Bill Gates Calls for Taxes on AI Tokens and Robots, With Jobs Reserved for Humans

Bill Gates is proposing a two-part intervention in AI-driven automation: tax the tools that replace workers, including AI tokens and robots, and reserve some work for humans. The plan aims to slow displacement and finance retraining, but it turns a familiar warning about labor into a difficult policy-design fight over which technologies and jobs would be covered.

Gates set out the ideas in an essay on Gates Notes. He argues that employers pay payroll taxes when they hire people, while they can generally write off robots as business expenses. In his view, that difference encourages companies to substitute machines for workers; a tax could temper that incentive and generate money for retraining and a stronger social safety net.

The latest version is broader than Gates’ earlier robot-tax argument. He first floated a robot tax in a 2017 interview, and now calls for taxes on both robots and AI tokens, the units used to measure model processing. He has framed the goal as slowing the replacement of human labor rather than stopping automation altogether.

A boundary around selected work

His second idea is a “Human Reserved” category for jobs or tasks where AI use would be restricted or barred. Gates says a designation could respond to economic disruption when many workers cannot easily switch occupations. It could also protect human involvement in sensitive encounters, such as telling a patient they have an incurable disease.

The category would not necessarily freeze those roles permanently. Gates suggests some jobs could initially be set aside, then receive AI gradually over years or decades while preserving some human employment. That makes the proposal a pacing mechanism as much as a ban: policymakers would decide where adoption moves slowly and where it does not.

The productivity objection

The International Federation of Robotics rejects the tax approach. It says taxing production tools would make technology investment more expensive and hamper productivity and competitiveness. The group has also argued that automation creates jobs through higher productivity, directly challenging Gates’ premise that a tax is needed to manage labor losses.

The rules are still missing

There is precedent, but not a settled template. The European Parliament rejected a robot tax in 2017 while calling for a framework to regulate robotics. South Korea’s 2018 law is described as the only law of its kind, and the United States has no national robotics strategy.

Gates’ proposal therefore remains a direction rather than an operating policy. The unanswered question is who would set the tax and Human Reserved rules, and what those rules would require. The answer would determine whether the idea becomes a narrow adjustment to automation incentives or a far more consequential limit on where AI can work.