The Quarterly: Jeremy, how big a deal are machine-learning algorithms for employment and the workforce? And what should we do about it?
Jeremy Howard: I think it is important to think about the policy implications here. Government leaders need to be aware that, right now, computers are as good as or better than humans at most of the tasks people involved in information-processing jobs do. That is 65 percent of the American workforce. So is this wonderful or is this a tragedy? It actually depends entirely on how governments respond. Scenario number one is a disparity in economic power, in which the folks with the data and the algorithms have—and add all of—the economic value, and the rest of the workforce adds little or none.
That scenario could create an awful social disruption. Scenario number two is to accept that in this new world, there’s a large group of people who can’t really add economic value anymore, but that doesn’t mean they don’t get to live a decent human life. So we have to start thinking about the policy implications—like a basic living wage, which Germany will be introducing, or a negative income tax, which has been off the agenda for decades but deserves to be back on it. I think people should start to think about these policy implications because the point at which we need to make decisions will be upon us suddenly.