Why I Still Write Film Criticism in the Age of AI

A room at three in the morning has a particular kind of honesty. Once every sound has settled — the street, the building, the ambient hum of a city that never fully sleeps — you’re left with the things you’ve actually been holding onto. I’ve spent enough of those hours at a desk, revising sentences under the pale wash of a monitor, to know what that confrontation feels like. At some point the question stops being about the sentence and starts being about the life around it. Is any of this worth continuing? The world already has something that writes faster, organizes more clearly, and optimizes more ruthlessly than I ever will. So why is a person still sitting here, hands on a keyboard, arguing with a paragraph at 3 a.m.?

I spent nearly a decade working in music. Not as a listener — as someone embedded in the industry, in the apparatus of it, in the specific heartbreak of building something in a field that rarely rewards the effort proportionally. People talk about failure as though it arrives as an event, a single definable moment when something ends. In my experience, failure is more like weather. It accumulates gradually, seeps into how you move through the world, quietly rewires the way you think. There were days when music felt like the organizing principle of my entire life, and days when that same world looked like a ruin I was picking through for pieces worth saving.

Somewhere in the middle of all that, I became interested in psychology — not as a therapeutic project but as a genuine intellectual question. Why do people repeat the same wounds? Why do certain emotions refuse to dissolve? I ran a psychology blog for a while, reading and writing across those questions, and somewhere alongside it I started writing about film and music more seriously. It began as something close to a hobby. But over time those pieces became the ones that stayed with me the longest — the writing that felt most like it was genuinely mine, not performed for anyone, not calibrated to an audience I was imagining. They were the sentences I came back to.

Then the AI moment arrived.

People still speak about artificial intelligence as a technological development, a productivity tool, a convenience. I don’t think that framing is adequate anymore. What’s happening is closer to what the Industrial Revolution did to physical labor — except this time, the work being transformed is cognitive and creative. The factory didn’t just replace arms and legs; it reorganized civilization. Large language models are beginning to do something analogous to the written word.

The standard reassurance goes: art, at least, will always belong to humans. I’ve started to doubt that sentence. After AlphaGo, professional Go players didn’t return to the game they had known before — the reference point shifted permanently, the range of what was considered possible expanded beyond what human intuition alone could reach. Something similar, I think, is already happening in writing. The change is not incremental. The gap is wider than most people are willing to admit.

What AI does now is not simply “help” with writing. In the domains of information synthesis, logical structure, SEO optimization, readability, keyword strategy, and architectural coherence, it has already moved to a level that most individual writers cannot consistently match. And more unsettling than its capability is its accessibility. A twelve-year-old with a single well-formed prompt can produce something that, five years ago, would have resembled the output of a polished professional publication. Across technology, health, self-improvement, general knowledge — the majority of informational content being produced on the internet is already, in some meaningful sense, AI-assisted or AI-generated. Most people aren’t ready to say this out loud. But it’s true. And in those domains, that content is frequently more accurate, more clearly written, and more search-friendly than what a single human writer produces in the same time.

The logical response might be despair. Writing is finished. Human authors are obsolete.

I’ve arrived, stubbornly, at the opposite conclusion. Over the next decade at least, writing will become more important — not less. But what it means to write will change completely.

If writing once meant the transmission of information, it increasingly means something closer to the design of thought. The relevant question is no longer how much do you know but how precisely can you translate your thinking into language. In a world where the production of content has been largely automated, the human role shifts to something like direction — the capacity to articulate, with sufficient granularity and intention, exactly what you want. Think of Renaissance painters who ran workshops full of skilled apprentices executing the detailed work: the master’s job was to hold the vision and communicate it clearly enough that others could realize it. That relationship is being reconstituted, except now the apprentice is a language model, and the quality of what it produces depends entirely on the clarity of what you hand it.

I find myself genuinely interested whenever I hear someone complain that their AI “keeps getting it wrong.” In almost every case, the limitation isn’t the model — it’s the language surrounding the request. Given the same tool, some people produce work that surprises them; others get back something flat and generic. The variable is almost always the quality of thinking brought to the prompt: how deeply someone has examined the problem, how precisely they’ve mapped its contours, how well they can differentiate between nuances that sound similar but point in different directions. The irony of the AI age is that the core competitive skill turns out to be something irreducibly human: the capacity for careful, unhurried, articulate thought.

This is what eventually made the direction of my own writing feel not just plausible but, in a strange way, necessary. I moved toward film criticism — and toward cultural criticism more broadly — not because I was optimistic about the field’s commercial prospects, but because it seemed like one of the last places where human writing still had something to prove.

Watch a film and stop for a moment on the question of what happens afterward. Not what you thought of it, in the summary sense, but what it did to you. Which specific shot made you hold your breath without deciding to. Which expression on which face you kept returning to after the lights came up. Why a particular sequence unlocked something in your memory that you hadn’t visited in years. These are not data points that can be easily harvested. Two people watching the same film in the same theater will have experiences that diverge in ways that no shared review could bridge. One person reads a long static shot as loneliness; another, in that identical frame, feels something like liberation. Art, at least for now, remains irreducible to consensus. It doesn’t collapse into a correct answer.

I don’t expect this to be permanent. There will likely come a point — perhaps five years from now, perhaps ten — when AI systems can track the real-time experience of watching a film, correlating visual attention with emotional response and memory architecture, learning what images do to specific nervous systems under specific conditions. The possibility of a machine that doesn’t just analyze films but genuinely experiences them the way a person does is not science fiction. It’s an engineering problem on a timeline.

But until that point, the human encounter with a work of art retains a value that isn’t fully legible to the tools we’ve built so far. And in that gap — which is narrowing, but has not closed — there remains something worth writing about with the full weight of a specific life behind it.

This is also why I think reading and writing remain survival skills in a way that has become more literal than metaphorical. I’m not dismissive of audiobooks or video essays — these are legitimate modes of engaging with ideas. But they don’t fully substitute for what happens when you read slowly and write seriously. Reading a text carefully means entering someone else’s cognitive structure, following the architecture of another mind as it builds its argument or image. Doing that repeatedly and over time expands your own linguistic resources — not in a decorative sense, but functionally. And writing is the process of reorganizing that expanded world into a structure that is distinctly yours. Together, they build something like a muscle for thought.

The danger of the current moment is that every ambient signal is telling people this muscle is unnecessary. Short-form video compresses cognition into bursts of stimulation. Algorithms remove the need for deliberate exploration. People consume more information than ever while doing less of what might actually be called thinking. The more this happens, the more thoroughly someone is transformed from a person who produces meaning into a person who consumes it — moving from interpretation to passive reception, from a maker of sense to a reader of whatever has already been made.

That, more than AI itself, is what frightens me. Not the technology. The abdication.

What will distinguish writers — and perhaps people more broadly — over the next decade will not be their access to better tools, since everyone has access to the same tools. It will be how deeply they can read, how long they can think, how precisely they can catch their own experience in language before it dissolves. The person who survives in this landscape is probably not the one with the most technical fluency. It’s the one with the clearest language — by which I mean the most honest, most specific, most carefully won relationship to the words they use and why they use them.

A sentence, arrived at slowly, at a desk, at three in the morning — that might be the last traceable sign that a human being was here, doing something that couldn’t be outsourced.

And I’m not ready to stop leaving that trace.