In 2023, concerns about the rise of generative AI sparked a writer’s and actor’s strike, and many ordinary workers feared they could lose their jobs across a wide swath of the entertainment industry. Three years later, these concerns are still very much alive and well, and some Hollywood employees are enthusiastic about AI training and are helping to improve the technology anyway.
As Hollywood adapts to technology, with certain areas of the business running toward it and others moving away, a small number of creators are beginning to spend time and go public in the world of reinforcement learning from human feedback (RLHF). In May, writer Ruth Fowler (small disaster, game rules) published a personal essay wired He talked about his experience working in AI training. It was this field he turned to when entertainment jobs dried up and he needed money to pay rent and buy groceries. That same month, screenwriter Robin Palmer, who has written TV movies for Disney Channel and Hallmark, spoke to CBS News about his work in AI training, although he acknowledged that some in the field might compare it to crossing a picket line.
Editor Gabe Sena is also an entertainment worker with a side job fine-tuning AI models. “I’m a mid-career guy and don’t want to be a dinosaur in my field,” he explains. hollywood reporter. “This is something that people are afraid of, and it seems like a black box to a lot of people who aren’t in the tech world. So it made more sense to me to try to immerse myself in it, rather than just thinking, ‘I don’t like it because it’s new.'”
As the traditional film and television job market shrinks, this type of gig work is on the rise, with current and former entertainment workers participating. This phenomenon raises uncomfortable questions in the creative community, where the use of technology can be a third-rail topic, as exemplified by the recent story of animator Jorge Gutierrez dropping out of a generative AI series he was planning to create for Amazon following backlash. Are these workers actively contributing to the ultimate job loss in their industry, or are they simply trying to survive in a system where AI adoption is being driven full-throttle by forces far greater than the individual?
For Senna, the decision was rooted in curiosity about the future and how to prepare. A graduate of the University of California, Los Angeles Film School, Sena typically edits small documentaries and videos for nonprofit organizations such as the Make-A-Wish Foundation and the Children’s Defense Fund. He says he wanted to learn more about AI in the summer of 2025, when he wasn’t working. “It was very genuine curiosity, but I wasn’t even 100% clear exactly what kind of work I would be doing,” he says. He signed up for a job facilitated through a recruitment platform for AI jobs called Mercor.
Melkor was founded in 2023 by three college dropouts and Thiel Fellows, participants in billionaire Peter Thiel’s program that provides funding and connections to people who drop out of college or don’t attend college at all and want to “build something new.” Backed by venture capital firms Benchmark, General Catalyst, Robinhood Ventures, Felicis, and Menlo Ventures, Mercor’s signpost provides subject matter experts to “organize human intelligence to power the AI economy,” or improve AI models. A $350 million Series C funding round in 2025 valued the company at $10 billion.
Though he can’t say exactly what he’s been working on, like many workers in the field, Sena has signed non-disclosure agreements — one of his recent tasks was to compare prompts given to a generative AI system with its output, to see how the two match up.
Asked if he was concerned about jobs being replaced by these types of jobs in the future, Sena said that while some roles may be “phased out,” he believes talented, hard-working and adaptable workers will be able to maintain their careers. “There are parts of my job that I’m willing to leave in very intelligent ways,” he added. “But there’s a lot of subjectivity in what I do, even if it seems like a rote job.”
Like Sena, Steven Woolworth, former head of development at HBO and Studio TF1 America, felt the need to dive headfirst into AI instead of shying away from it. After a year and a half of grueling and fruitless job searches in Hollywood, Woolworth came into this world when a friend who was a member of the Writers Guild of America forwarded an email from Melkor.
Woolworths, who says he is a supporter of AI regulation and guardrails in entertainment, said: “I saw this as one of two ways: I could bury my head in the sand, or I could go into this world and have AI You get a very inside view of what’s going on with training and how it works, and it’s also obviously a very small part of being part of a team of people who train this to be productive in a healthy, positive way.
The job “kept a roof over my head for the past year. [of] “I’m deeply grateful for this,” he says. He also says technology can’t replace in-person events like film festivals and plays to discover new talent, and it can’t replace his previous profession as a manager.
Left: Editor Gabe Sena works in the entertainment industry and helps fine-tune AI models on the side. Right: Steven Woolworth, former head of development at HBO and Studio TF1 America, found a job in AI training at Mercor.
**
RLHF operates as a three-part process. First, a human scores the model’s output. For example, assess how likely characters in a drama are to respond to tragic news with jokes or expressions of sympathy. Scores are given for various responses one after the other. Once the dataset is fully trained based on these responses, the AI itself trains a second “reward model” and uses it to train the original AI, removing humans from the process.
The actual work can be daunting for human enhancers. The machines make choices that range from the rational to the laughable, and humans tell them what was right and what was wrong. One veteran author who helped train the model described an environment similar to a high school standardized test. There’s a strict human director in the room telling people what they can and can’t do, with output flashing on the screen one after another. What at first felt like fun, like guiding a cute, wayward child, soon felt like a crazy psychological experiment.
“How many times can you tell a machine that it’s wrong without losing your mind?” the person recalled.
In recent years, AI companies have reportedly moved away from training models with the help of data annotators doing menial jobs in developing countries, in some cases allegedly experiencing exploitative and harmful working conditions. Companies are now looking for “experts” in their fields, such as doctors and lawyers, in addition to Hollywood experts. Depending on your expertise, you can earn upwards of $100 an hour doing tasks like helping AI tools sound more human or correcting mistakes in your field.
Unlike traditional Hollywood, this is a growing field. According to data provided by job platform Indeed, the proportion of AI-related jobs in the arts sector doubled from nearly 5% to nearly 11% between May 2025 and April 2026. This growth rate also outpaced broader AI job postings, which increased from 2.8 percent in May 2025 to 5.5 percent by April 2026. At the time of writing, “creative writers” can earn “up to $44 an hour” by helping with AI projects promoted through job platform Handshake. “Music professionals” with a master’s degree or higher can earn “up to $100 an hour.”
As AI training grows in the United States, labor complaints have dogged the field. In published news reports, workers describe highly precarious employment, harsh conditions where demand for work exceeds supply, rush work, and a lack of information about exactly who the customers are. Surge AI and Scale AI, two major companies providing RLHF jobs, are each accused in the lawsuit of unpaid wages, misclassifying workers, and exposing workers to traumatic material.
Even Hollywood’s critics of the generative AI economy have expressed sympathy for those working in this volatile field. One of them was breaking bad and pluribus Creator Vince Gilligan. “I’ve been very fortunate throughout my career and I don’t mean to criticize people who are just trying to support their families,” he wrote. hollywood reporter By email. “I just thank God I never faced that choice myself. And I’m also thankful that my career began in 1990 and not 2026.”
Added storyboard artist Sam Tung, who worked on citadel and twisters“People who are doing this work are not doing it because they want to disrespect other human workers. People are really struggling in the current recession, and they have mortgages to pay and children to feed.”
But Tim Friedlander, a voice actor who heads the National Association of Voice Actors (NAVA), warns that the work could provide short-term benefits with long-term consequences. While AI training opportunities are currently proliferating on go-to job platforms for peers like Voices dot com and Voice123, Friedlander said the organization’s recent survey found that about 20% of respondents were intentionally losing their jobs because of jobs performed by AI tools. NAVA’s current position is that there should be no licensing or training agreements in voiceover.
“I wouldn’t necessarily blame someone if all of a sudden they were given a potential $1,200 for four hours’ worth of work, which would be very tempting, but those four hours might be the only time they work on that job,” he added. “In the long run, I think training these systems is more detrimental to the creative sector as a whole.”
The Writers Guild of America is in a difficult position because its members are taking jobs in AI training, even though its leadership opposes the same AI systems.
**
The guild is in a difficult position. The illusion of members cooperating in training a system that the leaders oppose would appear to present an easy path for the group. That is, to issue a directive (if not an outright prohibition) against taking on the task. But at least two writers, both active in WGA politics, said they believed WGA leaders’ hands were tied with the RLHF, and that their group and others may, in fact, be stuck in a kind of Catch-22. They can’t tell their members not to use their skills to make money, especially at a time when so many people are struggling to find regular work. Of course, if they don’t stop it, people could face even more difficulties and they’re really going to need to get these gigs.
When asked if they have a policy for members working on AI training, major Hollywood unions said they either had no comment or did not respond to inquiries about the matter.
In some ways, the cat may already be out of the bag. In 2025, Disney and Universal filed a lawsuit against Midjourney, an image and video production company, alleging plagiarism after the company siphoned copyrighted material onto the Internet as training data. Later that year, Warner Bros. Discovery also sued the company over similar claims. Storyboard artist Phil Langone says:Thunderbolt*, skeleton crew), “I’m an artist and I know that my work is being crawled and it’s being stolen. I’m like, ‘Oh my god, that war was lost before I even realized it was being fought.'”
How much RLHF ultimately improves writing and other models is an open question among engineers. While there is little doubt that it is useful and necessary for humans to train machines, the technology is so powerful that its creator won the prestigious Turing Award last year, but it is far from foolproof. Some of the challenges of RLHF are pandering, requiring a model that fundamentally agrees with human trainees but still produces what you want in real-world situations. That model could then become weaker and actually less threatening to human entertainment workers who worry about their work being interrupted.
Still, AI companies likely need subject matter experts to turn their challenges into more sophisticated deliverables. As a result, more entertainment companies could be persuaded to cut back on human employment. And in desperate situations, people don’t tend to think what-ifs or 10 years ahead. When asked if he had any friends in Hollywood involved in AI training, Woolworths said, “Yes.” He added, “And a lot of people are definitely trying.”
