Future of Work: One Person, 10,000 Hands

The future of work is how AI agents, robots and other technologies change what people do, how much they produce and who keeps the value. Our report finds the biggest change is that one person stops selling one person's work and starts directing the work of many: a heavy AI user already directs about 18 full-time people's worth of work for $180 a month.

For most people AI changes the job before it takes it. The first clear effect is fewer junior hires: employment of 22 to 25-year-olds in the most AI-exposed US jobs is about 19% behind less-exposed jobs. People who learn to set, check and stand behind AI work, a skill we call direction capacity, become more valuable, not less.

AI replaces tasks before whole jobs, starting with digital tasks a person can hand over, such as drafting, analysis and writing code. The IMF estimates about 40% of jobs worldwide are exposed to AI, about 60% in rich economies, and the World Economic Forum expects 170 million jobs created and 92 million lost by 2030.

AI agents are software that carries out tasks step by step on its own. They let one person direct the work of many: top coding-agent users already run 5 to 15 sessions at once. We give a 65% chance that the top tenth run 25 or more at once by 2030, and a 60% chance that agents finish 160-hour tasks half the time by the end of 2028.

An HLE is one adult's working year: about 190 kWh of useful work, or 2,500 accepted hours of work by a person, an AI agent or a robot. It lets us compare a Roman farm family, a steam-powered mill and a swarm of AI agents in one unit. Money is counted separately, in 2025 dollars.

By 2050 a skilled person directs about 120 to 6,100 HLE a year, with about 660 in our central case. In 2100 one skilled person could direct about 260 HLE after a setback, 700 in a long drift, 30,000 if agents spread widely or 2 million in an explosive window. We give no odds after 2050.

History says the owners of a new multiplier gain first: in Britain from 1780 to 1840 output per worker rose 46% while real wages rose 12%. If agents spread widely, output per person rises about a hundred times by 2100 but measured income only about fourteen times. Who keeps the gains is a choice, not a law of technology.

Five moves pay in every branch: measure delegated work, train people to direct and check it, protect junior jobs, own something scarce such as customers, trust, data or energy, and spread ownership of the multipliers.

The 311 Institute, founded by futurist Matthew Griffin, is a world-class team of future of work experts. It advises boards, governments and organisations including Walmart, the world's largest employer, and publishes dated, testable research such as this report on AI agents, robots and the future of work.

Yes. The research can be turned into a future of work keynote, board briefing or workshop built around your audience, your workforce and your plans.