Hollywood Writers Become Mercor AI Trainers as…

A new Wired report by screenwriter Ruth Fowler describes a quiet shift inside the entertainment workforce: Hollywood writers who can't get a room staffed are logging into Mercor and similar platforms to train artificial intelligence for $50 to $150 an hour. Fowler calls it "the new waiting tables" — the side hustle every aspiring creator now seems to be doing between gigs. Two and a half years after the 2023 WGA strike ended, Los Angeles film and TV employment is still down roughly 25%, and many credited writers report going a year or more without a studio paycheck. AI labeling is filling the gap — and reshaping what creative work means in 2026.

What Does the Hollywood-to-AI-Trainer Pivot Actually Look Like?

The work is anonymous, fragmented, and almost entirely remote. According to Fowler's Wired piece, screenwriters log into platforms using opaque usernames, accept short tasks ranging from a few minutes to several hours, and rate AI outputs on tone, accuracy, voice, and "humanity." A typical assignment might ask a writer to rewrite a chatbot's flat dinner-table dialogue so it sounds like a real family argument, or to grade three competing model responses to a prompt like "write a noir opener about a private investigator in Glendale."

Pay arrives via 1099 contractor invoicing. There are no benefits, no WGA minimums, no residuals, and — crucially — no credit. Writers who spent decades building bylines now train models under a string of random characters. Fowler describes working under usernames like "ri611" while assessing whether a chatbot's voice sounds "natural or flat, affected or annoying." That anonymity is part of why this labor market has been invisible to most readers until now, even though Mercor alone disburses more than $1.5 million per day to its contractor base.

The texture of the work matters as much as the rate. Writers describe juggling three or four concurrent platform tabs, switching between rating dialogue, fact-checking a model's claim about Renaissance architecture, and rewriting a chatbot's customer service script — sometimes within a single hour. The cognitive whiplash is significant: a working screenwriter accustomed to spending weeks inside one set of characters now has to context-switch every twelve minutes to clear a daily task queue. Workers Fowler interviewed reported that the projects most prized by Mercor and competing platforms tend to involve emotional nuance, cultural specificity, or comedic timing — exactly the skills that took decades to develop and that no formal credential captures. For credentialed writers, that creates an uneasy bargain: the very expertise that should command a staffed-room salary is instead being purchased in fifteen-minute increments by a venture-backed intermediary they have never spoken to.

Who Is Mercor and What Does It Pay Writers to Do?

Founded by Brendan Foody, Adarsh Hiremath, and Surya Midha — all in their early twenties — Mercor raised a $350 million Series C led by Felicis in October 2025, pushing the company to a $10 billion valuation on a roughly $500 million annual run rate. Its customers reportedly include OpenAI and Anthropic, the two largest foundation-model labs. Mercor's core product is talent supply: it recruits subject-matter experts and routes them to AI labs for Reinforcement Learning from Human Feedback and evaluation work.

For creative writers, public Mercor listings show "Creative Writer — AI Trainer" roles paying $75 to $150 per hour, with "Creative Writing Expert" tasks closer to $50 per hour and senior screenwriting roles reaching $150. Across all professions, Mercor reports more than 30000 contractors at an average billed rate near $95 an hour. CEO Brendan Foody told Bloomberg in April 2026 that "millions" of workers will "spend the next decade teaching machines the judgment, nuance, and taste that only humans possess." The pitch lands harder for someone whose last produced credit aired in 2022.

What makes Mercor different from earlier crowdwork platforms is the deliberate emphasis on credentialed expertise rather than generic clickwork. Earlier data-labeling vendors recruited at scale and paid pennies per task; Mercor positions itself as a marketplace for verified subject-matter experts, screening applicants through interviews, sample tasks, and resume review before assigning model-training projects. That shift means a Mercor worker is not competing against an anonymous global pool — they are competing against other professionals in their specific domain. For Hollywood writers, that has translated into project pipelines specifically tagged for narrative, screen-format, and dialogue work, and pay bands roughly two to three times what equivalent crowdwork paid five years ago. The trade-off is intensity: the same vetting that pushes hourly rates up also pushes the pace, the calibration demands, and the implicit pressure to be available whenever a high-paying assignment surfaces in the queue.

Why Are Entertainment Workers Turning to AI Gig Work?

The short answer is that the work isn't coming back fast enough. Production volume cratered during the 2023 dual strikes and never fully recovered. The 2025 Otis College Report on the Creative Economy found California's film, television, and sound sector still about 25% smaller than in 2022, and the number of shooting days in Los Angeles County was down 42% in 2024 versus that peak. Streaming consolidation hasn't helped: Paramount laid off roughly 2,000 staff in 2025 following its Skydance merger, and the proposed Netflix–Warner Bros. deal could compound the squeeze.

Within that contraction, the WGA's own membership is feeling it. The guild's April 2026 contract protected the health plan and addressed unpaid "free work," but it can't manufacture rooms that studios refuse to greenlight. The result is a class of credited, experienced writers — many of whom built shows for Hulu, the BBC, or premium cable — taking AI gig shifts at 2am because the work is fast, remote, and pays this week. Fowler's piece notes that writers swap project links in Discord servers the way they once swapped agent recommendations.

There is also a geographic story underneath the contraction. Los Angeles, long the gravitational center of scripted television, has seen production migrate to lower-cost incentive zones in Georgia, New Mexico, the United Kingdom, and Canada — and that migration accelerated through 2024 and 2025 as studios chased margin during the post-strike rebuild. Writers who relocated to Los Angeles in the streaming boom now find themselves anchored to a city whose primary employer is shedding work faster than nearby industries can absorb it. Remote AI training fills that vacuum precisely because it is location-agnostic. A screenwriter in Burbank, a former showrunner in Atlanta, and a credited episodic writer in Vancouver are all logging into the same task queue at the same hour, competing for the same project assignment. The result is a quietly national, increasingly anonymous labor market for creative expertise that did not exist in any meaningful form before 2024.

Is the Pay Fair, and What's the Long-Term Career Cost?

On paper, $95 an hour beats most non-writing service work. In practice, the economics are messier. Writers describe unpaid onboarding sessions, vetting tasks that take hours but pay nothing, contract terms that can change mid-project, and sudden project cancellations that vaporize a week's expected income. Slack-based oversight, opaque rating systems, and "calibration" rounds — where workers must align with hidden answer keys — mean a high posted rate can mask a much lower effective hourly take.

There's also the credit problem. A staff writer position builds IMDB credits, union pension contributions, residual streams, and a portfolio that travels. An AI labeling gig builds none of those. Workers report that the more skilled they become at training a model, the closer they get to teaching it the very judgment that makes them hireable elsewhere. Several writers Fowler interviewed described feeling "professionally invisible" after a year of trainer work — present in the AI's output, absent from any human-facing résumé. For entertainment professionals trying to stay in the industry rather than out of it, that trade-off is the heart of the conversation.

Tax and benefit math compounds the credit problem. A 1099 contractor pays both halves of self-employment tax, carries no employer health contribution, accrues no pension service credit, and must self-fund retirement entirely. For a working writer used to WGA-administered health insurance and pension contributions tied to staffed-room work, the headline hourly rate at Mercor can look healthy and still produce a sharply lower net-of-tax, net-of-benefits annual outcome than a typical staff job. Writers also report difficulty smoothing income across the year, because project pipelines surge and vanish without notice; one month of $9,000 in trainer earnings can be followed by a month with zero qualifying tasks. That volatility makes mortgages, child care arrangements, and even basic budgeting harder, and it is one of the quieter reasons working writers describe the shift as financially exhausting rather than financially liberating, even when the posted rate looks competitive.

What Does This Signal About AI's Real Job Impact?

The Hollywood-to-Mercor pipeline complicates the usual "AI is creating new jobs" framing. The new jobs are real, well-paid by service-industry standards, and — most pointedly — being filled disproportionately by the same workers whose primary industry is shrinking. A 2025 TIME profile of Mercor flagged the same dynamic across law, medicine, and consulting: white-collar professionals are being recruited to teach models the tasks those models will eventually perform with less human oversight. In other words, the "AI is creating jobs" story and the "AI is destroying jobs" story can both be true about the same person in the same week.

That's a sharper picture than the macro labor data alone can capture. The latest BLS Employment Situation report still shows a generally healthy headline unemployment number, but it doesn't disaggregate "veteran TV writer earning $95/hour training a chatbot for 20 hours a week" from "veteran TV writer in a staffed room earning a WGA minimum plus benefits." The AI-trainer surge is partly invisible because, statistically, it looks fine. Read more on how the workforce is shifting at Metaintro.

The longer-term policy implication is that traditional labor classifications are not capturing how creative work is actually being performed in 2026. A writer who earns roughly half her income from a partial-season staff job and the other half from anonymous AI training projects shows up in one set of statistics as employed, in another as a contractor, and in none as a person whose primary craft is being absorbed into machine-learning pipelines. For employers, recruiters, and policy researchers, that gap matters: it obscures both the scale of the AI training workforce and the degree to which the most experienced creative professionals are quietly subsidizing model development at rates that may not survive the next funding cycle. For the writers themselves, the most pressing question is whether the pivot is a bridge back into staffed creative work as production recovers, or a one-way ramp into an evaluation economy where the model improves until it no longer needs them. The honest answer in mid-2026 is that nobody knows yet, and the data we need to know is precisely the data being produced under usernames like ri611.

For job seekers watching this shift unfold from outside Los Angeles, the practical takeaway is sober but useful. AI-trainer income can be a real bridge during industry downturns, but it works best as a complement to a primary craft, not a replacement for one, and it rewards workers who keep a public credential pipeline alive alongside the anonymous task queue. Treat the gig as supplementary income, document the skills the platform asks you to demonstrate, and protect time for the staffed work that still builds a durable career — that combination is what most working writers Fowler interviewed described as the only honest path forward in this market.