How tech companies are using poets, novelists, playwrights, and TV writers to engineer the personalities of AI virtual assistants
SAN FRANCISCO — Until recently, Robyn Ewing was a writer in Hollywood, developing TV scripts and pitching pilots to film studios.
Context & Ripple Effects
In 2016, the Washington Post profiled Robyn Ewing's jump from pitching TV pilots to scripting personalities for AI virtual assistants — an early signal that Silicon Valley saw voice and character as an engineering problem worth hiring Hollywood writers to solve. At the time it read as a curiosity at the edge of two industries that were otherwise on a collision course over AI's role in content.
The years since turned that curiosity into a labor market: data-labeling firms like Scale AI and Appen now recruit writers and poets specifically for their humanities backgrounds, and writers describe making up pretend chatbot answers to train models how to write. The through-line from Ewing's persona work to today's industrialized writing-for-models pipeline is what makes this piece worth rereading.
First-order effects
- TV and film writers like Ewing gain a new, well-funded employer outside Hollywood, redirecting narrative-craft talent toward assistant personality design at tech companies.
- Tech companies building virtual assistants turn character voice into a staffing line item, competing directly with studios for the same pool of working writers.
Second-order effects
- Hollywood's faction that opposes all uses of AI now faces an awkward split: colleagues taking tech paychecks to shape assistants while the industry fights over how AI may be used in production.
- Once personality-writing proved valuable, it scaled beyond in-house teams — firms like Scale AI and Appen built hiring pipelines around humanities graduates, commoditizing what began as bespoke persona work.
Third-order effects
- If the pattern holds, 'writer' bifurcates into two professions: creators of original work, and model-shapers who fabricate hypothetical dialogue to train systems — with the latter employed by labeling vendors rather than studios.
- The relationship runs both directions over time: some authors have come to treat chatbots as a modern version of the writer's muse, meaning the people who engineered AI voices end up negotiating creative identity with the tools they helped build.
The trend: Writing for machines has evolved from a niche 2016 experiment in assistant personality design into a standing labor market where data-labeling firms and tech companies employ writers to shape both what AI says and how it says it.