Key Takeaways
- The transition in Hollywood from human to AI writers is rapid and often exploitative.
- AI training work is precarious and unpaid for many of those employed in this sector.
- Labor laws are insufficient or non-existent, allowing companies like Mercor to exploit workers without legal repercussions.
In 2023, Hollywood went on strike partly due to fears over AI replacing writers and actors. However, the industry did not benefit from this; instead, it became a breeding ground for insecure, underpaid work opportunities. One of these was AI training, where workers like the author were hired as “taskers,” often with vague or nonexistent projects that would end abruptly.
The author began their journey in AI training after seeing an ad on Facebook for writers willing to train large language models. They were promised a steady income and flexible hours, but quickly discovered these promises were empty. The work was sporadic, unpaid, and often ended without any tasks assigned. This led the author to seek more stable opportunities.
One such opportunity came in September 2025 when they were hired as a “generalist data annotator” at $52 per hour. However, this job was also short-lived, ending abruptly after only two weeks without any explanation or compensation for the time spent on tests and background checks.
Several months later, the author was offered an “expert” role paying $70 per hour. This position promised more responsibility but quickly devolved into a similar pattern of sporadic work and abrupt terminations. The projects were often given nonsensical names like “Project Dead Language,” which only served to highlight the lack of genuine tasks.
Throughout these experiences, the author encountered various team leaders who seemed to have significant power over their employment status. These individuals would frequently demand immediate action or risk the worker being “off-boarded,” a euphemism for termination without notice.
The work environment was characterized by constant pressure and urgency, often involving tasks that were never completed due to technical issues or lack of available projects. The author described these experiences as akin to a “Hunger Games” scenario—workers competing against each other for the few remaining tasks before they disappeared entirely.
Despite their initial optimism about finding stable work in AI training, the author quickly realized that this sector was rife with exploitation and insecurity. The lack of clear labor protections allowed companies like Mercor to hire workers without proper contracts or benefits, leading to a cycle of sporadic work and sudden terminations.
Ultimately, the author concluded that while AI training provided some income during lean periods, it was not a sustainable long-term solution. The industry’s reliance on exploitative practices left many workers in precarious situations, often without any legal recourse or support from labor laws designed to protect workers.
Originally published at wired.com. Curated by AI Maestro.
Stay ahead of AI. Get the most important stories delivered to your inbox — no spam, no noise.