Labor unions’ fight against AI is nothing new

Disruptive technology is at the heart of the contentious negotiations between the Writers Guild of America and studios, networks and streaming services. Last week, those negotiations failed and the screenwriters went on strike. At issue is how writers get paid for streaming content and the role of artificial intelligence in the creative process.

The WGA has pushed for guardrails on the use of new generative AI tools like ChatGPT, which are trained on vast amounts of human-made creative work and could, some fear, end up replacing it.

It’s a concern that is popping up more and more across a number of different industries as the implications of this technology come into focus.

Marketplace’s Meghan McCarty Carino spoke with Virginia Doellgast, a professor at Cornell University’s School of Industrial and Labor Relations. Doellgast said the union’s efforts to contain the harm of AI echo past labor struggles with new technology.

The following is an edited transcript of their conversation.

Virginia Doellgast: There are a lot of examples historically of how workers have responded to new technologies. You could go back to the classic Luddites, English textile workers in the 19th century who destroyed textile machinery because factory owners are basically using machines to cut wages and replace skilled artisans. You have a lot of examples of robots being introduced in the auto industry or in manufacturing. You had the longshore industry being heavily mechanized. Thinking about the call center environment, you’ve already seen a lot of use of AI, as well as, you know, broader trends of digitalization.

Meghan McCarty Carino: Tell me about some of the ways that disruptive technology like this can affect jobs.

Doellgast: So I tend to break down these effects into a labor-replacing bucket, a labor-controlling bucket and a labor-displacing bucket. So labor-replacing effects are where companies are using AI to automate tasks, and labor-controlling are where companies are applying analytics and algorithms to manage or direct workers. This is often described as algorithmic management. And then the third kind of AI effect is labor-displacing, which means that companies find it easier to relocate jobs using this combination of AI-based management tools, cloud computing and faster data speeds.

McCarty Carino: And how can unions respond to some of those challenges?

Doellgast: So how unions and their members respond to technologies really depends on how much say they have and how technologies are used in their workplaces. They tend to resist new technology when it’s either likely to lead to job cuts or to make jobs worse, and they’ll cooperate, on the other hand, where they can negotiate good contracts. The Writers Guild strike is a really interesting example of conflict over how new technology will be used and who will benefit. It’s also a good example of how important unions are in fighting for worker voices in these decisions. And this is really a classic case of control over work, right, whether this new technology is used to deskill, make work more precarious. And then questions about how you share productivity gains and profits.

McCarty Carino: Do writers here sort of look like the canary in the coal mine?

Doellgast: Yes. I mean, I think there are specific issues coming out of this conflict which you can kind of see echoes of in other creative parts of industries. So I’ve been doing some interviews on the use of AI and game development, and you see similar issues. So because these image-generation tools can do 3D modeling, and there are early examples of games that use AI-generated art, so artists are feeling threatened. These are the more high-skilled creative workers. And in a sense, the kinds of issues they’re facing are similar to the writers, because the large-scale neural networks that are used to generate these images, they’re trained on huge numbers of images scraped from the web. And of course, those images are originally created by the artists themselves, right? So the question of who owns these images, is this copyright infringement when it’s a black box on how images are being used to produce AI-generated art, it is kind of an open question and something that’s being fought over in places where there are and where there aren’t unions.

McCarty Carino: What would it look like to incorporate new technologies in the workplace in a way that protects workers or shares the benefits with workers?

Doellgast: Having some process in place where you have dialogue, or you have workers really being consulted over technology before it’s adopted, it can be really useful in convincing managers not to just buy every shiny new toy and to really think through how these tools might really affect performance, change the work, how can they be used more productively, what kind of changes in skills need to be made. So I talked to a works counselor from an IT firm based in Germany who was saying, “You know, we have to constantly educate our managers. Our engineers are actually more productive when they don’t have to log into the company’s monitoring system, when they’re not recorded all day long while they’re doing their jobs.” And, you know, the works council gathered data, their engineers, they presented that data and said, “You know, look, in Germany, we’re actually more productive without the invasive AI-based monitoring system.” But, you know, this kind of idea often runs counter to the intuition of managers who are hungry for more data and more control. Or in the case of the writers strike, they want to have more future potential, control in the future over how they might use this new technology. I think in the longer run, it’s really an advantage to companies themselves or to the studios to have to step back and really think about how those technologies fit in with their broader strategies, what kinds of changes in skills or work design might be needed, and then to bring their workforce along with them, who really feel like they’re valued partners and feel like they can have some say and some control over those choices.

Related links: More insight from Meghan McCarty Carino

Last year, when automation became a sticking point in West Coast port negotiations, I looked into how the dockworkers have approached previous innovations, like the steel shipping container.

I know, compared to artificial intelligence that can write a screenplay, the shipping technology sounds pretty prosaic. But standardized shipping containers that could be moved by cranes were a disruptive innovation when they were introduced in the 1950s. They made moving goods a whole lot faster and safer, but also less reliant on the labor of dockworkers.

The powerful port union was able to get promises that no existing workers would be fired and that they’d be first in line to operate the new machinery.

Of course, the port is in a unique situation. Most workers in the United States don’t have the leverage on their side of holding up imports to the country. And not to mention, labor unions are pretty rare among white-collar workforces whose jobs could be most affected by AI.