You’ll Be Automated, But I’ll Keep My Job

There is a fascinating new survey out from the Pew Research Center on public perceptions in the US of the impact of automation on jobs. It expresses a “What, Me Worry?” attitude about the likelihood that jobs will be substantially changed by cognitive technologies. It’s yet another example of the “illusory superiority” bias—and another piece of evidence for why computers think more clearly than we do.

On the one hand, Americans seem to realize that automation will have a major impact on jobs. 65% of the respondents believe that robots and automation will “definitely” or “probably” do much of the work currently done by humans within 50 years. This seems to be a reasonable grasp of future reality.

On the other hand, however, 80% of these same respondents said that their own jobs or professions will “definitely” or “probably” exist in 50 years. In other words, they think that other people have to worry about automation, but their jobs are so difficult and complex that they could never be replaced by a machine. I am guessing that the same 80% of respondents feel that they are above-average performers in their jobs.

I heard about an example of this phenomenon this week. I had written an essay on the automation of marketing for a company I do a lot of work with. My sponsor for this work, who is in marketing herself, reviewed the ideas with other marketers in the company at a meeting. The consensus was that marketing automation is not ever going to happen—despite plenty of evidence that this phenomenon is already taking place at a substantial level in digital marketing, and is accelerating rapidly in other marketing domains.

I do happen to believe, as Julia Kirby and I discuss at length in a forthcoming new book called Only Humans Need Apply, (Pre-order with 1-Click!) that  there will be lots of good jobs over the next couple of decades (albeit probably not enough for everybody wanting one). But there is little doubt that they won’t be the same jobs that people toil in today. They’ll involve working closely with smart machines at every turn. Those smart machines will be capable of doing almost any single cognitive task better and faster than we can—in the same way that Google’s AlphaGo program just swept the world’s best Go player three games in a row.

But fifty years is a long time given the current rate of growth in cognitive technology capabilities. By 2066 I think it’s quite likely that we will have “artificial general intelligence,” in which computers and robots are better than us at pretty much everything. The “singularity” will have come to pass. That is likely to have an earth-shattering effect on labor markets. I will be dead by that time, but those who are alive and hoping to work will face some tremendous challenges at finding jobs.

We are in for some pretty scary changes over the coming years. More than anything we need to avoid a sense of complacency—“it can’t happen to me.” The humans who retain employment will be those who maintain a strong realism about the threat and the opportunity from smart machines, and adapt their own skills and objectives accordingly. It may make us feel better to whistle as we pass the graveyard, but it won’t keep us from inhabiting it someday.

Tom Davenport (@tdav), the author of several best-selling management books on analytics and big data, is the President’s Distinguished Professor of Information Technology and Management at Babson College, a Fellow of the MIT Initiative on the Digital Economy, co-founder of the International Institute for Analytics, and an independent senior advisor to Deloitte Analytics.