Researchers simulated an economy featuring two types of workers high-tech employees who produce new software code, and low-tech workers who produce human services (such as artists, teachers priests etc).
Over time robots “can leave all future high-tech workers and, potentially, all future low-tech workers worse off”, the researchers wrote.
“In short, when smart machines replace people, they eventually bite the hands of those that finance them,” they wrote.
At first, high demand for code-writing high-tech employees increases their wages. However, over time, the amount of legacy code grows.
As this happens, and as some smart machines become better able to learn tasks, writing new code becomes redundant, the authors said.
Demand for code-writing high-tech workers then becomes limited to those who are needed for general code maintenance like updates and repairs.
The rest of the high-tech workers end up going into the service sector, which consequently pushes down wages for employees in that industry.
And lower incomes reduce the amount of goods and services that workers are able to buy.
The model also predicts a long-run decline in labour’s share of income, which has been underway in the US since the 2000s, as well as “a growing dependency of current output on past software investment.”
“In the absence of appropriate fiscal policy that redistributes from winners to losers, smart machines can mean long-term misery for all,” the researchers, from the Boston University and Columbia University, concluded.