Human Reserved is Bill Gates's term for work deliberately set aside for people only. He introduced it in an essay published on 26 August 2026, and he says plainly that the name is his: he has started calling this domain Human Reserved. The analogy he reaches for is a nature reserve, ground that could carry buildings and roads and is left alone because the loss would be too great. It is the first serious proposal from a major technology figure that the boundary of automation should be decided rather than discovered, and that alone makes it worth defining carefully.
The answer, in one line
Human Reserved is Bill Gates's term for work deliberately set aside for people only. He introduces it in an essay published on 26 August 2026, writing that as AI and robots improve, we will set aside certain things for only people to do.
What the essay actually says#
The idea arrives late in a long essay about the risks of the AI transition, and it arrives through a story. Gates's father died of Alzheimer's in 2020. In the later stages he was cared for by paid carers who, in Gates's account, understood him even when he could not express himself, and knew he was hungry when he could not say so. He writes that something in that care was irreplaceably human, and that no robot could or should have done it.
That word, should, is doing the work. The argument is not that machines will fail at care. It is that some things ought to remain with people whether or not a machine could manage them. He makes the same move in health, imagining a robot delivering a diagnosis of an incurable disease and observing that there is no technical reason it could not, and that it should not.
He gives a second reason, economic and quite different in character, and in the essay it actually arrives first, immediately after the coinage. A role may be worth reserving because automating it would displace a large number of people who cannot easily move. His example is a 55-year-old who has worked in construction their whole career, who cannot reasonably be told to go and work in an elder care facility and expected to find it fulfilling. Reserving work, on this reading, buys time rather than protecting something sacred, and Gates suggests some reservations should be temporary, phasing AI in over years or decades.
What Gates did not say#
The essay names no figure for how much work should be reserved. It does not mention jury service. It does not mention childcare. Those all come from an interview Gates gave to Ina Fried at Axios, published the same day, and they were widely reported as though they were the proposal itself. The 40 per cent is his ceiling rather than his target, and he says so: in a very extreme form of it he could imagine 40 per cent of jobs reserved initially, but that is as high as he can get.
The distinction matters for anyone citing this. The interview supplies a ceiling of 40 per cent of jobs and two concrete examples. The essay supplies a principle, one illustration from the author's own family, and an admission that the hard questions are unresolved. A board paper quoting 40 per cent as Gates's written position is citing the coverage rather than the source.
The questions he says he cannot answer#
Gates lists four and claims none of them. Who decides what gets reserved. What criteria apply. How companies are stopped from using robots anyway. What happens to international trade when one country lets machines make something and another does not. He says these will need to be worked out in public as part of a transition plan.
He also expects the boundary to move by geography rather than settle globally. Some countries might insist that people care for the elderly. Japan, with a shrinking workforce and too few young people to do that work, may welcome a caregiving robot. The same task, reserved in one place and automated in another, on grounds that are demographic rather than moral.
Why it belongs in this research#
Capability debt accrues when the boundary of automation moves by default. Nobody decides to hand a task over. It becomes cheap, the reason not to disappears, and the decision is taken by arithmetic and noticed later. Human Reserved is the opposite procedure: name the boundary in advance, and defend it.
Gates pairs the idea with a tax proposal that makes the mechanism explicit. As things stand, an employer who hires a person pays payroll tax on their earnings, while an employer who buys a robot usually writes it off as a business expense straight away. He argues for taxing AI tokens and robots, on the grounds that the current system nudges employers towards replacement. That is drift described as a tax incentive.
The organisational version of the same instinct already exists in consulting practice. BCG uses the phrase AI-off zones for tasks a company designates as off limits because originality, ethical judgement or synthesis matter most there. Human Reserved is that idea at the scale of an economy, with the state rather than a management team drawing the line.
What the term does not settle#
Reserving work protects the job. It does not by itself protect the capability. A role can be formally reserved for people and still be hollowed out from inside, if the person in it spends their day approving machine output rather than exercising the judgement the reservation was meant to preserve. The reservation defines who holds the task. It says nothing about whether they are still practising the skill.
That gap is where this research sits. A policy that names protected work without asking how competence inside it is maintained will produce protected roles occupied by people who can no longer do them unaided, which is a worse outcome than either automating the work or leaving it alone.
Key sources
- Gates, B. (2026). The turbulent AI era is here. The choices we make now are critical. Gates Notes, 26 August 2026. The essay in which the term is coined.
- Fried, I. (2026). Bill Gates wants to keep some jobs off-limits to AI. Axios, 26 August 2026. The interview, published the same day, which is the only source for the 40 per cent ceiling, jury service and child care.
Explainer · SS-2026-183 · Graded against the published rubric
Hirji, R. (2026). Human Reserved. The SuperSkills evidence base, SS-2026-183. https://thesuperskills.com/research/what-is-human-reserved. Last reviewed 5 September 2026.
An evidence review by Rahim Hirji, not peer-reviewed research. For a material claim, cite the underlying study as well; every study here carries its own permanent link.
How citations and IDs work