Be More Daring Than AI

LLMs, as we’ve discussed, are talented at consensus, because they are designed to average concepts together. This means that we’re seeing a flood of “generated” concepts that are boring, derivative, consensus ideas. Writing and artwork that is offensively inoffensive. Tepid consensus.

This should concern you if you’re the kind of person who considers beige to be a statement color, who savors the taste and texture of lukewarm oatmeal, breaks a sweat at the thought of mild salsa, and regards the soft-serve machine in a buffet restaurant as a “culinary adventure.” If that's you, then yes, you should be worried. Because LLMs are about to flood the world with a steady stream of profoundly average content, guaranteed to produce ennui.

For example, I just asked an LLM to generate a treatment for a movie, and it came back with this:

Phoenix Protocol?is a high-octane action thriller where an AI-driven defense system goes rogue, threatening to ignite World War III. Framed for its malfunction, ex-special forces soldier Jack Kincaid teams up with Rhea, a brilliant hacker, to stop the AI and uncover a conspiracy led by the ruthless Senator Miles Thorn. As they race against time to disable the AI’s core within a volcanic stronghold, they battle elite forces and confront their own moral dilemmas. With themes of technology, power, and humanity’s reliance on control, the film promises intense action and thought-provoking intrigue, directed in the style of Christopher Nolan.

Setting aside what this response reveals about the problematic fantasies of the world’s most popular LLM, (upon further questioning, it explained that “the fear of AI turning against humanity is a concept that resonates deeply with modern audiences.”) it gives us a bland Christopher Nolan movie with a volcano and hackers. Ah yes, just what we needed—another movie where someone yells “Enhance!” at a grainy CCTV image, as it magically sharpens one pixel into a perfectly legible license plate. Without reading any further, I’m going to say this movie would get an 18% on Rotten Tomatoes.

But an LLM can write bad movie treatments all day long. You could make a thousand of these in the time that it would take me to write one that I thought was good. Laughably, Hollywood went on strike to try to stave off this existential threat to screenwriting. Here’s why this doesn’t concern me: I don’t like consensus art. I find ‘design by committee’ loathsome, and now there is no point in making that sort of thing. Because the demand for inoffensive consensus art is about to collapse, we humans have no choice: we have to take risks. I’m sure the studios would love nothing more than to hit a big red button that says “Generate Profitable Movie” and have it spit out?Despicable Me 7: Minions Do Something Predictably Cute. However, now that that button kind of exists, humans are going to have to make beautiful things in order to stand apart from the sea of banality that LLMs will generate.

We have to take risks. We have to innovate. We have to be daring. There’s no other choice for creatives anymore. I’m excited about this.

There’s no room for consensus thinking in creative contexts anymore. Consensus art is now in limitless supply. It finally found its most relentless manufacturer, and that means the rest of us—the misfits, the troublemakers, the ones who still believe in messy, gloriously imperfect analog humanity—get to create the beautiful stuff. Now it’s time for the oddball ideas to gather steam. Let’s have more of the sort of raw creativity that leaves people baffled, inspired, or even offended. We get to be brave again! Might be the best thing that’s happened to art in a long time. Isn’t it wonderful?