Shaping Markets

Ideas for Shared Economic Prosperity in the AI Transition

04. 09. 2026 • Becky Chao, Director of Policy and Research

How society builds and manages AI is a public choice. Government must ensure it benefits all Americans, not just a select few.

Download Policy Brief

Modernize the Social Contract
Protect & Empower Workers
Shape AI Markets
The Affordability Crisis

Artificial Intelligence (AI) is transforming our economy. By some measures, breakthroughs in AI are already delivering initial benefits. In health care, for instance, AI is assisting in diagnosing illnesses, which could relieve pressures on medical staff and help patients get treatment sooner. In San Jose, Calif., AI tools are helping buses reach green lights more quickly, allowing them to run at higher speeds and reducing riders’ commute times by 20%. According to one estimate, generative AI is projected to boost productivity and the U.S. GDP by 1.5% by 2035, nearly 3% in 2055, and 3.7% by 2075 respectively, by automating tasks and increasing efficiency.

At the same time, the rush to build AI tools and systems comes with real costs and consequences. Some corporations are exploiting and profiting off people’s data, labor, and energy to power the AI industry, already demonstrating instances of deepening inequality. Renters paid higher rent when corporate landlords used RealPage’s algorithm to coordinate prices; some families are paying higher electricity bills as data center expansion drives energy costs up; and preliminary research suggests that AI may already be lowering paychecks while displacing entry-level workers.

People are rightfully nervous about AI and the disruptions that accompany this economic transformation; the public is pushing for rules, safeguards, and accountability mechanisms that will protect them from its impact. A majority of voters want more AI accountability, with 57% in favor of more government oversight, even if it means slower technological progress, and people are more concerned about who AI benefits than what its tools can actually do. Investors are making real-time moves based on predictions of how AI will impact the economy. Headlines are reporting that tech companies may already be making decisions to lay off millions of workers in anticipation of AI’s potential, stoking workers’ fears that they’ll eventually be displaced or negatively impacted by AI. Without policy interventions, AI’s impacts have the potential to exacerbate an already fragile economy, where many Americans are living on the brink and making difficult trade-offs about whether to forgo medical care and meals or rely on payday loans and cash advances just to survive.

While it remains uncertain how many workers will be impacted by AI, policymakers must be prepared for various scenarios, as even minor shocks to people’s economic security can derail a generation. 58% of voters agree government’s top priority should be funding the creation of new jobs and basic needs like healthcare to help displaced workers, even if that means limiting how much American tech companies can profit from AI.

Given the universal uncertainty of what the future of AI will bring, this moment demands both (1) policy experimentation to ensure fair outcomes and (2) strengthening the systems at the foundation of our social contract to navigate this transition.

Ultimately, how society builds AI and manages this transition is a choice the public should make. Civil society and government must not allow a small set of corporations to dictate the means and ends of how AI is deployed, and should instead put mechanisms in place to ensure AI is built to benefit all Americans, not just a select few. Policymakers, researchers, and advocates at the federal, state, and local levels must be prepared to critically examine AI’s likely impacts on society, and actively explore solutions that level the playing field and guarantee economic stability for all Americans.

This brief lays out policy ideas to meet the moment, building on Economic Security Project’s previous work on the political economy of AI. These ideas require additional analysis and experimentation to inform robust policy design and implementation. As the ways in which AI begins to reshape our economy become clearer, policymakers, researchers, and advocates will need to build on these ideas and innovate around new policies to address AI’s impact. The ideas outlined in this brief identify pathways for policymakers to (1) modernize the social contract for an AI-driven economy; (2) protect and empower workers; (3) shape fair and competitive markets; and (4) mitigate AI’s role in exacerbating the affordability crisis.

Modernize the Social Contract for an AI-Driven Economy

Prioritize Resiliency and System Readiness for AI Disruptions

AI has the potential to disrupt work and introduce more economic precarity for workers and families.The scale of this disruption remains to be seen: some argue AI has the potential to make work obsolete entirely, while others predict that its widespread adoption may facilitate the creation of new jobs. The Brookings Institution estimates that 6.1 million U.S. workers working in clerical and administration roles with high exposure to AI will struggle to adapt and find new jobs. Regardless of the scale of AI’s impact on workers, policymakers must prepare economic policies that provide people with the dignity and care they deserve. As Jacob Leibenluft writes in learning from the failure of past trade adjustments to help displaced workers, deploying policy solutions that match the scale of disruption, prioritizing speed in delivering the assistance, and supporting workers’ agency and autonomy will be key in addressing AI-related job loss.

Policymakers can ensure a smooth transition by evaluating whether existing economic policies sufficiently support people when they need it, and exploring new interventions to fill any gaps. Emerging ideas that build on existing social insurance programs include:

Improve uptake and offerings of unemployment insurance: Unemployment insurance will provide workers with critical bridge support until they find new work if they are displaced by AI. To ensure unemployment insurance is effective, however, policymakers must address the gap in uptake, as the majority of unemployed workers may not even apply for benefits, and explore complementary policies to support employment gains like reskilling and upskilling.
Build towards a more robust public healthcare system: Decoupling health insurance from private employers can ease disruptions arising from AI-related job loss while ensuring continued access to health care services. For U.S. residents under 65, employer-sponsored health insurance is the largest source of health coverage. If AI leads to job loss, policymakers must contend with the fact that many workers will also lose their health coverage.
Establish an income floor for displaced workers: To mitigate the uncertainty workers face in the age of AI, direct cash transfers could also provide an economic floor allowing people to weather the shock of labor market changes. Direct cash interventions, however, shouldn’t be used as a justification for labor displacement from automation or substitution for including impacted workers in conversations around AI adoption and regulation.
Reevaluate the traditional models of work: AI boosts in productivity could help employers shift to a four-day work-week, which could also help reduce unemployment from job displacement by sharing the remaining work more widely. There are important questions about whether a four-day work-week would also lead to a reduction in workers’ salaries to reflect four days’ worth of pay. As conversations about AI’s potential evolve, it’s critical to wrestle with how workers can ensure that they also benefit from AI’s productivity gains on measures such as free time.

Reform the Tax Code to Ensure Shared Wealth

Tax policy can be an important tool to generate tax revenue and address growing inequality in the age of AI. Economists have studied how tax policy can actively shape labor markets. Daron Acemoglu, Andrea Manera, and Pascual Restrepo found the U.S. tax code incentivizes companies to invest in robots and software over people and labor, potentially leading to too much automation that displaces workers. There’s a risk that as AI leads to wealth gains for a select few, it’ll leave workers behind, collapsing the labor-share of income and compounding existing inequalities in our society.

Researchers, scholars, and advocates have started to explore different tax vehicles, like corporate taxes or a wealth tax, to address this problem. Tax policy can also raise revenue to incentivize and potentially support high-quality, in-demand jobs in sectors like healthcare, childcare, and the trades like construction and maintenance. Preliminary proposals for various forms of an “AI dividend” have already been introduced to potentially compensate people for the use of their data to train AI models. An “AI dividend” is modeled after the Alaska Permanent Fund, which compensates Alaskan citizens using investment returns on a sovereign wealth fund funded by oil and gas royalties. These ideas surface important debates about how policymakers can ensure AI benefits people whose data it’s derived from while giving people more agency over their lives.

Protect & Empower Workers

Secure Robust Worker Rights and Protections

Incorporating workers’ perspectives into the policymaking process can help ensure more equitable and safe AI development and implementation. AFL-CIO and TechEquity have developed principles to protect and empower workers in the AI transition that policymakers should consider. People should not have to submit to intrusive, invasive surveillance when they go to work. Policymakers should consider requiring employers to provide full transparency about how AI is used to monitor and surveil workers in the workplace, both in real life and in virtual workplaces. Policymakers should also prioritize prohibiting common, harmful uses of AI surveillance tools, including: facial recognition, biometric surveillance, predictive behavior modeling, and monitoring and quotas that violate health and labor laws. High-quality AI workforce and digital literacy training and education programs will also be key to ensuring a smooth transition for workers impacted by AI-related shifts in in-demand job skills.

Regulate AI-Driven Wage-Setting and Scheduling

Employers can use AI algorithms on gig platforms like Uber to determine rideshare drivers’ pay and work schedules, ultimately encouraging workers to work for lower pay and less predictable work schedules than if they were full-time salaried workers. These gig-work labor models and the algorithmic management technologies that these apps use to pair workers with gigs are also being exported to other labor sectors, like health care. For instance, many nurses use on-demand nursing apps to indicate interest in picking up shifts in various medical facilities. These gig platforms advertise this algorithmic scheduling software as cutting edge AI that matches understaffed facilities with nurses looking for work. But there’s usually little transparency into how these algorithms set nurses’ pay: these apps show workers different shifts on different phones, often for different amounts of pay, and sometimes encourage workers to bid against each other by indicating the lowest hourly rate they are willing to work for. Policymakers should regulate these AI-driven wage-setting and scheduling practices to protect workers.

Shape Fair and Competitive AI Markets

Strengthen Public Investment for Public AI

Public investment can steer AI innovation in the public interest. Ideas like a national investment fund or similar initiatives like the National Artificial Intelligence Research Resource (NAIRR) can leverage private-public partnerships to offer researchers, educators, entrepreneurs, and small businesses critical access to resources like computing power, datasets, and training to build public AI. Through an application process, NAIRR decides which U.S.-based researchers and educators get access to these resources. States can also explore interventions like CalCompute, a public option for cloud computing established by California state legislation SB 53 in 2025. These forms of government support can help lead to new AI solutions that address pressing public problems, like critical medical research and benefits administration.

Deploy Antimonopoly Tools to Foster Innovation & Check Bad Actors

Consumers may think the AI sector is competitive based on the plethora of affordable AI models and applications available on the market today. But competition can be fragile as AI markets evolve, and the AI products most consumers interact with are only one part of a multi-layered, interconnected AI stack. One of the most essential keys to competition lies elsewhere in this stack, specifically in the infrastructure layer, where a handful of gatekeepers provide cloud services, computing power, and chips. Without access to this critical infrastructure, models and application developers would not be able to even build their products and deliver their services to end users. Crucially, these third-party models and applications may also be competing with the infrastructure providers’ own offerings at the models and applications level. Model developers and infrastructure providers are also leveraging investment partnerships to secure access to these important inputs, raising important antitrust questions about whether these deals may create competitive bottlenecks.

Antimonopoly solutions can ensure a level playing field for nascent and early stage start-ups and entrepreneurs in a dynamic marketplace and should be considered in specific contexts. For example, there may be a risk that infrastructure providers can cut off or restrict critical access for applications reliant on their services, so that they give their own applications preferential treatment, known as self-preferencing. In this scenario, enacting non-discrimination rules for infrastructure providers would require them to provide equal service and equal terms to all similarly-situated customers without favoritism. In another scenario, the Federal Trade Commission is investigating whether Microsoft’s business practices make it harder for customers to use Windows or Office on rival cloud services because it hasn’t made its own software products interoperable with them. These dynamics underscore the importance of data portability and interoperability rules in enabling users across different platforms to plug and play across the different systems. Ultimately, antitrust and competition policy can help ensure a thriving, competitive AI marketplace.

Mitigate AI’s Role in Exacerbating the Affordability Crisis

Prohibit Companies from Using AI to Charge Consumers Higher Prices

AI supercharges certain pricing strategies and increases their speed, scale, and scope. AI has made it possible for companies to leverage personal data from multiple sources on an unprecedented scale to set personalized prices based on real-time signals like consumers’ locations, demographics, credit history, and browsing history. As a result, companies can charge consumers higher prices for products based on their search history or rideshares based on smartphone battery level, as Uber has been accused of doing. Separately, companies have used pricing algorithms for decades now. They’ve historically needed to program parameters and rules for pricing algorithms. But machine learning, a type of AI, has made it possible for algorithms to learn from data without explicit programming. Using these AI tools, companies can set prices based on changing market dynamics like supply and demand in real time.

These AI-enabled algorithms can also learn to maximize revenue over time, sometimes facilitating collusion to hamper competition and set higher prices for consumers. It used to be that companies could only collude by agreeing to manipulate the market through direct communication. Price-fixing is one type of collusion where companies agree to set prices together. With the rise of AI-enabled algorithms, companies are now able to fix prices even without competitive data directly exchanging hands or being instructed to cooperate instead of compete. Corporate landlords, for example, were able to coordinate using RealPage’s AI-enabled algorithm and charge renters 4% higher prices, an average of $70 a month, in housing markets across the country. Policymakers can save consumers money by exploring interventions that prohibit these AI-enabled pricing strategies.

Ensure Households Don’t Pay for AI Data Centers’ Energy Costs

Electricity prices have increased across the board, but households are experiencing bigger increases than commercial users like data centers, even when data centers’ electricity demand is growing at a faster rate. The typical U.S. household was already paying $142 per month for electricity in 2024. Yale Climate Connections calculates that between 2022 to 2024, residential electricity prices increased 10% while commercial users like data centers experienced a 3% increase in comparison. U.S. data centers’ electricity use tripled from 2020 to 2024, whereas residential users’ electricity demand increased only 1% between 2021 and 2024. Goldman Sachs projects that relative to 2023 levels, global data center power demand will grow by 220% in 2030, with about 60% of this demand coming from the U.S.

These substantial loads will require significant investments in U.S. utility grid expansions, like new power plants and transmission lines, to meet demand. Forecasted capital investments to support AI-related data center demand have varied from over $600 billion between 2026 and 2030 to about $3 to $8 trillion by 2030. In certain markets, these capital investment costs are likely to be subsidized by consumers who are already seeing electricity price increases of 6.9% in 2025 year-over-year, even as data centers are driving 40% of this electricity growth demand. Households are on track to see an additional 6% increase in electricity prices through 2027. The Natural Resources Defense Council estimates that the average family serviced by grid operator PJM will pay approximately $70 more per month on their electricity bills because of forecasted data center growth.

While energy is a state-by-state issue, policymakers should explore potential policies that could shift the burden away from households and back onto data centers. For example, states and localities could mandate transparency into energy usage and pricing, to inform decisions about data center development, including how to manage forecasted load growths stemming from data centers. Policymakers could also introduce fair cost allocation, including possibly creating new classes of users on electrical grids to distinguish between data centers and households. Localities could then choose to charge higher rates for clients with greater usage. States and localities could also impose a large load utility tax on clients that use disproportionately more energy to fund heat pumps and rooftop solar in nearby areas.

Conclusion

AI is rapidly developing and becoming increasingly integrated into our society. The speed and scale of AI can be immense, and the uncertainty about how it will transform our culture is already overwhelming our politics. But its impact on the economy is something policymakers, advocates, and researchers can shape. We have navigated major economic transformations in the past, and we know better the tools that can help ensure those transformations benefit people. The basic rules of a fair economy still apply. As shown, competition, worker power, block abusive behavior, and having the economic bedrock to manage disruptions while benefitting everyone will remain as essential as ever.