Experienced and award-winning screenwriters, directors and producers are taking gig work teaching AI models film-industry skills — how to build a shooting schedule, how to assemble a pitch deck — because entertainment work has dried up. The reporting, published at 06:00 UTC on 22 August, names the intermediaries: Mercor, Micro1 and Handshake, which hold contracts with AI companies including Anthropic and OpenAI. Pay runs from $12 to $200 an hour. One live listing sought producers with five or more years of credited experience at up to $85 an hour. An anonymous documentary director described being "essentially handed a shovel and asked to dig the grave of my profession."
The vendor layer is the part nobody covers
Between the labs and the workers sits a contracting industry that is where the money actually changes hands, and it is almost absent from coverage of AI training data. These are not scraped corpora. They are hourly contracts with named professionals, sourced deliberately for domain expertise the open web does not contain.
What the common framing gets wrong
The two big numbers do not measure what a headline will say they measure. The 28% fall in US motion picture and sound recording employment — 450,000 in July 2022 to 326,000 in May 2026 — runs from the post-pandemic peak. It spans the 2023 writers' and actors' strikes and the streaming spending pullback, and neither the statistical agency nor the article attributes any of it to AI. AI enters as the thing filling the vacuum, not the thing that created it. Likewise the projection that 10-15% of US jobs could be eliminated is a consultancy forecast, not an observation. And the $12-$200 band is a range across wildly different task types quoted by a handful of interviewees — this is a qualitative feature with named sources, not a survey, and it carries no denominator for how many creatives are doing this work.
What is measurable
Los Angeles shoot days fell 48% between 2021 and 2025 on the local film office's own figures. Netflix has said it used AI in 300 of its 1,000 titles in 2026. Named on-record participants include Ruth Fowler, creator of a BBC One series and co-writer of a Paramount+ title, and Jody Wheeler, 56, a screenwriter and former USC instructor.
Why the framing matters to the argument
Treating the employment collapse as AI-caused makes the story simpler and weaker. The accurate version is harder and more damning: an industry contracted for reasons largely unrelated to AI, and the resulting pool of unemployed expertise became cheap enough to buy by the hour and encode.