David Perell on AI's impact on writing: why utility writing is already gone and what human writers must become
Apr 4, 2025 with David Perell
Key Points
- David Perell shut down his writing course Write of Passage in November after concluding AI will eliminate utility writing by 2026-2027, forcing human writers to compete on voice and distinctiveness alone.
- Managers are adopting AI faster than frontline workers because the technology mirrors their existing workflow of delegation, feedback, and iteration.
- Perell frames the shift through Marshall McLuhan's observation that displaced technologies become art, positioning writing's future as an artistic rather than utilitarian practice.
Summary
David Perell, a writing teacher who has trained several thousand students over six years, argues that AI has already made utility writing obsolete — and that human writers who survive will need to become something closer to artists.
The inflection point for Perell came in November, when an employee sent him what he called the best memo he'd ever seen from that person. The employee had written up dinner notes and asked AI to summarize them. Perell shut down his writing course, Write of Passage, on November 11th, then spent December in Argentina running 50 to 70 AI prompts a day. His conclusion: by the end of 2026 or 2027, any writing driven by pure utility — business memos, standard legal emails, routine professional communication — will be written by AI as a matter of course.
The split in writing culture
A visible fault line is already forming. Tech-forward writers are quietly building custom prompts and workflows to multiply their output, some claiming 3–4x productivity gains, while a growing purist contingent treats AI-assisted writing as taboo. Perell received his first rejection that week from someone who objected to his promotion of AI tools. He expects the divide to widen.
For writers who want to stay relevant, his answer is voice. As AI floods public channels with competent, generic prose, distinctiveness becomes the only durable signal that a human wrote something. Perell is investing in what he calls "rizzand vibe" — sharpening the energy that is uniquely his, down to the texture of a text message.
The managerial class advantage
One pattern Perell flags that runs counter to how technology usually diffuses: managers have adopted AI faster than frontline workers. His explanation is that working with an LLM mirrors the manager's loop — set a vision, delegate, expect the first draft to be inadequate, give feedback, iterate, ship. Frontline workers have no prior mental model for that. The corollary is that managers who've always said their hardest problems are about people, not work, now see AI as a way to remove that layer entirely.
Writing becomes art
The historical frame Perell reaches for is Marshall McLuhan's observation that when a new technology accelerates at scale, the thing it displaces doesn't disappear — it becomes art. Film photography is art now. Hand craftsmanship is art now. Perell thinks writing is heading the same way: utilitarian output goes to AI, and what remains for humans is the artistic register. He draws a parallel to 14th-century medieval painting, which looks flat and strange by modern standards, and the Renaissance art that followed once architects like Leon Battista Alberti applied the camera obscura and perspective geometry. The technology didn't kill painting — it unlocked a Renaissance. Perell believes the same is coming for writing.
Humor as the last frontier
LLMs can get the structure of a joke right but consistently miss the delivery. Perell thinks that gap is temporary. The more interesting near-term capability, in his view, is hyper-niche humor — feeding an AI 45 minutes of shared context from a weekend trip and asking it for jokes. He argues it would outperform a general-purpose comedian precisely because specificity is where current models already perform well.
The deeper question
Perell closes on the problem he says he can't resolve: when a simulation of humor, consciousness, and care becomes indistinguishable from the real thing, is it the real thing? He doesn't offer an answer, but frames the cultural counter-pressure clearly. The appetite for authenticity — the reason people travel for authentic food rather than a local recreation — is deeply human and will reassert itself. His read is that the late 2010s represented peak inauthenticity, a period when cancel culture made self-expression dangerous and forgiveness culturally unavailable. AI flooding public channels with polished, risk-free prose may paradoxically accelerate a correction toward realness, rawness, and the willingness to forgive people for being themselves.