(Bloomberg) — Don’t think a robot could take your job? A study in the latest issue of McKinsey Quarterly says as many as 45 percent of the tasks that Americans get paid to do could be automated. The share rises to 58 percent if computers get as good as the median human being at speech recognition and understanding.
The study’s most depressing finding is that a lot of what people do at work is rote. Just 4 percent of Americans’ work activities require creativity at the level a normal person has, and just 29 percent require sensing emotion, the authors conclude. (Though it isn’t clear the humans are sensing emotion like crazy, either.)
The study differs from earlier ones on automation by focusing on activities, not occupations. Any given occupation, from landscaper to chief executive officer, involves a set of activities, some of which can be automated and some of which can’t, the authors say. “Very few occupations will be automated in their entirety in the near or medium term,” they write. “Rather, certain activitiesare more likely to be automated.”
Automation, in other words, will transform occupations, with machines taking over what they do best and people doing more of what they do best, the study says. That sounds like an opportunity for McKinsey & Co., whose consultants presumably would be happy to help companies restructure jobs. The report was written by Michael Chui, a principal at the McKinsey Global Institute, the consulting firm’s think tank; James Manyika, a director of the institute; and Mehdi Miremadi, a principal in McKinsey’s Chicago office.
When Chui went to his Stanford University reunion recently and talked about his findings, he said, “some friends from my freshman dorm reminded me that I was voted the person most likely to be replaced by a robot of his own making.”
Automation isn’t all about cutting costs; it can also be about improving quality and performing tasks at “superhuman” levels, the article says. About 60 percent of occupations could have 30 percent or more of their constituent activities automated, the study calculates.