business
technology

The Robot Revolution Is Here, But It Is Still Early Days

Economics, Oxford University
George Mason University
Genesis
Response
Penultimate
Finale

Robin Hanson

George Mason University

January 5th, 2020
Carl Frey says “robots are already here”, “impact on jobs and productivity will grow larger as their use spread” and “they have reduced employment and wages.” I don’t dispute such modest claims.
But surely Frey at least hints at larger claims when he compares robots to the industrial revolution, and says “robot revolution … is still early days” and “New technologies have always caused much hype which might have seemed excessive at the time but less so in retrospect”. That last claim sounds like Amara’s Law, “We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.” Yes, most tech that eventually gets big started small and grew slowly at first. But most hopeful new techs don’t actually cause much hype, and most that do never get big. Instead, most techs fail. So it is just not true that new techs “always caused much hype” or that hype usually seems “less so in retrospect.“ And regarding short versus long run change, in some recent Twitter polls I found respondents tending to say that they underestimated change over the last five years, but got it about right over the last thirty. Let me now state my position. The category of robots seems to me less interesting than the larger category of automation, most of which has long been computer-based. For over a century, we’ve seen a roughly thirty-year cycle of booms regarding media, pundit, and investor concern regarding automation. Over that period, progress in automation and other computer-aids has been far steadier, and has made large contributions to our relatively steady economic growth, which has long caused a relatively steady rate of large job disruptions. A continuation of these trends seems to me the obvious null hypothesis for the next few decades: we’ll see roughly the same rate of tech advances, job disruptions, and economic growth. Yes, many pundits predict large trend-deviating changes soon, just as they did in each prior boom. But expert retrospective views tend to see slower and steadier progress. As I discuss in my first book The Age of Em: Work, Love and Life when Robots Rule the Earth, we may eventually see huge trend-deviating tech advances and resulting social disruptions. But tech progress is usually smooth enough to offer large clear warnings well before such revolutions. Do we see such warnings now? No. I’ve just released a data analysis of automation levels in most U.S. jobs over the last two decades. We found that we can predict automation well, using mundane predictors which have not changed over this period. We also find that increases in job automation do not on average predict changes in wages or employment. This is consistent with prior literature, which does not find consistent effects, and sees them as varying with the elasticity of demand for particular jobs. So even if an automation revolution comes, it's not likely to cut jobs on average.
0 Comments