Why AI Researchers Need To Hear More From Workers and Unions

At a time when many nations aim to lead the global race around artificial intelligence, quantum science, and other emerging technologies, policymakers need to better understand how to maximally serve the public interest. Central to that goal is the need for taxpayer-funded research and development leaders to understand the downstream ramifications of their technologies on the frontline workers who will use them.

To do that, policymakers should task government agencies that fund this work to study, pilot, and scale approaches that foster partnerships between the labor sector and R&D leaders in academia, industry, and technology-based economic development.

Governments around the world have long invested heavily in technological R&D that has fueled patents, startups, jobs, and economic growth. In the United States, innovations like the Google search engine, smartphones, and even the internet itself all trace their roots to government investments in R&D. Roughly one-third of US patents stem from government-funded research — a share that has increased steadily since the 1970s. Even today’s AI zeitgeist traces back to early funding from the US government in the 1960s.

However, as MIT economists and 2024 Nobel Laureates Daron Acemoglu and Simon Johnson have carefully documented, taxpaying workers rarely have a say in the direction technology development and implementation takes, even when it most affects them.

Instead, policy decision-making concerning R&D strategy is overly concentrated among the Big Tech lobby and academic elites, two groups that polling suggests many Americans increasingly distrust which may not align with public sentiment. Pew found significant gaps between how the public and experts think AI will impact their jobs, and a separate poll found that most Americans believe that tools like generative AI will hurt rather than help them in the workplace.

Science and industrial policy affect workers, and it’s time for policymakers to show they understand that. Publicly incubated emerging technologies like AI can alter working conditions for better or worse. Work augmenting technologies can enhance workers’ capabilities and eliminate dull or dangerous tasks. But poorly designed systems or automation processes that provide little value for consumers or businesses can make jobs more difficult, result in ‘deskilling’ of middle-class jobs, erode civil liberties, and degrade worker morale.

American businesses are already suffering a crisis in worker morale that justifies a new approach to understanding their needs. Gallup has found that US worker engagement levels — the degree to which employees feel committed and motivated by their work — hit a 10-year low in 2024. US Labor force participation — the percentage of the working-age population employed or seeking employment — is down significantly from its height in the 1990s. Workers are also jumping ship more often. Gallup found that 42% of employee turnover is preventable but ignored.

It doesn’t have to be this way. For example, a 2023 study examined the impact of industrial robots across 14 industries over 16 years and found that workers who had a level of control over their use expressed a greater sense of agency and competence in their jobs and felt more collegiality with colleagues than those with less oversight. Another study by Partnership on AI found that when given a voice, call center workers genuinely appreciate AI at work and that it improved their performance.

Researcher-worker collaborations can help technologists identify productivity bottlenecks, overlooked implementation challenges, and solve industry problems that also meet worker needs. They can also mitigate unintended biases that are known to erode trust and buy-in from workers and identify upskilling needs for workers.

Leveraging lessons from policy entrepreneurship

While President Donald Trump is pushing for steep cuts to both government funding for R&D and worker protections, his predecessor, former President Joe Biden, oversaw some of the most substantive and innovative policy experiments to boost both, all while bringing workers and R&D leaders closer together. Those moves warrant continued attention globally.

Under the Biden administration, the National Science Foundation (NSF), the US federal agency primarily tasked with supporting science research and education, signed a historic agreement with the AFL-CIO, America’s largest labor union federation, to foster collaborations between technology researchers and workers.

The first of its kind five-year memorandum of understanding should serve as a model for how organized labor can partner with other government agencies that fund R&D, such as the National Institutes of Health (NIH), the National Aeronautics and Space Administration (NASA), and the Departments of Energy, Defense, and Commerce.

AFL-CIO President Liz Shuler told Politico she sees AI as a way to reinvigorate the labor movement, calling the partnership a way to help ensure that “federally funded science and technology innovation incorporates the needs of workers, creates good union jobs, and provides workers pathways for training,” including around AI.

Her sentiments mirrored what I heard from international leaders while studying how AI can improve job quality at the World Economic Forum. Tim Noonan, Director of the International Trade Union Confederation, the world’s largest trade union confederation, and a member of the Forum’s Global AI Action Alliance, rejected the stereotype that labor unions are made up of technology Luddites.

“Where the change is negotiated, the outcomes are much better for workers and employers. We are seeing this activity grow as the prevalence and power of workplace technologies increases," he told me in an interview.

After decades of backlash, American support for labor unions reached a record high of 70% in 2024, and the number of Americans supporting expanded influence of unions has steadily gone up over the past 25 years.

In some corners, the tech sector has also softened its historically adversarial posture towards labor. In December 2023, AFL-CIO and Microsoft, a substantial investor in OpenAI, signed a historic partnership around AI where workers will inform Microsoft's AI developers. The public is coming along too. The stars have aligned for policymakers to better harmonize labor, science, and industrial policy.

Unions and workers aren’t resisting the fourth industrial revolution — they seek to help shape it. Recent moves to elevate worker voice around AI in labor contracts won by the Writers Guild of America and the Communication Workers of America illustrate what’s possible and set a precedent for other unions to follow.

But facilitating substantive collaborations between technology researchers and workers is easier said than done. The labor sector will need to continue to build its capacity to proactively collaborate with researchers, better understand the intricacies of technology development, shake their policy reputation as Luddites, and emerge as good-faith collaborators if it wishes to be given credence as a value-adding steward of technology. Otherwise, it risks being sidelined by skeptical policymakers and business leaders who aren’t keen to be left behind in the global technology race.

Resources are emerging to support that capacity-building.

MIT professor Thomas Kochan partnered with the AFL-CIO to produce a free online course called “Bringing Worker Voices into Technology and Employment Strategies.” During a joint New America and World Economic Forum event, Kochan encouraged the use of such resources, stating that labor, business, and government leaders "haven't had dialogue about how you bring worker voice into the design and development of technologies.” Meanwhile, the University of California Labor Center has generated resources to incorporate technology provisions into collective bargaining agreements.

Still, more needs to be done to upskill labor leaders’ know-how around R&D policy and technology development.

Including workers in technology development

Since only 10% of Americans belong to labor unions, collaborations must go beyond just collective bargaining within individual unions. Fortunately, policymakers have a range of methods to unite workers and R&D leaders, including participatory technology assessments, public deliberation forums, and collaborative design methodologies.

These techniques are worth policymakers’ attention. They have been used to meaningfully engage the general public around science and technology policy decisions at NASA, the US Energy Department, and beyond. They’re primed to be tested with workers and unions.

Associations like the Expert and Citizen Assessment of Science and Technology (ECAST) and the Public Interest Technology University networks convene practitioners who can provide expertise around applying these approaches in worker contexts. Policymakers should study, test, and consciously improve these by providing dedicated funding for pilot programs, evaluation studies, and through technology governance.

For example, Congress could codify the National Science Foundation’s Enabling Partnerships to Increase Innovation Capacity (EPIIC) program and increase appropriations for it to fund team-ups between labor unions, industry, and university researchers. Federal technology-based economic development investments created under the CHIPS Act, such as the NSF’s Regional Innovation Engines and the Commerce Department’s Tech Hubs, would be well-suited to test these approaches.

From climate change to public health, business and policy leaders face inordinate public problems that urgently warrant greater public investments in scientific research, technological innovation, and worker well-being.

However, without policymakers supporting more worker engagement, these technologies are likely to concentrate power and benefits in the hands of far too few.