Artificial intelligence is making some workers faster at their jobs, and the question in Washington is who keeps the gain. A Politico opinion piece argues that the productivity boom from AI needs a worker dividend, a way of routing part of the value created by the technology back to the people whose work it changes. The AI worker dividend is not a single policy, and its supporters describe several versions of it, from higher pay and shorter hours to direct payments funded by taxes on AI use.
What an AI worker dividend would do
The basic idea is that AI raises output per worker, and that the extra output now flows mostly to company owners and shareholders. An AI worker dividend would redirect part of it. OpenAI set out one version in a 13-page policy document released on April 6, 2026, titled “Industrial Policy for the Intelligence Age: Ideas to Keep People First.” It proposed a levy on automated labour, a public wealth fund that would hold equity in AI companies and pay annual returns to citizens, and government-backed trials of a four-day workweek at full pay.
The wealth fund in that plan is modelled on the Alaska Permanent Fund, which has paid state residents a yearly dividend from oil revenues for more than four decades. OpenAI chief executive Sam Altman has said publicly for years that AI will be deflationary, labour-replacing and wealth-concentrating unless policymakers intervene.
New York lawmaker Alex Bores has put forward a related proposal for an AI dividend, funded by a tax on commercial AI use measured in tokens, by equity stakes in leading AI firms and by changes to how labour and capital are taxed. Bores has said the programme would switch on when clear triggers are met, such as sustained falls in labour force participation or wage compression in affected sectors. He previously pushed through the RAISE Act, New York’s AI safety law. Bores calls the AI worker dividend insurance for the workforce.
The evidence behind the AI worker dividend
A 2025 study by Erik Brynjolfsson, Danielle Li and Lindsey Raymond followed 5,172 customer-support agents and found that AI assistance raised the number of issues resolved per hour by 15 percent on average, with gains of about 30 percent for less experienced and lower-skilled workers. The study did not establish economy-wide wage or employment effects, and the authors were careful to say so. It is the clearest number behind the AI worker dividend case.
A Goldman Sachs report cited by Fortune estimated that AI had already contributed to the loss of around 16,000 jobs a month over the past year. A July 2026 working paper from the International Monetary Fund put the labour-cost equivalent of time currently saved by AI at about 2.7 trillion dollars a year, or 3.4 percent of GDP across the countries studied, and found the gains concentrated in higher-paid occupations. That gap between output gained and wages paid is the case for an AI worker dividend.
The International Labour Organization’s 2025 assessment found that one in four workers is in an occupation with some exposure to generative AI, while only 3.3 percent of global employment sits in its highest exposure category. Its conclusion was that job transformation is more likely than complete job elimination, and that sharing the gains should not wait for mass unemployment.
Why the AI worker dividend faces long odds
None of these proposals is law. Congress has shown little appetite for new taxes on AI, and OpenAI’s own document is a recommendation, not legislation. A July 2026 framework from the pro-human.co group argued for capturing part of the capital income and rents created by AI, the funding base for an AI worker dividend, paying some of it out at once through existing tax and benefit systems and investing the rest in a permanent citizens’ wealth fund.
Other writers have proposed a Pigovian automation tax, named for the economist Arthur Pigou, which would charge companies for the demand destroyed when workers are displaced, with lower bills for firms that instead cut hours, lower prices or pay displaced workers. The paper behind that idea, “The AI Layoff Trap” by Brett Hemenway Falk of the University of Pennsylvania and Gerry Tsoukalas of Boston University, argues that firms face an automation arms race they cannot stop on their own.
The argument behind every version of the AI worker dividend is the same: productivity gains from AI are real, they are measurable in hours saved and output raised, and the institutions that decide who keeps them have not kept up. Whether that leads to a dividend, shorter hours or higher pay will depend on choices made in Congress and in state capitals, not on the technology itself.