Fitting that first criterion is why public opinion research from Blue Rose Research is so crucial. They recently tested 79 different possible policy responses with American. Each time, they used a best-practice common to Blue Rose reports: giving arguments for and against a policy, and measuring how exposure to these arguments (meant to represent the debate likely to occur once a policy is actually proposed) changes their minds.

Generally speaking, exposure to the debate made public support for policies fall. This makes sense; lots of things seem like a good idea at first glance but less promising when you weigh the details a bit. But even after that, the vast majority of the ideas (61) still were supported on net after exposure to arguments, and twelve ideas were supported by a margin of 40 or more points, suggesting a deep consensus.

So we have some ideas that the public starts out very receptive towards. Great! The question then becomes: are these ideas sound, and sized to the scale of the problem? Here the news is a little less encouraging.

The full list of policies earning a margin of 40 or more is below; you can click to see a more detailed description.

Expand Apprenticeships (+66)

Some policymakers are proposing to scale up “earn-while-you-learn” apprenticeships into new fields — including AI infrastructure and roles that work alongside AI. Under this policy, more workers could train for skilled jobs by working a paid position under experienced mentors.

Require Severance for Automated-Away Jobs (+63)

Some policymakers are proposing to require that companies provide severance pay or transition support to workers they replace with automation. Under this policy, a company that eliminates a job through AI or automation would have to give the affected worker a defined payout or help finding new work.

Sector-Based Job Training (+60)

Some policymakers are proposing training programs built around specific high-demand industries, run with employers who agree to hire graduates, plus support like childcare and transportation to help people finish. Under this policy, workers would train as a group for openings in growing fields.

Employee Ownership (ESOPs) (+50)

Some policymakers are proposing to expand employee stock ownership plans, which give workers an ownership stake in the companies they work for through tax-advantaged trusts. Under this policy, more workers would own a piece of their employer and share in its profits — so that as AI boosts company value, employees benefit alongside investors.

Data Dividend (+48)

Some policymakers are proposing that tech and AI companies pay people for the personal data they collect. Under this policy, the data companies use to build and train AI would be treated as something you own, and firms would have to compensate you when they use it.

No Billionaire Should Pay a Lower Rate Than a Nurse (+47)

Some policymakers are proposing a rule ensuring the very wealthiest pay at least as high a tax rate as middle-class workers. Under this policy, billionaires and the ultra-rich could no longer use investment income and loopholes to pay a lower effective tax rate than teachers, nurses, and other working people.

Invest in the Care Economy (+46)

Some policymakers are proposing major public investment in care work — home care, childcare, and eldercare — fields that are labor-intensive and hard to automate. Under this policy, the government would pay for higher wages in care jobs, expanding a part of the economy where human workers remain essential even as AI advances.

Modernize Disability Benefits (+44)

Some policymakers are proposing to update the rules for Social Security Disability Insurance and Supplemental Security Income so people can build modest savings and work where they’re able. Currently, these programs limit how much money recipients can have in savings, with limits set decades ago, and recipients can lose benefits if they save a few thousand dollars or take part-time work.

Guaranteed Jobs Caring for Family (+44)

Some policymakers are proposing a jobs program that guarantees paid work caring for people — looking after young children, aging parents, and sick or disabled relatives. Under this policy, the government would pay people to do caregiving that families now do unpaid or can’t afford to hire out.

Make Big Corporations Pay a Minimum Tax (+43)

Some policymakers are proposing to require large, highly profitable corporations to pay a minimum amount of tax. Under this policy, big companies reporting large profits would owe at least a set minimum rate, even if deductions and credits would otherwise lower their tax bill.

Fund the IRS to Crack Down on Wealthy Tax Cheats (+43)

Some policymakers are proposing to fund the IRS so it can pursue wealthy individuals and corporations that don’t pay the taxes they legally owe. Under this policy, the agency would have the resources to audit complex high-end returns and collect unpaid taxes.

Universal Retirement Accounts (+42)

Some policymakers are proposing universal, portable retirement accounts that follow workers from job to job, with employer and government contributions, paying guaranteed income in retirement. Under this policy, everyone — including gig and part-time workers without an employer plan — would build retirement savings automatically.

A response scaled to the problem

If you go by raw popularity, the message you get from this list is that job retraining should be at the center of our response to AI disruption. Larger apprenticeship programs are the most popular single policy, with sector-based retraining programs meant to help people pivot into newly growing fields not far behind.

I have a few reservations, though. Alongside the policy polling, Blue Rose made short 20-60 second video ads, each focused on a different aspect of AI’s economic disruption, and tested them against three advocacy outcomes: whether voters see AI development as economically harmful, whether they want Congress to regulate AI, and whether they name AI as the more important issue facing the country.

The first two hardly need help. 70 percent already say AI development hurts the economy and 68 percent already want Congress to act. Only 13 percent rank AI as the more pressing issue. The ads moved concern about AI’s economic effects by 3 to 6 points and support for congressional action by 0.4 to 2.2 points. On salience they moved nothing: the best result in the field was 1.8 points, from the longest and most maximalist ad tested, and the apprenticeships ad moved it 0.3 points.

To be clear: job retraining, if it worked, and if there really are sectors where demand for workers is growing, is a fine policy to promote. But those are big “ifs.”

The federal government has funded workforce retraining programs for a very long time. The Wagner-Peyser Act in 1933, a key part of the New Deal, kicked efforts off; then there was the Manpower Development Training Act of 1962, the Comprehensive Employment and Training Act of 1973, the Job Training Partnership Act of 1982, the Workforce Investment Act of 1998, and finally the Workforce Innovation and Opportunity Act of 2014. A new bill in the litany, A Stronger Workforce for America Act of 2026, was marked up by the House Education and Labor Committee this past April; a version almost passed in late 2024 until Elon Musk and president-elect Trump killed the funding bill it was a part of. The programs these bills authorize are funded per the appropriations process, but generally get between $3-5 billion a year.

There is, correspondingly, a large literature on how effective this spending has been. David Roodman and Maxim Massenkoff at Anthropic just released an excellent evidence review summarizing what we’ve learned. The results are mixed. Overall, they conclude that job retraining programs subjected to randomized trials increased employment rates by 1.7 percentage points, against a baseline of 63 percent in the control group, and increase earnings by $800 per year on average

But when small programs are scaled up, they tend to lose effectiveness. The intensive Adult and Dislocated Worker programs first authorized under the 1998 law were subject to a large national randomized trial, which found no positive impacts after 30 months. A similar randomized evaluation began in the 1990s for Jobs Corps, a program more specifically targeting youth. Follow-ups after four years found gains to participants’ employment and earnings levels, but a longer-run review of study participants twenty years on found null results, with earnings gains fading away rapidly.

The best results have come from sectoral programs (backed by a margin of 60 points in our poll), which a recent evidence review led by Harvard’s Larry Katz estimates as leading to persistent 12-34 percent boosts to earnings. But as Roodman and Massenkoff note, these sectoral programs aren’t an across-the-board answer. They generally screen out the overwhelming majority, over 80 percent, of applicants, and last months rather than years. They aren’t capable of imparting skills that require longer to master. They also have a lot of variability in their effectiveness, and attempts to copy them have often ended in failure.

They also require an ability to identify and target jobs for which demand remains. We don’t know what, if any, such jobs will be left after the AI shock. “Sector-based training works when there is strong demand for jobs people can, and are willing to, step into,” the economist and AI analyst Molly Kinder noted recently. “The moment a shock destroys the demand, or scrambles which jobs are safe, or creates a clash with the jobs people truly want, the very mechanism that makes sectoral training work is the mechanism that breaks.”

I’d have no objection to job training programs being part of the government’s AI response — but there are major landmines here that we’ll need to avoid. We’ve learned it’s very easy to develop training programs with weak to no effects. My fear is that a hastily deployed expansion of these programs developed in the next couple of years might look more like the programs with dismal evaluation results than the carefully designed ones Katz evaluated.

Deploying ineffective or unpopular programs could be worse than doing nothing at all. The Comprehensive Employment and Training Act of 1973 (CETA) led to a wave of hundreds of thousands of CETA jobs that often went to ex-welfare recipients, which led CETA jobs to become demonized and unpopular much as welfare had been, and set the stage for the Reagan administration to wind the program down. That had a long-lasting, damaging effect on the US’s ability to do job training at scale. We can’t let a failure like that happen again.

More to the point, the size of the AI shock could be large enough that no realistically sized job retraining program is adequate to handle it. Our current job training infrastructure is sized to deal with hundreds of thousands of people, not millions, and certainly not tens of millions. This is fine if AI reduces employment by, say, 1 percent. But what if it reduces employment by 10 percent? What if there is a broad collapse in the return to cognitive labor, and workers from accountants to lawyers to engineers have to pivot into care work or other manual labor? There is no realistic job training system that could handle that kind of shock.

Then again, plenty of other popular policies in the above list seem inadequate to the moment too. Universal retirement accounts are a perennial proposal in the retirement policy world, but beyond serving as one form of capital redistribution, how exactly are they a specific response to AI? Why would broader employee ownership, regardless of its other virtues, help people whose entire companies are struggling due to AI?

The next stage of AI policy development should focus on identifying popular options that actually do show promise at dealing with sweeping job loss (things like required severance for AI job loss, or a jobs guarantee targeting care work) and developing them into viable plans capable of becoming law. For instance, how do we determine which job losses are due to AI? Should we require severance for all job loss at certain companies? How will we finance it?

These are answerable questions. But we have to keep our focus on ideas scaled to the problem, not familiar options from the past.

Subscribe now

CSAIP is working to close that gap. We sit at the intersection of public opinion research and policy design. Policy should respond to the concerns (and hopes!) Americans are expressing every day about how the economy should work in a potential era of major disruption. We partner with Blue Rose Research to continuously test how Americans are thinking about AI. And we help incubate responsive economic policies to ensure that the gains from AI flow to all Americans.

Importantly, while CSAIP publishes public opinion research, we do not simply endorse policies with the highest levels of support. Of course, the most popular policies may not be the best policies for responding to and redistributing the benefits of AI. Some policies that match the scale of the challenge face an uphill battle to earn support from the American public. Policymakers shouldn’t abandon these policies, but they should be honest about the political headwinds. We plan to bring together polling and rigorous policy analysis to identify solutions that are both popular and effective. That starts with knowing where people are today.

How do we measure public support for AI economic policies?

We’ve tested 79 policy ideas that could address the impact of AI on the economy. Our national survey of 56,000 respondents measured public support on 79 policies that range from AI-specific taxation to the creation of a U.S. sovereign wealth fund, from “data dividends” to wage insurance. Our survey results are available here. We went broad because all popularity is relative, so we need to understand where people are on a diverse battery of economic policies before we can say much about any specific intervention. We will update the rank order (relative popularity) of these policies over time and expand the economic policies we test as the field evolves.

Our surveys include pro and con arguments. Blue Rose Research measures how support changes after respondents read very short arguments for and against the policy. Net support reported in this post, and elsewhere in CSAIP research, reflects net support after argumentation. After reading the arguments, respondents are asked to choose whether they support or oppose the policy without an undecided option. Net support expresses the difference between support and opposition in percentage points. This helps us understand how fixed opinions already are on any given policy idea, and how vulnerable a policy might be to basic opposition. More information about the polling methodology at Blue Rose Research can be found here.

We test AI policies against generic economic proposals. To understand what the public thinks about AI specific economic policies, we’ve also included other economic policies in this battery as useful sign posts for baseline support. These baseline policies do not include specific language related to AI disruption.

Americans support most progressive economic policies addressing AI disruption

Our polling indicates that Americans are broadly supportive of economic policies addressing AI disruption. Of the 79 economic policies we tested, 61 hold net approval even after strong opposition arguments. The dozen policy ideas that top the chart all maintain an approval margin of +40pp in the face of opposition arguments. And 48 policies maintain a net approval of +15pp or more.

In particular, we see significant support for policies that are straightforwardly redistributive (meaning they transfer resources from corporations and the wealthiest Americans to support workers and families), and that put AI companies on the hook (meaning they specifically ascribe responsibility to AI companies for disruption they may cause). When we look at policies that directly mention AI, 31 of 48 economic policies had majority support, and 19 of 27 of the policies specifically mentioning AI economic disruption had majority support.

Our bottom line finding is this: Americans want an active government response to protect people from AI disruption, and they are open to a big range of tools to do this. Americans strongly support job retraining and compensating workers who are impacted by AI automation, strengthening the existing social safety net, and funding training, apprenticeship, and care work through progressive taxation schemes.

For this initial polling instrument, CSAIP focused on four major policy areas:

• Taxation and revenue

• Income support and safety nets

• Labor market and workforce development

• Ownership, governance, and stakeholder models

Labor market and workforce development policies had significantly more support after argumentation than the other three categories tested. Taxation and revenue policies and ownership governments and stakeholder models also provide the least familiar mechanisms for addressing AI economic disruption.

What this polling doesn’t tell us

We’re clear-eyed about what this polling cannot tell us:

Will this policy match the scale of the disruption we may face? A popular policy is not the same as an effective policy. This polling helps us understand where the public is already open to a policy idea, and where bold policy ideas will need strong organizing to gain traction.

What message will break through? How will public opinion shift over time? A one-sentence, pro-con argumentation in our polling provides an initial read of how AI economic policies stack up against each other, rather than messaging guidance for any specific policy. Think of this polling instrument as a jumping-off point for future polling and message testing which will give us a better sense of what kinds of arguments make these policies more or less compelling to Americans.

What’s next?

Policy should be guided by listening to the American public. CSAIP is committed to sharing public polling that helps us build economic solutions that meet the scale of the challenges we face, and we’re committed to sparking new policy development that harnesses public opinion to meet this moment. Explore our policy polling here, and join us by submitting your ideas.

Subscribe now

The safety net we have today won’t bail us out of widespread AI disruption

Even before AI began to raise the prospect of widescale economic displacement, our social safety net was failing to effectively deliver economic security.

Let’s look at unemployment insurance (UI), the stalwart New Deal program that forms an essential part of the social safety net. Even before AI, UI was in crisis. It had failed workers in the transition to a gig economy. The share of unemployed workers who actually receive UI has fallen to just one in four, driven in large part by the growing share of workers who are classified by their employers as independent contractors. Benefits are too low to keep people financially stable, and assistance is time-limited. The program is administered through a miserably complex patchwork of state laws that make the benefits extremely difficult to carry across state lines if you’re trying to relocate to find a new job. And the program basically denies help to any recent college graduate who cannot find work but also doesn’t have a work history.

Put simply, today’s UI is not capable of absorbing even a modest level of AI related job disruption. That’s ominous, because UI is ostensibly the most important safety net program for an era in which automation may swell the ranks of the unemployed, millions of Americans may need to shift careers, and young workers are already graduating into an economy where entry-level opportunities are shrinking.

But UI is not an outlier. Many of the most important safety net programs are built around the assumption that most Americans can find work that will sustain them. Medicaid and food stamps? Locked behind work requirements. Social Security? Tied to your lifetime wages. The social contract between the American welfare state and the American people assumes deregulated labor markets will expand prosperity, and provides inadequate help for the people who can’t cut it in the private market. But a safety net built around the primacy of work is set up to fail when the future of work itself is shifting under our feet. Our social safety net is profoundly ill-equipped to provide economic security if automation upends our working lives.

So we have a social safety net that is structurally misaligned with the direction our economy may be heading. The question for policymakers now is how do we harness a moment of rupture and outrage to finally transform the safety net for a 21st century economy?

AI economic policy should start with listening

This is where real-time, iterative public opinion testing becomes so valuable. Practically speaking, we need to be clear-eyed about which policies are popular enough to have a viable political pathway. But there’s a principled reason, too. As AI continues to tip the balance of power in our economy and democracy towards an emerging tech oligarchy, we should pay extra care to ground any AI policy agenda in the aspirations, fears, and opinions of the American public instead of the oligarchs’ vision of a post-AGI utopia.

CSAIP recently polled 56,000 Americans on their views on potential AI economic policies. You can access the data and learn more about our polling methods here. This poll is a treasure trove for policymakers: it tests baseline public opinion on 79 economic policy ideas for managing the AI transition. And it shows that public opinion is, once again, way ahead of the policy process.

What we’re hearing: Americans broadly support bold progressive policies to protect people from AI displacement

CSAIP’s data shows that the public is demanding a new economic vision for a world fundamentally transformed by AI. Americans overwhelmingly support dozens of policies to redistribute the gains of AI and to protect workers and families from economic displacement. The good news for policymakers is that Americans seem open to a wide range of policy tools: progressive taxation of AI profits and wealth, a stronger safety net to help workers impacted by automation, expanded investments in workforce development, and new ownership and governance models.

People want to be workers in the AI economy

Here’s what jumps out to me in this poll: The policies that appear to resonate most deeply with Americans speak to their aspirations to be workers and owners in an economy that AI is transforming, generally outperforming policies that are untethered to work.

The most popular policies in CSAIP’s poll center around protecting employment and ownership: expanding apprenticeships in new fields shaped by AI (+66pp), providing retraining or severance pay for automated-away jobs (+63pp), delivering sector-based job training in high-demand fields (+60pp), and creating employee stock ownership plans to give workers an ownership stake in the companies they work for (+50pp). In fact, the most popular policy in this survey expands apprenticeships in new fields that work alongside AI. So while Americans are anxious about the AI transition, they also want policies that deal them into a changing economy and ensure they’re not left behind.

By contrast, policies that expand access to benefits but don’t emphasize work were repeatedly underwater in this poll. A universal basic income policy ranks second to last at -33pp. Even a proposal to provide guaranteed income only to workers who have specifically lost a job to AI underperforms at -20pp. For those of us who believe expanding access to cash assistance is an important way to modernize the social safety net, these findings reflect a public skepticism we haven’t overcome yet.

The poll also raises big policy questions. We have been told a federal jobs program is politically impossible. That just doesn’t appear to be true. A proposal to create a modern version of a Works Progress Administration jobs program to employ workers who lost their job to AI scores an approval margin of +32pp. How can we design policy to match the public appetite for a program like this? At the same time, a proposal to expand paid apprenticeship programs scores an even higher approval margin of +66pp. These programs offer something similar to workers who are worried about being left behind in a changing economy. So why does one policy design command even higher public support?

An uphill battle for the boldest policies

Our poll found that strengthening programs that Americans are already familiar with sometimes outperforms novel policies that would establish new systems of redistribution. Support for policies that strengthen familiar programs like food stamps and disability insurance were more resilient against the opposition’s argument than proposals that ask Americans to trust something new, like a social wealth fund or AI profit windfall taxes.

One explanation for this dynamic is that trust in government is very low, and Americans may not believe that bold policies can actually be delivered well. Another explanation is that because the AI transition sparks deep anxieties about rapid change, programs that feel familiar and predictable may seem preferable to policies that themselves promise radical change.

But this dynamic should raise alarms for those of us who worry that the safety net we have today, even if improved, is not structurally up to the task of delivering security and prosperity. What this poll shows us is that there may be a mismatch between the baseline comfort that Americans feel with status quo programs (even programs that already underperform!) and the need for policies that are proportionate to the scale of the disruption we may face. How we avoid falling into a familiarity trap will be a key test for safety net redesign in an AI economy.

Sparking a new social contract

What an era of disruption calls for is not just pumping up the existing programs we have; AI may well demand that we reimagine the foundations of our social contract. In the most prosperous nation on earth, economic security should be a birthright, not conditional on the decisions corporations make about the future of work. And if the AI revolution does in fact usher in the wealth and productivity that its boosters promise, that should catalyze deep reinvestments in our social safety net so that this prosperity is shared. If this moment of transition is to become a moment of transformation, we’ll need an ongoing practice of listening and testing our ideas with the public.

Subscribe now

At CSAIP, we believe there is uncertainty around how precisely AI will impact the economy in terms of both timing and magnitude. We also concur with David Autor, the MIT economist quoted in the piece, who says CEOs “may have realized it was simply bad business to say that your great new product will destroy the economy.” Has anything in the last six months fundamentally altered the economics of AI-driven jobs displacement or is there simply a new narrative emerging?

Blue Rose Research, which advises CSAIP, has been tracking that public sentiment on AI, and its trackers point to a worsening scenario in how voters perceive and experience AI. Over the last six months the share of voters who personally know someone who lost a job or had work replaced by AI rose from 7 percent in mid-January to 9 percent by early July, a jump of roughly a third and the clearest single movement in the series. Concern about one's own job climbed in the same window, from 30 to 33 percent for the next ten years and from 21 to 23 percent for the next year, with the increase concentrated among the most concerned.

We also see shifts in public trust in AI. Optimism about AI’s future impact fell from 47 to 43 percent since late November. Belief that AI data centers benefit one’s community dropped from 30 to 25 percent. Topline trust in AI companies barely moved, but its composition changed. The share who trust these companies “not at all” rose from 27 to 30 percent as softer skeptics hardened, and only 3 percent now say they trust AI companies a lot. Just 13 percent say the benefits of AI outweigh the risks, against 39 percent who say the reverse.

Other polling shows that the public isn’t looking for doom narratives either. What the last six months show is a public whose direct experience and stated trust are both moving against the industry. The approaches that succeed will be those focused on concrete accountability over reassurance and abstract fear. Anyone proposing policy or trying to reset the story about AI and the economy will be working against that current rather than with it.

I became a CSAIP advisor because Americans seem so far ahead of the political system at this moment of technological change: they are demanding real civic imagination from leaders and policymakers. Americans have a long history of fighting for economic rights, social citizenship, and just entitlements. Together we created highly flawed yet remarkable government programs from Social Security to Obamacare, tax-and-spend programs that can create a temporary sense of security.

Still, I was most suspicious that Americans were genuinely open to new taxes to make these visions real. Would we be able to think about taxation differently in light of AI? The past decades of fights over the tax code have been downright depressing. I have been worrying at night about the feasibility of any effort to expand the necessary taxation to support the best parts of the welfare state. Just last summer, Congress locked in $3.8 trillion in extended tax cuts — and then turned around and used the resulting deficit as justification to cut Medicaid and food stamps.

So one of the first CSAIP polls pressure tests a core idea: Does AI create a genuine window of opportunity to organize around new taxation to support plans that address job market disruption?

The initial answer is strong yes, but with some caveats.

At baseline, support for AI-related taxes is robust. An energy tax on AI data centers gets overwhelming endorsement, with 76% of Americans in favor, resulting in a +64 percentage point net margin of support. A wealth tax on AI fortunes above $50M comes in at +53. A higher corporate tax rate on AI companies sits at +44. These are not subtle findings. Taxes are just not a hard sell. People want a fairer economy.

I was intrigued to see how specific AI-related taxes seem to be quite popular. Where Americans are already frustrated (especially around data centers) there is genuine enthusiasm for new taxes targeting the people and companies at the cutting edge of AI. A lot more research is needed to know how much we should extrapolate from the poor performance of the inheritance tax loophole.

Figure 1. Net support margin (support minus oppose) at baseline1

The caution: moveable opinion and many undecided on policy

Our partners at public opinion research firm Blue Rose Research have developed a thoughtful approach to measuring attitudes, which CSAIP will use often. Polling questions measure both a baseline support for a policy with a basic description, and then measure support for a policy after an interviewee has read strong arguments for and against any given proposal. This approach gauges if the opinions expressed are deeply felt, or weakly held and likely to crumble in the face of opposition.

When voters are exposed to generic pro-and-con argumentation for and against a policy, support drops on most proposals, sometimes substantially. Support for an AI data center energy tax falls from +64 to +36, and support for a wealth tax on AI fortunes drops from +53 to +24. Similarly, on nearly every proposal, roughly a quarter of respondents said they didn’t know. That uncertainty is consequential.

These realities are not shocking for nascent policy debates, and it is not necessarily bad news for tax enthusiasts. Americans are still forming their opinions, and taxes are not poison pills. In fact, support for new AI taxes is starting off from a point of strength. Still, proponents cannot assume that such strong baseline support for any new taxes will hold once opponents mount a campaign.

Figure 2. Net support margin at baseline and after generic pro-and-con argumentation2

AI-specific taxes versus broad tax reform: A generational fault line worth watching

There are two buckets of taxes that policymakers could advance in a world disrupted by AI deployment: AI-specific taxes (aimed at addressing the expected accumulation of wealth among the people and corporations that will profit from AI) and more general taxes (aimed at broadly capturing more of the wealth held by the largest corporations and the super-wealthy). There is an early policy debate occurring regarding which bucket of taxes make the most sense when it comes to raising revenue to address AI-driven job loss. CSAIP started by testing a number of AI-related taxes.

Beneath the aggregate support numbers, the data reveals a generational divide. All age groups agree that something needs to change, but start to diverge on what. When asked which tax philosophy best reflects their view, younger voters (18–34) are nearly as likely to favor targeted, AI-specific taxes (30%) as broad tax reform (32%), while older voters (65+) tilt heavily toward broad reform (47%) and are least likely to support AI-specific taxes (21%). Most importantly, the “status quo” camp is small across all ages.

Figure 3. Tax philosophy by age group — "When it comes to taxes and Artificial Intelligence (AI), which of the following comes closest to your view?"3

Younger Americans are clearly worried about a starved state. They are seeing AI wealth expand rapidly. Their openness to AI-specific taxes may reflect greater direct familiarity with the AI capabilities and the companies building it. For them, the question may not be whether the wealthy should pay more in general, but whether AI specifically has created a new kind of wealth that warrants its own form of responsibility to reinvest in society.

A window of opportunity

These results confirm that AI taxation is not a fringe position. It is where public opinion is already pointing. But “already pointing” is not the same as “already arrived.” I am excited to get into the nitty gritty of these policies and combine public opinion research with economic policy research to craft a new vision for progressive taxation in a world drastically changed by AI.

The first step of that work will come as a wide range of folks respond to the CSAIP RFI. We are hoping economists, policymakers, historians, and other creative thinkers will submit ideas for addressing key questions like: How would a token tax actually work? Is a value-added tax (VAT) possible in the United States? How would the IRS administer a robot tax? What would the impact of AI-specific taxes be on the development of AI in the United States and around the world?

The data center energy tax may be the most potent near-term entry point: highest baseline support, a concrete and visible target, easily tied to utility costs voters already feel. The token tax is getting more and more momentum. The robot tax (which would force companies using advanced AI to pay Social Security and Medicare taxes on the expense, the same way they would on a worker) sees its support fall the least after argument. It may be the most durable idea under pressure, grounded in Social Security and Medicare protections that voters across generations trust. But does the economics of a robot tax make just as much sense as the rhetoric of a robot tax? That’s exactly the type of issue CSAIP wants to explore.

The larger opportunity is not any single proposal; it is the chance to shape the terms of a debate that is still being written. CSAIP is committed to tracking this as it evolves, and to making research public. Join us by submitting your ideas.

Subscribe below to receive CSAIP research directly in your inbox.

Subscribe now

1

Figure 1 — Baseline Policy Support (AI tax proposals: support and opposition (N = 6,000)

2

Figure 2 — Policy Support After Pro and Con Argumentation (AI tax proposals: support and opposition (N = 6,000). Same proposals after respondents read generic arguments for and against each policy.)

3

Figure 3 — Tax Philosophy by Age Group (N = 2,753)

“When it comes to taxes and Artificial Intelligence (AI), which of the following comes closest to your view?”

• AI is creating so much new wealth that it needs its own targeted taxes on AI billionaires and AI corporations.

• AI doesn’t change the underlying problem — we should overhaul the entire tax code so everyone pays their fair share.

• AI is just another industry and the tax code should stay as it is.

• Not sure