May 9, 2017 - Economy

Why it's dangerous to fear the robot revolution

AP Photo / Alastair Grant

Robert Atkinson, president of the Information Technology and Innovation Foundation, co-authored a new report published Monday that contrary to widespread popular opinion, the adoption of new technologies is displacing fewer workers than ever before in American history. The data he presents shows that:

  • Technology always directly kills more jobs than it creates, though it is doing so at a lower rate today than at any point since 1850.
  • The true engine of job creation is that technological displacement of labor makes those who still have a job richer, and therefore creates demand for more already existing occupations. For example, the societal wealth that is created by no longer needing movie projectionists has created the wealth to support many more teachers.

Why it matters: This study could cause policymakers to reevaluate the public fear of coming waves of automation, and instead focus on the threat of overreacting rather than underreacting to these changes. Axios talked with Atkinson to get his take on the implications of his research:

Were you surprised by your results?Yes, totally. Part of what we wanted to do was just look at the changing nature of the labor market. Were there a lot of blue-collar jobs being destroyed and when, and things like that. When we really started looking at the overall amount of churn and occupation loss, we said, "wow, that's amazing," because the narrative you hear all the time is that the period we're in is this unprecedented period of occupational disruption. I expected today's numbers to be on par with the past, but it's actually a lot lower, and that's a big surprise. What's the reason for this gap between perception and the evidence?The labor market is complex and opaque. People tend to think about technology and jobs from their own very direct, personal experiences. The major one they think about nowadays is Uber, and they see taxi drivers losing their jobs. What people forgotten is that if you go back to the 1970s, for example, there was this huge decline in farm workers. 400 or 500 thousands workers lost their jobs, telephone operators lost their jobs. There was this occupation called 'file clerks," that don't exist anymore. What is your opinion of economist Robert Gordon's thesis that the real danger is that technological innovation is actually on the decline, and that we should prepare for very slow economic growth?I agree with Gordon that we're currently in a period of low productivity growth, but I don't buy into this notion that we've run out of important innovations. Using the example of autonomous cars, I think the most disrupted job in coming years will be auto-body repairman, as that technology will dramatically reduce the number of accidents. But there are also many, many jobs that won't be destroyed in coming years. What the techno-optimists would say is that the economy of the future is fundamentally different because we are entering an age in which we are inventing technologies that will actually further their own development. Does this prospect mean we have to rethink our assumptions about technology, or are these people underestimating how transformative previous waves of innovation were?People really do forget how mind-blowing technologies of the past were. It's easy to be ahistorical and say that the things we're inventing are bigger than those of the first industrial revolution. People also tend to overestimate the reach of the technologies to come. I've talked with a lot of AI scientists, and one thing they all said to me is that it's not as if you are going to create machines that have anything close to human capabilities. It will have better than human capabilities in a lot of areas, but doing things like writing a complex legal brief, it. won't be able to do.Many of the jobs you argue will be irreplaceable by AI are those in the public sector like teachers, fireman, and police officers. Doesn't this mean that we'll have to become more comfortable with the public sector being a more important source of employment?We shouldn't set a goal of how much money we want to spend on education or safety. We should have goals in terms of output. And what people forget is that if we automate and become richer as a result—and we're able to distribute those gains equally—then people will have more disposable income and can tolerate higher taxes to pay for public goods.Is the media overhyping this issue?A lot of people are. The media is often a reflection of who's giving them stories, and a lot of people feeding stories are snake oil salesman. The more outrageous your prediction, the more likely you are to get quoted, so they put these very inflammatory predictions out there.What are the dangers of overreacting to this issue?Bill Gates recently argued that we should tax robots, and [Nobel Prize winning economist] Robert Shiller has argued that automation technology is like alcohol—it has negative externalities and therefore should be taxed. I could see just some guy saying that, but for a respected economist, that is a fundamental change in how economists have historically looked at technology. We've always looked at it as a net positive, and this attitude will be harmful to the economy.

Go deeper