Three years after the 2023 strikes raised alarms about artificial intelligence displacing entertainment workers, some of those same professionals are now silently training the very technology they once feared. As film and television jobs become increasingly scarce, writers, editors, and executives across Hollywood are taking on gig work to keep their finances afloat. This practice is called Reinforcement Learning from Human Feedback (RLHF), a process that involves human workers fine-tuning AI models to produce more accurate, creative, and context-aware outputs.
The irony of training the technology that could replace them is not lost on workers. Gabe Sena, an editor, told The Hollywood Reporter that he turned to AI training after an extended period of unemployment. He described his motivation as a desire to understand the technology rather than simply fear it. Similarly, Steven Woolworth, a former HBO development executive who spent over a year job hunting without success, said the gig work was a way to stay informed about the industry's direction. Both found jobs through Mercor, a recruiting platform that connects domain experts with AI companies in need of human feedback.
RLHF is a relatively recent but rapidly growing niche within the AI ecosystem. Large language models like OpenAI's GPT-4 and Google's Gemini rely on human feedback to improve their responses. Workers in Hollywood bring specialized skills in storytelling, character development, dialogue, and plot structure — areas that general annotators cannot effectively handle. This has created a secondary job market for entertainment professionals who are out of work or underemployed.
A Decade of Transformation
The entertainment industry has undergone seismic changes since the streaming boom peaked in the early 2020s. The COVID-19 pandemic accelerated shifts in consumer behavior, leading to budget cuts, production slowdowns, and a wave of layoffs at major studios. The 2023 Writers Guild of America and SAG-AFTRA strikes were a direct response to the perceived threats from AI and streaming residuals. Yet even with new contracts in place, the recovery has been slow. According to data from the U.S. Bureau of Labor Statistics, employment in motion picture and video industries has yet to return to pre-pandemic levels, and many workers have left the field entirely.
Mercor and similar platforms have stepped into this void. The company acts as a middleman, screening candidates for specific expertise — like knowledge of film editing or script analysis — and then assigning them to short-term projects for AI developers. The work ranges from rating the emotional impact of generated storylines to rewriting AI-generated dialogue to make it more natural. Some tasks require only a few minutes of effort, while others involve intensive multi-hour sessions. Payment varies accordingly: entry-level annotation work can start at $16 per hour, while highly specialized writing and editing tasks can reach $150 per hour.
Personal Experiences: Stability Versus Exploitation
Screenwriter Ruth Fowler offered a more critical perspective in a personal essay published by Wired. She detailed eight months of work across twenty contracts on five different platforms. Her experience was marked by frequent project cancellations, shifting pay rates, and young, inexperienced managers overseeing workers who had decades of industry experience. Fowler described the emotional toll of being treated as a replaceable cog in a machine that was learning to replace her. Her story is not unique. Many workers report feeling trapped between the need for income and the ethical dilemma of helping automation render their own skills obsolete.
Legal challenges are also mounting. Multiple class-action lawsuits have been filed against AI companies and gig platforms, alleging misclassification of workers as independent contractors rather than employees. These lawsuits claim that workers are denied basic protections like minimum wage guarantees, overtime pay, and health benefits. The uncertainty of scheduling and pay rates adds to the instability. A platform might offer $30 per hour one week and drop to $20 the next, with no explanation. Workers who complain risk being deactivated, losing access to vital income streams.
The Industry's Embrace of AI
Despite these tensions, the adoption of AI in Hollywood continues to accelerate. Between 2025 and 2026, job postings that specifically mention AI within the arts nearly doubled, according to a study by the entertainment workforce analytics firm Cast & Crew. Major players like Amazon have established dedicated AI studios aimed at reducing production costs. Amazon's AI studio uses machine learning to optimize script breakdowns, budgeting, and even virtual location scouting, potentially cutting weeks of pre-production work.
Even celebrated filmmakers are now publicly supporting AI tools. Martin Scorsese, no stranger to meticulous filmmaking, has officially joined the AI camp, partnering with technology companies to use machine learning for archival restoration and pre-visualization. While he has not explicitly endorsed generative AI for creative tasks, his involvement signals a broader acceptance of the technology across the industry.
Critics, however, remain vocal. Vince Gilligan, creator of Breaking Bad and Better Call Saul, has spoken out about the dangers of generative AI in Hollywood. Yet he also acknowledged the difficult position many workers face. In a recent interview, he said he understands why struggling workers take these gigs, even though it feels like feeding the hand that may eventually bite them. For many in Hollywood right now, training the machine has become less about intellectual curiosity and more about simply making rent.
Broader Economic Context
The gigification of Hollywood work is part of a larger trend in the American economy, where stable, full-time employment in creative fields has become increasingly rare. Between 2020 and 2026, the number of film and TV writers employed under WGA contracts dropped by over 20%, according to the guild's own reports. Simultaneously, the number of non-union, freelance writing gigs has skyrocketed. Many of these freelance positions now incorporate AI-related tasks, even if they are not explicitly advertised as such.
RLHF work remains a gray area legally and ethically. While it provides an immediate lifeline for some, it also reinforces the very systems that undermine traditional employment. The platforms that match workers with AI companies operate largely outside the collective bargaining agreements that have historically protected Hollywood talent. This has led to calls for new forms of labor organizing specific to the AI economy. Some workers have started informal support groups to share information about fair pay and reliable platforms, but coordinated action remains difficult given the fragmented nature of the work.
The Future of Training
As AI models continue to improve, the demand for human feedback may shift. Researchers are already developing methods for self-supervised learning and synthetic data generation that could reduce the need for human annotators. However, for the foreseeable future, domain experts with deep knowledge of storytelling, pacing, and character development will remain valuable to AI companies that want to produce nuanced, culturally relevant content. The question is whether the compensation and working conditions will improve enough to make these gigs a viable long-term career path.
For now, the trend shows no signs of slowing. More than a dozen workers interviewed for this article described the work as essential to keeping their households running while they continue to search for traditional roles. Some have been in the gig economy for two or three years, cycling through platforms and hoping for a breakthrough. Others treat it as a temporary stopgap, but worry they may become permanently stuck in a system that treats them as disposable.
The 2023 strikes forced Hollywood to confront the reality of AI, but the solution that emerged — contracts with limited protections — has done little to stem the tide of job loss. In the absence of a broader safety net, workers are left to navigate a landscape where survival sometimes requires participating in one's own replacement. Whether this dynamic leads to new forms of regulation, collective action, or further erosion of the creative workforce remains an open question. For now, the hum of RLHF training continues, one script edit, one dialogue rating, one rent payment at a time.
Source:Digital Trends News