Bill Gates Wants Jobs Set Aside for Humans, Like Nature Reserves

Bill Gates has called for "human-reserved" jobs in certain sectors to prevent AI replacing them, and expressed concern that governments are not prepared for the technology's impact.

The argument appears in a 6,000-word essay titled The turbulent AI era is here. The choices we make are critical — his first substantial piece on AI in three years.

The proposal

"Many jobs will disappear forever," Gates writes. "I believe that as AI and robots improve, we'll set aside certain things for only people to do."

He likens the idea to conservation. "I've started calling this domain human reserved. I like the phrase human reserved because it makes me think of nature reserves — places where we could put" boundaries around what technology is permitted to displace.

Gates also questions whether the institutions required to implement any of this exist, writing that "it is fair to wonder whether the world's institutions are up to the task of designing and implementing this new architecture."

Gates wrote that 'it is fair to wonder whether the world's institutions are up to the task of designing and implementing this new architecture'. Photograph: Denis Balibouse/Reuters

The metaphor, examined

The nature reserve comparison is more revealing than it may be intended to be.

A nature reserve is what remains after a species has lost the rest of its habitat. It is a managed remnant, defined by boundaries drawn by the same forces that destroyed everything outside them, and its inhabitants persist at the discretion of whoever maintains the fence.

Applied to employment, the analogy concedes that the default outcome is displacement and that the preserved category is an exception granted rather than a position held. That is a considerably bleaker proposition than the reassuring framing suggests.

The question it leaves open

Gates does not specify which jobs would be reserved, and that is where the difficulty lies.

Protecting work that AI cannot yet do requires no policy — the market handles it. Protecting work that AI can do, because society decides a human should do it anyway, requires deciding which occupations qualify, enforcing the restriction against employers with strong incentives to evade it, and accepting a higher cost for the affected service.

Care, teaching, judging and nursing are the categories usually proposed, on the grounds that the human presence is part of the service rather than a means of delivering it. That is a defensible principle and an extremely hard rule to draft.

Coming from Microsoft

The essay arrives from the co-founder of a company that has invested more heavily in AI than almost any other, which invites the obvious observation.

It also makes the intervention harder to dismiss. Gates is not forecasting displacement from outside the industry; he is describing what he expects the technology his company is building to do, and arguing that governments should constrain it. The gap he identifies — between institutional capacity and the pace of deployment — is one he is in an unusually good position to measure.

The alternatives on the table

Human-reserved work is one of several proposals circulating, and it is worth seeing where it sits.

Universal basic income accepts displacement and compensates for it. Retraining programmes assume displaced workers move to new sectors, which historical evidence supports over decades and not over the careers of the people actually displaced. Robot taxes attempt to slow adoption by pricing it.

Gates's proposal is distinct in preserving the work itself rather than the income — an argument that employment provides structure, status and purpose that a transfer payment does not replace.

The political feasibility

The essay's most defensible claim may be its most pessimistic one: that institutions are not equipped for this.

Designing reserved categories requires legislation, enforcement and international coordination, since a country that protects an occupation unilaterally simply exports it. Nothing in the record of the past two decades suggests that machinery can be assembled at the speed the technology is moving.