Missing AI Leadership on Labor Policy

Betsy Masiello is a founding partner of Proteus Strategies, a boutique tech policy strategy and advocacy firm.

There are a number of storms brewing for AI companies, whose public policy profiles have gone from zero to sixty faster than any tech companies that came before. Copyright lawsuits have piled up, privacy regulators are having a look, legislators and regulators are actively drafting new rules around the world, and of course the existential safety debate continues to rage.

But a slower-motion storm is brewing: the likelihood that an increasingly animated US labor movement will come for AI.

Disruption to Labor Markets is Coming

The economic impacts of AI, while likely to be vast, are impossible to predict with certainty. Much of the existing research has focused on analyzing and predicting how many jobs AI might take, and which ones. Major management consulting shops and investment banks have written reports, as have economic policy think tanks and academics. VCs are penning thought-leadership pieces hailing a new era of startup scaling, where fewer people than ever are needed to run a business. And among the AI leaders, OpenAI deserves credit for releasing its own analysis of this question.

My goal is not to rehash any of these reports, so instead I offer a very simplified TL&DR:

(1) Roughly 30% of today’s jobs involve a significant number of tasks that AI could do, with greater impacts felt in traditional “white collar” and higher-income jobs;

(2) These impacts may not result in 1:1 job replacements, but significant task replacement will undoubtedly have an effect on the nature and number of jobs available; and

(3) The speed of impact should be expected to be startlingly fast, challenging our historical expectations about the timeliness with which novel job categories will emerge.

There seems to be an emerging and relatively broad consensus that AI will reshape the nature of jobs and of our economy as a whole, and that there is a very real risk some workers will be left behind. Industry deserves some credit for not beating around the bush about this quite as much as it may have done in the past, but the problem remains.

Tech Companies Will be Blamed

For well over a decade we’ve seen leading technology companies take the brunt of criticism for rising economic inequality and complex social problems. Putting aside the merits of such criticism, it might be wise to understand the last few years of techlash as a prelude to a more profound conflict that is starting to gather steam.

A colorful example that many in tech will remember is the 2013 protests against the Google shuttles in San Francisco, seen at the time as an emblem of rising inequality and tech largesse. Around the same time, the tech industry found itself headlining a global debate about all corporations paying their fair share of taxes, a debate which it’s worth noting has yet to be resolved by actual policy reform. And of course the rise of the tech-enabled gig economy has directly pitted low-wage workers against the technology companies they depend on for jobs, leading to an extended and complex policy movement to regulate gig economy labor practices.

Labor has Wind in its Sails

Across the economy, labor is resurgent in the United States. Even as union membership remains at historic lows, recent green shoots in the labor movement suggest shifts in the power dynamics between labor and capital.

Most notably, perhaps, is the United Auto Workers (UAW) strike, the first strike across the Big Three manufacturers in history. But this strike comes on the heels of years of labor organizing activity in ecommerce and services companies:

• More than 250 Starbucks retail stores have been unionized;
• Workers at numerous Amazon warehouses have unionized or signaled interest in unionizing; Las Vegas strip workers have just voted to strike;
• Scores of novel union-like organizations (e.g., NDWA, USSW) have emerged as ways to build collective power in non-unionized workforces;
• Aggregate data shows increases in the number of strikes and National Labor Relations Board (NLRB) petitions to unionize, and public opinion data indicates popular support of unions is reaching all-times highs.

All of this suggests the landscape is primed for a new conflict centering around tech companies, holding AI technology responsible for harms suffered by workers — 20% of whom now say they fear being replaced by technology. And mobilization against AI could come more easily than movements against automation we’ve seen before. The likely broad impact of the technology suggests such a backlash may be more distributed across the economy. Anger at the tech firms developing the foundational models may risk their bottom lines: the customer support agent at American Airlines who is already training the AI that will do his job for him will be an easy target for political mobilization, and not just mobilization against the airline, but also the developer of the AI software it uses. Other easily mobilized workers will be the software engineers, graphic designers, technical writers, marketing managers and other workers whose jobs stand to be impacted by AI. When these workers mobilize, it might take the form of simple protests or increased political engagement, but certainly it will come with growing demands for better working conditions, better pay, and better benefits.

That may sound like a bigger problem for companies in other sectors than it is for the tech firms themselves, whose workers generally aren’t unionized. But we’ve already seen an example of how things could play out as workers organize against what they perceive as the threat of generative AI. In Hollywood this summer, both the WGA and SAG-AFTRA unions recognized that AI might jeopardize their members’ livelihoods, and as a result AI was a significant issue in their negotiations. Rather than oppose the use of generative AI outright, the WGA took a more nuanced position, and ultimately got concessions in the final deal that says generative AI can be used to complement, but not replace, writers. It is not hard, though, to imagine future worker mobilization efforts to call for greater restrictions on the use of AI altogether, and in fact this has already begun to happen.

While the Hollywood strikes played out in Los Angeles this summer, 400 miles north in Sacramento a different conflict over automation has been playing out. In June the California Assembly passed a bill supported by the Teamsters union that limits the use of self-driving technology in trucks unless a driver is present. Governor Gavin Newsom vetoed the law, pointing to an already-started regulatory process to resolve any issues, and salvaging a business case for autonomous trucking in the state. But that veto doesn’t at all suggest the path for autonomous vehicles is clear. Last year, California’s Occupational Health and Safety Standards Board declined a petition to update regulations and allow autonomous tractors in the agricultural sector, pushing off additional regulatory review until 2026, and putting that business opportunity on ice — at least in part because of industry’s failure to involve labor interests.

These early efforts at restricting the use of AI reflect a fear that may already have penetrated the hearts and minds of workers across the economy. It is thought that over 100 million people have actively used ChatGPT already, a fact that white collar workers are likely unsurprised to hear. Who among us hasn’t experimented to see what bits of their work ChatGPT or Bard can take on? Even leading politicians have given it a try. As Gov. Newsom recently described, upon asking ChatGPT to draft a speech (a draft which was “10x better”), his speechwriter “came back very distressed.” That speechwriter surely felt the fear of AI coming for his job, and workers of all stripes undoubtedly took the signal that the tech is moving quickly.

An Expected Strategy: Kick the Can Down the Road

This landscape is what the hypergrowth AI companies find themselves slowly wading into. Some deserve credit for putting forward novel policy ideas and championing experiments in redistribution, such as universal basic income. But sooner or later, opponents will widely highlight gaps between the policy ideas and practical realities, the challenge in financing redistributive solutions without tax reform, and that companies often have a tendency to gesture at solutions without putting their advocacy muscle behind them.

These statements aren’t intended to be criticisms. For tech companies to go out on a limb and, for example, advocate for higher taxes to finance novel retraining or redistribution policies may be counter to their short-term self-interests. It’s not hard to imagine some companies facing shareholder lawsuits over such activity. Instead, companies have incentives to navigate these issues slowly, quietly, and somewhat reactively.

International tax policy advocacy of the past decade points to examples of strategic engagement on the part of these companies. Though the writing was on the wall for years in the form of public opinion, media and political pressure, and even in messaging from tax authorities, many companies chose to wait and close down their tax haven structures only when required. And some companies ultimately prevailed in the courts, despite facing years of headwinds. Similarly, while leading gig economy companies have engaged politically on labor policy in very high-profile and even controversial ways, they have really only done this in response to clear and direct legal threats on their businesses.

A Different Way Forward

It would be understandable and unsurprising to see tech companies engage with labor policy issues in slow motion, waiting until the last minute to pick the can up off the road. But the landscape above suggests to me that for companies developing AI systems that will disrupt the labor market, the challenge is great enough to warrant early and strategic engagement.

Industry is acknowledging risks of AI and the need for policy change in ways that it didn’t at similarly early stages of technology development. While companies are certainly exuberant about the technology, they have shown more willingness to see regulation and policy as critical to making the technology a success and addressing real concerns.

That willingness to accept the necessity of regulation is an asset these companies should bring to bear on labor and tax policy. It would be a step forward to simply acknowledge that today’s system is not up to the challenge of the AI era. Some individual leaders have gestured in this direction, but I’m talking about something different: imagine the industry collaborating to publish a seminal report detailing the concrete ways in which current policy frameworks may not be up to the task.

As one example, academics have pointed to the tax system that currently incentivizes investment in capital over labor — is there a future in which AI companies might acknowledge that while this helps their business in the short-run, it risks harming the economy at large in the long-run and needs reform? As another example, nearly everyone has acknowledged that the AI era will require a vast retraining effort, but proposals on how to finance that and implement it equitably are lacking —is there a future in which AI companies might advocate for tax policy changes that incentivize industry-led retraining efforts, and/or finance publicly-led retraining efforts?

There are of course a range of options for how these companies might choose to engage on these issues. Some might see value in direct engagement with labor unions, particularly in countries with high labor union participation. Others might continue to pursue their own research on the likely impacts, commission others to study it, or work with think tanks and organizations like OECD to advance the research, putting faith in the economic research. Surely a handful – if not all – of these companies will choose to support retraining efforts by partnering with NGOs in ways that build political alliances and generate good press. These efforts are all in the standard playbook, and would be unsurprising.

But I would love to be surprised. It would be an idealist’s dream to see this generation’s leading business and technology minds take a stand, regardless of political affiliation. The AI economy needs to work for everyone if it’s going to work at all.

Imagine a 2024 where the leading tech companies have banded together to stand up an independent, apolitical commission on labor and tax policy. Such an effort could be one step removed from the corporate brands, and thus not require up-front commitment to supporting any particular policy proposal. Doing anything on these issues in an apolitical way would be a real challenge, but an admirable goal and one that, if clearly stated upfront, could be optimized for. As many of these companies are headquartered in the US, one goal may be feeding research and a set of substantive policy proposals into the next President’s agenda — regardless of which party winds up winning.