A friend of mine spent three months writing a short film. She had a script she was proud of, a small budget, and a cast of actors she trusted. What she did not have was the certainty, nagging and persistent, that a sufficiently clever prompt could have produced a similar draft in forty-five seconds.

She finished the film. It is very good. But the experience had changed something for her — not in the work itself, but in the way the work felt while she was doing it. "I kept asking myself," she told me, "what's the point of the struggle if the struggle is optional?"

This is the central psychological crisis of creative work in 2026: not that machines are taking our jobs (though they are taking some), but that they have made the suffering optional. And suffering — or at least, difficulty — has always been part of how we justified the work to ourselves.

The Old Bargain

There was an implicit bargain in the pre-automation era of creative work. You put in time. You developed taste through failure. You learned to distinguish your good ideas from your bad ones through the grueling process of executing both and living with the results. The difficulty was not incidental; it was pedagogical. You learned by doing, and doing was hard.

That bargain is now renegotiable. If you want to write a novel, you can write every scene yourself, or you can draft with AI assistance, or you can use AI to generate a complete first draft that you then revise. All three paths can produce good books. Only one of them involves the particular kind of learning that comes from wrestling with a blank page at two in the morning.

Whether that learning matters depends on what you think creative work is for.

"The tool does not care what you make. That is its power and its poverty."

If creative work is primarily about the artifact — the film, the novel, the painting — then any tool that helps you produce a better artifact is unambiguously good. The audience does not experience your process; they experience the work. And if AI assistance produces better work, then using it is simply rational.

But most people who make things seriously do not believe that the artifact is the whole point. They believe something happens in the making — something that changes the maker, that clarifies what they think, that develops capacities that cannot be shortcut. The artifact is evidence of a process; the process is the thing.

The Paradox of Infinite Possibility

Here is something that sounds like good news but feels like bad news: we are living in the most creatively abundant moment in human history. The barriers to making and distributing creative work have never been lower. A teenager with a phone can produce music, film, and writing that would have required a studio, a camera crew, and a publishing contract twenty years ago.

And yet creative work feels, to many people, more anxious and uncertain than before. The problem is not access to tools. The problem is the elimination of constraint.

Constraint is generative. When you can only do a few things, you get very good at doing those things. A poet working in a strict form — a sonnet, a villanelle — is not limited by the form; the form is the engine. The resistance creates the shape. Remove all resistance, and you do not get better poems; you get an infinite, undifferentiated stream of possibility that never coalesces into anything in particular.

The most interesting creative response to AI I have seen is not to use it as a generation engine but to use it as a constraint machine. Give it strict parameters — this palette, this structure, these words and no others — and then work within what it produces. The AI does not replace the creative act; it sets the stage for one. The constraint is artificial but no less real for that.

What Cannot Be Automated

There is a category of creative work that AI cannot do, not for technical reasons but for conceptual ones. It cannot make work that is about the experience of making it. It cannot document a life from the inside. It cannot produce the particular texture of attention that comes from spending a decade thinking about a single problem. It cannot be changed by what it makes.

This is not nothing. In fact, it is almost everything that we care about when we care about art at its most serious. We read certain writers not because they give us information but because they give us the experience of a particular consciousness moving through the world. We cannot get that from a system that has no consciousness, no world, and no movement.

The question is whether that distinction will matter commercially, culturally, or both. There are reasons to think it will, and reasons to think it will not. The history of photography suggests that a new medium does not eliminate an old one; it redefines it. Painting did not die when photography arrived. It became more painterly. It leaned into what only painting could do.

Creative work in the age of AI may follow a similar path. The work that survives will be the work that is irreducibly human: not because it was made without tools, but because the humanity is the point, and no tool can simulate a point.

Back to the Film

My friend's film screened last month at a small festival. After the screening, someone in the audience asked her about her process. She described the three months of work — the rewrites, the failed shots, the argument with her cinematographer that turned out to be right, the scene she cut that she still thinks about.

Nobody asked whether she had used AI. The question did not occur to anyone, because the work was so clearly the product of a particular person's particular obsessions, pursued with a particular stubbornness over a particular span of time. The struggle was visible in the work. That visibility was not a flaw.

It was, I think, the whole point.