Will artificial intelligence lead to unemployment?

Ever since the Industrial Revolution, machines have replaced human labor, but some scholars believe artificial intelligence will have a much bigger impact on employment than 19th-century steam engines and textile mills ever did.

MIT economics professor David Autor argues that intelligent machines are “hollowing out” the economy, with new jobs coming at the bottom of the economic pyramid and the jobs in the middle being lost to automation and outsourcing to lower-wage countries. [Footnote 16]

And this trend, some analysts say, is only going to increase as machines become more sophisticated. “The economic impact will be huge,” said Carnegie Mellon's Mitchell. “We're at the beginning of a 10-year period where we're going to transition from computers that can't understand language to a point where computers can understand quite a bit about language.” [Footnote 17]

American software entrepreneur and author Martin Ford thinks the economic impact of intelligent machines is likely to be nothing short of disastrous, and not just for displaced workers. As companies turn to lower-cost machines to replace people, according to Ford, it could be the beginning of the end for traditional capitalism. “It's not just about unemployment,” Ford says. “It's about consumers. As people lose their incomes they can no longer go out and buy things.”

While most intelligent machines are being used to perform midlevel tasks, Ford says low-wage, unskilled employees also are highly vulnerable. It “doesn't mean that machines won't also figure out how to flip burgers,” he says. Ford argues that job losses throughout the economy will be so widespread that “ultimately the only solution will be some form of taxation and redistribution of income.”

Yet some economists argue that the negative economic effect of intelligent machines will be transitory and that their development can improve society. Robin Hanson, a professor of economics at George Mason University in Fairfax, Va., says that while machines eventually will be able to “substitute cheaply for … most human cognitive tasks,” the transition will have “a lot of overall positives.”

If intelligent machines are able to produce more at lower cost, he says, “total wealth goes up even if wages go down. This becomes a much more rich, powerful and capable society.”

The trick, says Hanson, is to get through the transition period. The tendency throughout history has been that successive technological innovations produce progressively less inequality, he argues. “Transition periods in and of themselves produce inequality,” but “the first societies to adopt new ways of doing things gain great advantages from that.”

Hanson expects a slow, predictable transition that causes less economic disruption than a rapid one. But other economists are skeptical.

“Artificial intelligence leading to gloom and doom? I think it's possible,” says MIT's Brynjolfsson. “My first instinct, as with most economists, is that historically every time technology has improved and eliminated some jobs, there have been a lot of other things that people can do and, as a consequence, the net effect has not been unemployment. But I don't think that's an iron law of economics. It's just the way it has worked so far.”

Brynjolfsson concludes: “It is possible that at some point we start having machine intelligence that replicates or, for that matter, surpasses enough of the tasks that humans are doing in enough areas that it starts becoming more and more difficult to find gainful employment for a lot of people.”

Click here for more information on the CQ Researcher report on "Artificial Intelligence" [subscription required] or purchase the PDF.



[16] Markoff, “Armies of Expensive Lawyers,” op. cit.

[17] Ibid.