Seth Rogen is right about the part that matters most.
Writing is the process. The wandering, resisting, testing, throwing away, circling back, discovering-what-you-meant process is not some inconvenience stapled onto the side of the work. It is where the work becomes itself. Any tool that promises to remove that process deserves suspicion.
That is why his comments about AI in filmmaking landed with such force. In the clip shared by Variety, and reported by TheWrap from his Cannes appearance promoting Tangles, Rogen pushed back hard against the fantasy of AI as a way around writing. His sharpest line was also the most revealing:
“The idea of a tool that makes me write less is not appealing to me.”
Seth Rogen, via TheWrap
There is a real standard inside that sentence. Rogen is not defending typing as a sacred physical act. He is defending contact with the material. He is defending the part of writing where taste develops under pressure, where bad ideas expose better ones, where jokes become scenes, scenes become arguments, and the author slowly becomes accountable for what the work is trying to say.
The mistake is assuming that serious creative use of AI exists to avoid that contact.
Anyone who uses AI creatively for more than five minutes knows the actual experience is almost the opposite. The lazy use case is real, of course. You can ask for generic pages, accept the first pass, and call the output writing. That deserves every ounce of contempt it gets. But that is not the interesting version of the tool, and it is not the version serious writers, designers, filmmakers, or developers recognize from daily practice.
The serious use case does not make you write less. It makes you confront more.
AI opens avenues. That is the gift and the problem. You can test ten versions of a scene premise, three different Main Character framings, four possible Relationship Story pressures, two tonal registers, and a half-dozen ways of sequencing a reveal before lunch. None of those possibilities absolves you from authorship. They demand more of it, because every option requires judgment.
The writer has to decide which path belongs to the Storyform and which is merely attractive. The writer has to recognize when a line is fluent but emotionally false, when a scene is vivid but structurally irrelevant, when a generated turn gives the appearance of conflict while quietly dissolving the actual Throughline. More options do not reduce the burden of meaning. They multiply the number of moments where meaning has to be defended.
This is the part the shortcut argument keeps missing.
People often talk about AI as if it were a vending machine for finished pages. Type in the desire, receive the artifact, skip the agony. That does happen, and it produces exactly the kind of hollow work critics are reacting against. But in a real creative workflow, the value of AI is not that it hands you the answer. The value is that it lets you stay longer inside the question.
That is a different kind of labor.
It is the labor of exploration. The labor of comparison. The labor of asking why one version feels dead and another suddenly reveals the center of the piece. It is closer to having a wall covered in index cards than having a ghostwriter in the room. The machine can help populate the wall. The author still has to know what the wall means.
Anyone using Codex to write code understands this immediately.
From the outside, it can look like software development has been compressed. A prompt goes in, code comes out, and people imagine the developer has skipped the work. From the inside, the work often expands. You ask for an implementation, then read the diff, trace the architecture, reject the wrong abstraction, adjust the test, inspect the failure, tighten the boundary, check the route, verify the browser, and discover three related problems the first request exposed.
You are not doing less engineering. You are doing more engineering at a higher level of leverage.
Codex can generate a patch, but it cannot be responsible for the system. It does not know the product promise, the hidden production constraint, the user expectation, the thing that broke last month, or the reason a seemingly cleaner abstraction would violate the grain of the codebase. The human developer still has to carry the continuity of meaning across the work. The tool increases reach. It does not inherit responsibility.
Creative writing works the same way.
An AI assistant can suggest a scene, but it cannot know what the scene costs the story unless the writer gives it a structure of accountability. It can propose language, but it cannot feel when the language has drifted from the emotional truth of the Main Character. It can outline possibilities, but it cannot decide which possibility best expresses the inequity at the heart of the Objective Story. It can generate, but it cannot mean.
That is where Dramatica becomes more important, not less.
Dramatica gives writers a way to keep exploration from turning into drift. A Storyform is not a shortcut around authorship. It is a model for protecting authorship while the creative field expands. When AI offers ten possible directions, Dramatica helps the writer ask which direction preserves the argument, which one strengthens the Throughline pressure, which one exposes the actual source of conflict, and which one only sounds good because fluent language is very good at pretending to be insight.
Without that kind of structure, AI can become a slot machine. Pull the lever, receive a scene-shaped object, mistake novelty for progress. With structure, the same tool becomes a way of testing intention against possibility. It gives the writer more ways to discover what does not belong, which is often how the real work finally comes into focus.
So yes, reject the fantasy of AI as a process-free substitute for writing.
Reject the pitch that creativity is just having an idea and letting software do the rest. Reject the flood of image slop, scene slop, joke slop, and story-shaped noise pretending that output volume is the same thing as art. Rogen’s contempt is pointed at something real, and writers should not be embarrassed to share it.
But the serious response cannot stop at contempt.
The future of creative work will not be protected by pretending these tools only serve people who hate the process. Plenty of serious creators are using them precisely because they love the process enough to want more of it: more angles, more attempts, more pressure tests, more ways to see the thing before deciding what the thing is. They are not trying to write less. They are trying to explore more deeply before they commit.
The line worth defending is not between AI and no AI. That line is already too crude to carry the weight of the moment. The meaningful line runs between replacement and augmentation, between outsourcing judgment and expanding inquiry, between letting the machine impersonate authorship and using the machine to make authorship more deliberate.
Rogen is right that a tool designed to make a writer care less about writing is artistically poisonous.
The tools worth building, and the practices worth defending, move in the opposite direction. They make the writer answer more questions. They surface more contradictions. They expose more weak choices. They widen the field until judgment becomes even more necessary than it was before.
That is not a shortcut.
That is more writing.