I Work in Hollywood. Everyone Who Used to Make TV Is Now Secretly Training AI

Editor's take

Hollywood creatives are finding themselves performing data labeling and annotation tasks for AI companies, a trend illustrating the growing demand for human input in refining large language models. This development highlights a subtle but significant shift in the labor market, where creative professionals, accustomed to intellectual property and artistic creation, are now engaged in the often tedious, repetitive work of training AI systems like those powering generative text and image tools. The availability of such "AI gig work" offers a temporary income stream for those facing industry downturns, such as the recent WGA and SAG-AFTRA strikes, but raises questions about the long-term sustainability and ethical implications for creative careers.

The prevalence of these contracts on platforms like Appen, Lionbridge, and specialized AI training services suggests a burgeoning, albeit low-profile, sector. It underscores the dependency of advanced AI, including models from OpenAI and Google, on vast amounts of human-curated data. What remains to be seen is whether this trend will lead to greater integration of AI tools *within* the creative process, as opposed to simply using creatives to build the tools themselves. The critical question is whether these data labeling roles will evolve into more collaborative or supervisory positions, or if they represent a permanent underclass of human labor supporting an automated future.