So far users keep most of AI's value; workers' pay and jobs show no broad break, except for the young in the most exposed jobs, though some labour readings are moving
This layer is people: the workers whose tasks AI can do and the consumers who use it. Consumer surplus is the gap between what people would pay for something and what they are charged, the value they keep. Labour's share of income is the part of what businesses earn that goes to pay rather than profit. David Ricardo, an English economist who asked in 1817 why landlords grew rich when bread grew dear, found that the owner of the scarcest input collects a rent: profit above what it takes to keep a business going, earned because something holds rivals off. William Nordhaus, an economist who measured how much of the value of new inventions their makers keep as profit, found that it is a small part and that users keep most. Pascual Restrepo, an economist who models how automation changes wages, argues that once machines can do any task, labour's share of income falls towards nothing even as total wages rise. This site judges that for now consumers are Nordhaus's winners, and that workers are paid in wages only as long as their time stays a scarce input. For a layer of people, the direction word describes the split of AI's gains: holding means the split has not yet moved against them, with users keeping most of the value and workers' pay and jobs showing no broad break.
- Getting cheap and interchangeable
- This site judges that AI's answers are becoming a commodity to the people who use them, free or cheap and offered by many firms, which is why users keep most of their value. Nordhaus's finding is the account of it: competition and copying pass an invention's value to its users as lower prices and better products. For workers, the tasks AI can already do, such as drafting, summarising and routine coding, earn less as the machine's price falls.
- Staying scarce
- This site judges that what stays scarce is work where the person is part of what is bought (care, teaching, hospitality), work someone must answer for, and the skill to direct and check AI across many tasks. Hamilton Helmer, a strategy consultant who studied why some firms keep earning more than their investors could expect elsewhere, named the lasting advantages of firms, not people, so none of his powers applies here; the question is only whether workers' time stays scarce enough to be paid for.
- What would make it last
- Consumers keep their share only while AI firms compete; a few firms coming to own the assistants people rely on could let them raise prices and take back part of the surplus. Susan Athey and Fiona Scott Morton, economists who study competition, find that when AI is sold by firms with market power, the workers it displaces can lose twice: once when cheap AI cuts their pay, and again when its price later rises. Workers gain a durable share if AI makes their time more valuable than it makes it replaceable, which is what the rise or fall of relational jobs and of pay in exposed jobs would show.
- What would prove this wrong
- AI firms' profits growing as fast as the value users say they get; or labour's share of income falling steeply with employment in exposed jobs, across ages, falling well behind the rest while output rises.
- Binds on (how tight)
- Applications (no reading; a model judges it probably slack), Agents and tool use (easing)