Bill Gates has spent decades betting on technology to change the world. Now, he argues that AI should not be allowed to replace people in certain jobs, even when it can.
In an essay published Aug. 26 titled “The turbulent AI era is here. The choices we make now are critical,” Gates warns that the AI transition could become one of the most disruptive periods in modern economic history. He argues that automation could spread from software and office work into areas such as construction and hospitality as robotics improves.
His answer is not to stop AI outright, but to establish limits around some forms of automation before mass displacement becomes harder to manage. For employers, the proposal raises questions about which decisions and services should always retain human involvement, regardless of what AI can do.
Gates calls the idea “Human Reserved,” borrowing from the concept of nature reserves.
In his argument, society could decide that some work is worth preserving for humans even when machines can perform it. He also proposes taxing AI tokens and robots to reduce the financial incentive to replace workers as automation changes the labor market.
AI can do more, but Gates says that doesn’t mean it should
Gates’ concern is that this wave of automation may be different from earlier technological shifts that created new forms of work that still required human labor and cognition. Gates argues that AI can increasingly perform tasks that depend on human cognition, while advances in robotics could extend that pressure into physical work as well.
That is where his “Human Reserved” idea comes in.
Gates proposes that society deliberately keep some jobs for humans even when AI becomes capable of performing them, much like a nature reserve protects land that could otherwise be developed. He points to childcare, jury service, education and health care as areas where society might decide to preserve meaningful human involvement.
The reasoning is not simply that humans will always perform better. Gates argues that some work carries a social or human value that efficiency alone cannot measure.
That distinction also explains why Gates is not presenting AI as inherently harmful. He still sees the technology as a powerful tool for improving lives, but argues that whether it reduces or deepens inequality will depend on how the transition is managed.
More must-read AI coverage
- SS&C Intralinks DealCentre AI vs. Datasite: Which platform is built for the future of dealmaking?
- SS&C Intralinks FundCentre AI vs. Juniper Square: Which platform better supports modern private markets fund managers?
- Why Data, Not Models, Determines AI Success
- The Rise of the AI-Native Factory: How Physical AI Is Transforming Manufacturing
Gates wants to make replacing workers with AI less attractive
Gates’ answer to AI-driven job displacement isn’t to stop companies from automating, but to change the financial incentives behind it.
He argues that existing tax systems can make automation financially attractive because employing people creates payroll-tax costs, while investments in AI and robotics may qualify for tax deductions.
Gates therefore raises the possibility of taxing AI tokens and robots to narrow that difference. The revenue could potentially support workers affected by automation, although his essay does not lay out a detailed tax rate or funding mechanism.
Without changes to the incentives, he argues, the more capable AI becomes, the greater the incentive for companies to replace workers with AI.
Some big questions remain unanswered
Gates’ proposal leaves a bigger question than which jobs should stay human: who gets to decide where AI stops?
Governments are the most obvious candidates, although Gates does not provide a detailed process for selecting protected jobs or enforcing the boundaries.
But the decision could also involve independent regulators, companies setting their own limits, industry or worker advocates, or even consumers rejecting AI in roles where they want a human involved.
Governments could establish enforceable standards, but achieving cross-border consistency may be difficult. Regulators could set sector-specific rules; companies could voluntarily offer human-only services; and consumers could influence adoption by what they choose to pay for.
That matters because the answer could directly shape how people work and use essential services. If society decides some jobs should remain human, someone still has to draw that line — and enforce it.
For now, Gates has proposed a principle rather than a workable regulatory system. Employers do not need to change their automation plans because of it, but they can already identify which decisions require human accountability, where AI should assist rather than replace workers, and where automation could create legal, safety or reputational risks.