Until last year, TV writer Danny Tolli never imagined that AI could threaten his career. It wasn’t until ChatGPT came out at the end of 2022 and he saw its uncanny ability to instantly generate scripts and dialogue that he took notice. Suddenly, he worried that TV studios in Hollywood would use AI to replace writers, destroy the career ladder he was rising, and erode the compensation and job stability he counted on.
As Tolli’s fears grew, his union, the Writers Guild of America West, was negotiating a three-year contract with Hollywood studios. Suddenly, AI became a top priority for Tolli and thousands of Guild members who went on strike in spring 2023.
In a November interview, Tolli shared with me what was at stake for him in the strike:
The contract the Guild secured in September included sweeping, first-of-their-kind guardrails on generative AI. Ultimately, the contract spells out that generative AI should complement, not replace, the writers. If generative AI is used, the contract stipulates that writers benefit with full credit and compensation.
In a new Brookings case study and worker storytelling series, I distill lessons from the writers’ victory and what they mean for other workers and unions in industries that may face similar disruptions from AI – from journalism to customer service, sales, finance, law, creative industries, media, STEM, and clerical work.
Photo by Phil Cheung
The key lessons include the following:
-
Technology isn’t destiny: Workers can shape its design and deployment through collective bargaining. Too often, generative AI is framed as unstoppable, with inevitable impacts on workers. By exercising their power, the writers pushed back on this narrative and succeeded in setting their own terms and guardrails for the use of AI. Other unions can look to the WGA agreement as a key precedent about the possibility of collective bargaining over the use of AI. Despite technology being considered an issue of “management prerogative,” and outside the required purview of collective bargaining, the writers’ contract proves that it is possible for unions to bring employers to the negotiating table for technology like AI.
-
The writers sought to regulate, not outright ban, technology. One of the elements of the Guild’s success was their decision to influence—not ban—the use of a powerful new technology. The WGA’s approach of shaping the use of AI and its benefits is a model for other unions.
-
Proactive, not reactive. The workers acted early, before AI was even being deployed in writers’ rooms. Just months after the release of ChatGPT, Guild leadership moved quickly to formulate bargaining positions to protect their members. By acting before the horse left the barn, the union was able to shape the use of this new technology more or less from the beginning, and establish contract terms for future negotiations. Other unions can learn from this proactive approach. Several have already moved quickly to develop AI positions, educate members, and negotiate terms of AI use in collective bargaining agreements. For instance, just months after the screenwriters and studios reached their agreement, the Communication Workers of America (CWA) published a set of AI principles and recommendations.
- Solidarity with other workers and unions across industries. While on strike, the writers were helped by solidarity from other workers and unions spanning a range of different industries: SAG-AFTRA members, the crew union IATSE, the Teamsters Local 399, and vocal support from the biggest unions in the country, including the AFL-CIO, and by a diverse group of workers on the picket line including teachers, nurses, hotel workers and Starbucks baristas.
Photo by Phil Cheung
- Protections that can enable workers to experiment with AI and benefit from it: Because the contract protects writers’ employment as well as their compensation and credits, it creates an enabling environment where writers may feel “safe” to try out the technology and explore ways that it might be useful to them in their jobs, without worrying about losing their job. Compared to other technology, generative AI lends itself especially well to worker-led experimentation and innovation. This is good for workers, but also good for employers. A growing body of research illustrates the benefits of engaging workers in the design and roll out of new technologies, compared to implementation from the top down. With workers leading in shaping its use, generative AI may be a helpful tool.
While groundbreaking, the replicability of the writers’ victory is undermined by the very low rates of union representation in industries that are most exposed to generative AI. Across the whole economy, only ten percent of workers have union representation – this falls to an even smaller fraction for the industries that AI will likely disrupt the most.
This mismatch highlights the importance of labor law reforms at both the federal and state levels that enable workers to exercise power in their workplace and shape this technology. It also underscores the need for a greater number of institutions --- from worker groups to policymakers, civil society organizations, employers, industry and consumer groups – to open opportunities for worker voice, collaboration and guardrails.
We don’t have to look far back into history to see how unmitigated technological advancement can fuel inequality and cause lasting pain for workers and their communities. Whether workers can benefit from generative AI and avoid harms will depend on workers’ ability to shape it. Hollywood writers show what is possible when workers have a voice in this technology’s future and the power to set their own terms.
