How we can protect workers in the age of AI

California is home to both the country’s highest concentration of billionaires and the highest poverty rate. Despite promises that AI would flood the state with jobs and tax revenue, the gap between the haves and the have-nots is only widening.

AI is being used to increase profits for shareholders and executives while workers are laid off, surveilled, and exploited—including the workers behind AI.

Given widespread plans for AI adoption across industries, it has never been more urgent for elected officials to act to prevent harms and ensure the economic potential of AI does not rely on exploitation to succeed. 

That’s why we wrote a memo articulating the harms workers are experiencing from AI and outlining solutions that elected officials can adopt to improve the lives of working people, while also better serving businesses and consumers. These were the top takeaways.

What’s at stake?

AI and digital technologies impact nearly every aspect of people’s livelihoods, from their workplace rights to whether they can find and keep a job. These are some examples of how employers’ use of AI is determining the life chances of workers in California now:

AI may result in job displacement and increased precarity: The same CEOs who design and sell AI technology are simultaneously raising alarm bells about the potential for massive job displacement, and we are beginning to see how these AI technologies pose a direct challenge to workers. For instance, AI tools have contributed to replacing actors and writers and also used their creative output to generate new content for which workers are not paid. Similarly, computer programmers are employed to train AI models, which are likely to deskill or possibly displace their current roles. This isn’t to say AI-related layoffs aren’t real and already happening, but it is to say there’s more to it than meets the eye.

AI tools are being used by employers to demean and exploit workers: AI systems are now widely used by recruiters at all stages of the hiring process, from sourcing applicants to final selection. According to a study by Upturn, applicants can be exposed to a range of biases during hiring that harm their prospects. Employers are also using AI systems to surveil and manage workers—with dire consequences on their financial, mental, and physical well-being. In a survey by PowerSwitch Action and Gig Workers Rising, 40% of Uber drivers reported continuing to drive when tired and even in pain to earn enough to make ends meet under the uncertain conditions created by algorithmic management tools. 

Global AI supply chains contribute to a race to the bottom on labor standards: Data workers enhance datasets and train the algorithms that form the basis of AI. Yet, data work tends to be low-wage (as little as $1.50 per hour in the Global South), and workers also report severe mental health impacts of this work. This is not just a problem for workers overseas. U.S.-based data workers surveyed and interviewed by TechEquity, Alphabet Workers Union, and Communications Workers of America shared similar challenges, including struggling to make ends meet, inadequate training, pressure to meet unrealistic work quotas, and limited access to benefits. 

The importance of prioritizing worker voice

In addressing these issues, it’s important to prioritize the voice of workers throughout the process of developing technology that impacts them at work.

Studies show that when clear guardrails are in place and workers are given a voice, “input from workers can increase the likelihood that organizations use generative AI tools effectively, and workers’ job quality improves.” Employers and technology companies should authentically engage workers and unions, who are closest to the technology and best suited to speak to their own needs more broadly.  

Meanwhile, an MIT Project NANDA study found that 95% of enterprise generative AI pilots failed, with major contributing factors including a lack of trust in the products and the failure to align with employee workflows. 

Excluding workers from decision-making can also result in significant harm to communities. For example, in a survey conducted by National Nurses United (NNU), 69% of nurses who tried a tool that used algorithms to analyze electronic health records reported disagreeing with the computer-generated estimate of a patient’s illness severity and the number of hours of nursing care they needed. These measures are used to make staffing decisions, so incorrect estimates can lead to understaffing and critical gaps in patient care. 

Engaging workers in the technology adoption process can help stem the tide of job loss. For example, unions have already negotiated collective bargaining agreements that include job security protections, pay protections, and severance programs, as well as comprehensive technology transition agreements and training and workforce development for technological change.

We need stronger guardrails to make sure there is human review and oversight of any decision made by a machine that impacts a worker’s job and paycheck.

Lorena Gonzalez President of the California Federation of Labor Unions, AFL-CIO

Solutions to support California’s workers and economy

So what should policymakers do next? Policymakers must take a holistic approach, with a focus on enhancing worker agency, to address the risks of AI in the labor market

To improve conditions for workers and society at large, it is necessary not only to regulate current technological development but also to address the underlying economic and political conditions that make it easier to extract and exploit workers. 

Priorities for elected officials in California should include:

• Shifting the balance of power in the workplace by expanding collective bargaining and access to unions: Policies should expand access to collective bargaining and protect workers’ right to unionize, and workers and their representatives should be included on any boards, commissions, or other multistakeholder entities established by the state related to AI and other digital technologies. 
• Making a plan to address the potential for AI-related job loss: One early step would be to require employers and AI developers to provide transparent data around the use and impact of AI products within enterprises. This data can help identify workers most likely to be affected, design effective workforce training programs, and develop new models of safety-net transition assistance that can aid affected workers. 
• Establishing and enforcing workers’ technology rights and proactively preventing the most harmful impacts of technology: Workers need transparency about how tech affects them at work—from hiring and firing decisions to how their wages are set to whether and how they are surveilled—as well as the right to access, correct, and control the use of their data. California should ensure that no one is subject to the most egregious applications of workplace technology. 
• Ensuring that California-based AI companies are held accountable for improving labor conditions throughout their supply chains: Supply chain labor abuses are not new or unique to the AI industry, and policymakers can learn from improvements in the garment industry and other sectors.
• Investing in state regulatory and enforcement infrastructure: While there are existing laws on the books that protect workers from many workplace harms, enforcement can be challenging when those harms are perpetrated by technology. California’s leaders must ensure that the people vested with the authority to investigate and enforce our labor laws have the time, money, and talent they need to do so. 
• Incentivizing employers to retain workers: Nobel laureates Daron Acemoglu and Simon Johnson have proposed new tax structures to incentivize employers to retain workers and to disincentivize the use of AI that is merely there to displace workers. 

With these policies, we could build a future in which AI benefits workers, rather than being used to lay them off, surveil them, and exploit them.

The future is ours to build

California’s leaders hold the future of the state’s workers in their hands. As AI spreads throughout a labor market in which employers already have enormous power over workers, the health of the state’s economy hinges on our leaders making bold policies that protect and empower workers.  

That’s why we wrote “AI & California’s Labor Market: Reckoning with AI’s Impact on Workers,” a memo outlining ways California leaders can take the reins to protect workers as we move into the age of AI. This memo is part of TechEquity’s California’s AI Future series, which includes three other memos addressing the political dynamics of AI in California, how AI usage is making California less affordable, and an alternative vision for California’s tech economy.

Have a personal experience with AI and work? When you share your story, California leaders have more information and are more emboldened to stand up for everyday Californians.