When a mind can be copied, its wage falls toward the price of its computer

Today, if a company wants another excellent surgeon or engineer, it has to train one or lure one away from somewhere else, which takes years and costs a fortune, and that scarcity is why skilled people get paid well.

Hanson calls his brain copies "ems," short for emulations. "Scan a human brain, then run a model with the same connections on a fast computer, and you have a robot brain, but recognizably human," he writes, and "an existing em is easily copied into that hardware, resulting in a new em with exactly the same memories and mental habits" 1. If the best surgeon in the world can be copied a million times, the surgeon stops being scarce, and so does every other skill.

Hanson's conclusion is stark: "almost all wages fall to within roughly a factor of four of the cost to rent em hardware and supporting utilities," which he notes is the kind of "'Malthusian' wages" that "have described most humans before the industrial era" 1. In other words, a worker earns roughly what it costs to keep the worker running. He also expects an em economy to double in size every few weeks 1.

Nick Bostrom reached a similar conclusion in his book Superintelligence: human labor would earn "wages at around the price of the substitutes," while "capital (including AI) will earn all of the income, which will be a lot" 2.

If capital ends up earning nearly all the income, it helps to know where we start from. The International Labour Organization puts the share of global income going to workers at 52.6% in 2025, below its 2019 level of 53.0% 3.

Here's what surprised me: you may have heard that workers get about two-thirds, but that's a long-run average for rich countries 4, and even that has slipped. In the US, the share of business income going to workers hit "52.8 percent in the second quarter of 2026, the lowest level ever recorded," down from a post-war range of "60 to 66 percent" 5.

To be fair, economists don't all agree that copyable minds mean poverty for workers. Anton Korinek and Donghyun Suh modeled the transition to AI that can do human work and found it depends on one thing, which is whether there's an endless supply of harder tasks that only humans can do. "If the distribution of task complexity exhibits a sufficiently thick infinite tail, then there is always enough work for humans, and wages may rise forever. By contrast, if the complexity of tasks that humans can perform is bounded and full automation is reached, then wages collapse" 6.

Picture tasks as a ladder that gets harder with every rung. If the ladder goes up forever, people always have a higher rung to move to, and if it has a top, the machines eventually reach it. Korinek and Suh add that "declines may occur even before" full automation if automation outpaces the building of new machines 6.