, and Apple TV+ is not creating R&D but rather filtering out the garbage.

The vast majority of new art is terrible. This is not a fault—R&D is allowed to be terrible, that’s the whole point. But it creates a problem when trying to commercialize. Because there is no objective metric for what makes art “good,” we have to wait for the market to decide what sticks.

This isn’t my fringe out-of-touch high-tech take. Just listen to the argument made by critics like Anthony Fantano, who frames the overwhelming volume of new music as a fundamental distortion of our perception. In his article “There’s Too Much Music,” he cites a core statistic driving this shift: “According to a Music Radar study talked about in NME, apparently more music was released in a single day in 2024 than in the whole of 1989”.

When you realize the sheer scale of modern output, you realize that the common sentiment that 80s music was “better” is actually just a classic case of survivorship bias. As Fantano and other analysts have noted, we aren’t comparing the average song of 1985 to the average song of today; we are comparing today’s unfiltered, total output to the hand-picked, top 1% highlight reel that was good enough to survive forty years of cultural filtering.

An AI model that can predict what art will be good ahead of time will make billions. Except, such a model is impossible. In music and film, there is an incredibly low signal-to-noise ratio. Studios are already trying to solve this with big data, but it has led them straight into the “Safe-Hot Trap.” They only greenlight “Safe” bets—sequels, reboots, and established IP—because they have a predictable, albeit mediocre, floor. They are terrified of “Hot” projects—the original, weird outliers—because they can’t predict them. This is why we have twelve Fast & Furious movies but almost no high-budget original sci-fi. Finding a black swan outlier is essentially random noise, and no algorithm can predict a “vibe shift” before it happens. The market is the only engine capable of doing this reliably because it’s the only system large enough to process millions of human reactions simultaneously.

A model that would have an accurate enough understanding of the world to predict success accurately wouldn’t be predicting art. It would be predicting stocks and wars. It’s the closest thing to an AI god.

So what does the future of AI look like in this space? I argue its purpose will not be to make R&D cheaper, nor will it predict success. AI will be used to throw everything on the wall and see what sticks. This is how I predict the pipeline will go. A “director” will create 100 movie trailers using generative tools, upload them to TikTok, and see what the market decides is most compelling. They will then rework and flesh out the concept into a finished product and release it in theaters. This will happen to everything from music to books and perhaps even video games. Game jams already fulfill a similar purpose there.

Critics argue that humans require a “soul” to form the deep obsessions that lead to the Immortality Jackpot. This is nonsense of course. A “soul”, so much as it can be said to exist - can be manufactured. We already see this with Hatsune Miku or K-Pop groups, where “Human Will” is distributed across a corporate architecture. In my pipeline, the Director becomes the face. People won’t follow the AI; they will follow the Director who has a 100% track record of picking the best vibes from the noise. The “brand” moves from the performer to the director.

There is an elephant in the room I have not addressed. Artists hate AI. They usually give a host of reasons for this, but the most common rhetoric is that AI steals art. This doesn’t make any sense because it implies the artists were getting paid in the first place. But if you recall our economic model, you’d know that money was never part of the contract. The cold hard truth is that you didn’t become an artist to make money; you did it because it’s what you loved. The real reason artists don’t use AI is because they don’t find it enjoyable. The aspect of creation they hold most dear is going to be completely automated in my proposed pipeline.

This is fantastic news! It means my pipeline kills the passion tax. Today’s displaced illustrators are the manual typists of the 1950s; they find fulfillment in the tactile nature of the work, but that labor is now obsolete. This new industry will attract an audience who don’t care about “brushstrokes”—they call themselves Product Managers, Creative Technologists, or Prompt Architects. They want to optimize a product and get paid reliably. They find their fulfillment in Strategic Success rather than execution. The art industry stops being a casino and starts being an investment bank.

Runway’s “The Hundred Film Fund” is the first true VC of this new industry. By promising to fund 100 films that use AI, they aren’t looking for a “star” to subsidize; they are providing the capital to accelerate the “throwing everything at the wall” phase. They are the high-frequency traders of the art world.

Critics point to the current “signal-to-noise” crisis on social media as proof this won’t work, but they are looking at the messy “dial-up phase” of the industry. As we decouple “Labor” from “Art,” the industry will stop being a cruel lottery and start being a high-precision engine.

I’m sure artists are going to mourn and dub this “the death of art”. In many ways they’ll be correct. The wild west of art wonderland they hold dear will be gone. Personally, I won’t be too bothered. People who love creating for the sake of creating can still do so. The difference is that their passion will no longer be exploited by the fantasy of commercial success. If you want a soul, go to a museum; if you want the future of culture, look at the data-stream.