A Hollywood showrunner details 20 AI training contracts across five platforms (Mercor, Outlier, Turing, Micro1) in eight months, finding the gig economy chaotic and exploitative.
Key Takeaways
Pay ranges from $52/hr (generalist annotator) to $150/hr (expert creative), but erratic task availability means actual earnings collapse far below advertised rates.
Projects launch with no notice, end without warning, and tasks are finite – fast workers claim all available work, leaving others with near-zero income.
Onboarding is heavy: background checks, multiple apps, unpaid test hours, AI recruiter interviews – all before earning a dollar.
Red-teaming tasks include generating harmful content (bomb recipes, CSAM) under safety-testing contracts, with strict NDAs and anonymized worker IDs.
Team leaders frame the arrangement explicitly as “tasks” not jobs, insulating platforms from labor law obligations while demanding immediate availability.
Hacker News Comment Review
Commenters note the article title is misleading: this is about displaced writers using AI data-labeling gig work for income, not about AI being built for Hollywood production.
Some commenters connect Hollywood’s production exodus to Canada and abroad with the writer surplus driving people toward AI gig platforms, though causality is debated.
One commenter with direct experience confirmed the chaotic conditions described, validating the piece as representative rather than exceptional.
Notable Comments
@kylecazar: clarifies the article is about gig AI training as income replacement, not AI tools disrupting cinema production directly.
