■ SAFE RISK ■ Creative Arts
No, but the economics were already brutal and AI does nothing to help. Theatre commissions a named human voice; the machine competes for the writing work that surrounds it, not the plays themselves.
“AI writes dialogue. But a play about the human condition needs a human who's survived it.”
Our AI replacement risk score — how we score jobs
Playwriting is a long apprenticeship in rooms. A script goes through drafts, dramaturg notes, readings with actors, a workshop where lines that looked fine on the page collapse aloud, rehearsal rewrites, and previews where an audience's silence in the wrong place sends you back to the laptop overnight. Most playwrights subsidise this with teaching, screenwriting, adaptation work, script reading for theatres, and grants — commissions are few and the fees rarely cover the months involved.
Generative models write competent dialogue, and they are quick at structural tasks: scene outlines, synopses, character bibles, alternative endings, and the grant and funding applications that consume weeks of a writer's year. In the wider industry, AI already threatens the adjacent income streams playwrights rely on — corporate scripts, some screenwriting and content work, script coverage. Translation and surtitling are being automated too.
The plays themselves resist for two reasons. First, craft: a good play is not fluent dialogue but structure under pressure, subtext, and a specific perspective earned by having lived something. Models produce plausible speech with nothing underneath, and it shows in a rehearsal room within an hour. Second, the market: theatres commission writers, not scripts. Programming decisions are about whose voice is in the season, which communities a work speaks to, and whether the writer can sit in rehearsals and rewrite. A play also has to survive collaboration with directors and actors who ask why a character does something — a question that needs an author. Union agreements across the performing arts have moved to restrict generated material as well.
Automatability: our editorial assessment of current and near-term AI capability
The squeeze this decade lands on the side work — coverage, corporate scripts, some screen and content writing — that keeps playwrights solvent between commissions. Original stage commissioning stays human well past 2040, protected by union agreements, programming culture and the need for a writer in the rehearsal room. The profession's danger remains what it has always been: not enough paid commissions.
It can produce something with the shape of one — scenes, entrances, plausible dialogue — and experiments staging generated scripts have been done. What they consistently lack is subtext, structural pressure and a reason to exist, which becomes obvious the moment actors have to play it. Theatres commission a writer's perspective, so a script without an author has nothing to sell.
Only for a small number, and that predates AI. Commissions are scarce and fees rarely match the months of work, so most playwrights combine writing with teaching, dramaturgy, screenwriting or unrelated jobs. The genuine new pressure is that some of those supporting income streams — coverage, corporate and content writing — are being automated, which narrows the financial base.
Chiefly as an assistant on the administrative and preparatory layer: grant applications, synopses, character notes, research, formatting and alternative structural options for a scene. Some writers use it as a sparring partner to argue against. Using it for the dialogue itself tends to be self-defeating, since the recognisable voice is the commissionable asset.
Three things: theatres select writers rather than manuscripts, rewrites happen live in rehearsal with a human author answering questions, and performing-arts unions have moved to restrict generated material in contracts. Add the fact that audiences and funders increasingly value authorship and lived perspective, and the stage is one of the better-defended corners of the writing profession.