Artificial intelligence threatens to cause a jobs shock across offices, warehouses, call centers, clinics, classrooms and more. Faced with backlash among displaced workers, the industry and its lobbyists will likely lean on cash‑transfer schemes — universal basic income and its like — as a release valve.
That path would paper over dislocation while entrenching gatekeepers. It would weaken bargaining power at work and create new and unintended social and fiscal dependencies.
We can do better than that. We have an opportunity instead to bend the trajectory, level the playing field and ensure that the public that pays for the externalities of AI also benefits from its returns.
What does this entail? Open and contestable markets; modern data rights; liability that governs control; guard-railing technological change to serve the shared good; and most importantly, public sharing in AI’s upside.
The U.S. should establish a professionally managed, firewall‑insulated Public AI Wealth Fund that holds diversified stakes across the AI stack and distributes an annual dividend tied to the AI upside (i.e. realized returns).
This would put the focus on public dividends, not basic income; it would give the public a meaningful, lasting stake in AI’s returns, while using competition, data, labor and energy rules to keep the field open and the harms in check.
The deal between Nvidia and AMD and the U.S government, which involves payments of 15 percent of revenue from sales of some advanced chips, provides precisely the platform to kick-start the creation of this Public AI Wealth Fund.
Capitalize it with equity warrants from firms receiving large subsidies, tax credits or preferential contracts; a compute excise above thresholds set by rulemaking; a windfall‑profits surtax on extraordinary AI margins; and location‑based fees where data centers draw on subsidized energy or water.
Adopt a simple fiscal rule: preserve principal and pay a universal dividend from realized returns. Dividends will start off modest and grow as the fund matures, aligning public returns with performance rather than with permanent deficit finance.
Sharing in the upside should complement — not replace — open markets and individual economic choice.
Congress should treat exclusive deals as presumptively anticompetitive once market‑share triggers are met; bar self‑preferencing in AI marketplaces; and scrutinize “model‑plus‑platform” tie‑ups as potential vertical restraints. Interoperability and portability — APIs, export formats, and the ability to move fine‑tunes — should be the default, with clear timelines and contractual exit rights. These rules ensure the public is not forced to buy into entrenched barriers.
Individuals need a clear opt‑out for the use of personal data in training, a right to know whether their data was used, and remedies for unauthorized use. For high‑risk uses — hiring, housing, health care, finance and critical infrastructure — deployers should bear a duty of care: documented impact assessments, human‑in‑the‑loop plans, incident reporting, and audit trails. The burden should rest on those who profit from deployment to show they have met these duties.
The Federal Trade Commission can scrutinize deceptive AI claims, dark patterns and data misuse. The Justice Department and the FTC can challenge exclusionary bundles and exclusivity in contracts. The Securities and Exchange Commission can require disclosure of material workforce impacts from AI deployments, not just AI “strategies.” Civil‑rights agencies can enforce explainability, auditing and appeal rights in automated decisions. Energy and water regulators can require facility‑level reporting from hyperscale data centers and condition interconnects on demand‑response and conservation plans.
The executive branch and states should move in parallel. States should seed their own AI wealth funds with data‑center and related revenues and equity warrants tied to subsidized projects, coordinated with — but independent from — the national fund. A federated research cloud across universities, national labs and state innovation hubs would give academia and students access to shared AI infrastructure without having to depend on vendor credits.
Ask schools, employers and hospitals which systems they use, how they were evaluated, and how you can appeal decisions. Insist that public dollars tied to AI come with ownership, transparency and exit rights.
Talk to your elected officials and those that are running for office; tell them it is time for our nation to build a Public AI Wealth Fund.
John deVadoss was a general manager at Microsoft for two decades. He did his Ph.D. work in AI, at the University of Massachusetts at Amherst, specializing in Machine Learning.
