You’ve probably already seen news stories floating around a couple weeks ago about how 47% of jobs are in danger of being automated. The stories are based on a report that looked at 702 different occupations and ranked them based on how likely they are to be automatable based on advances in machine learning, machine vision and robotics. I took a look through the report and thought I’d share some thoughts.
Caveats Methodology aside, the report only predicts that probability that a job could, eventually, be automated, not that it will be automated. More on that later, but for one thing it means they can’t predict how long it will take for a particular job to be displaced (they suggest it will take a decade or two, and the most automatable jobs will go first), or what percentage of people in a field will be replaced. Also, they don’t talk much about economics of replacement — whether it might be cheaper to pay humans than to buy and maintain robots for some positions. One thing I’m not sure about is whether the 47% is meant to apply to the total number of job types (if so, I’m not sure what the cut-off point is) or if it means 47% of all currently employed people are at risk of being replaced (which I think is what they actually mean).
Automation Winners and Losers According to the report the safest jobs, predictably, are in engineering, health care and creative work. Managers and supervisors are also pretty safe. Journalists are relatively safe, but my old profession — computer support — is in danger. The hardest hit will be the working class. Even skilled workers like plumbers, welders, machinists and truck drivers — the sorts of skilled workers there are reportedly shortages of — are in danger (electricians, however, are relatively safe).
Other Potential Losers It doesn’t address other supply and demand issues. For example, there are more law school graduates now than ever, so although most legal work can’t be automated, it doesn’t make law a “safe” profession. It also doesn’t address the effects of some portion of work becoming automated, thus reducing the number of people needed. To use law as an example again, software tools make the discovery process easier, reducing the number of lawyers and paralegals required to do that task. In journalism, some types of reporting have already been successfully automated. I’ve argued before that most types of writing and reporting will still need to be done by humans, but more sophisticated tools could reduce the total number of journalists required to run a profitable publication. Many types of professionals, including engineers and doctors — could be vulnerable to such disruption as well.
Social and Cultural Factors It also doesn’t address social or political trends that might protect some workers. It’s my understanding — though I could be wrong — that freight trains could operate with far fewer human workers than they do, but the union keeps humans in many roles. Likewise, unions or professional organizations could protect some careers, like taxi drivers and truckers, by pushing for legislation that requires a human operator ride along with self-driving vehicles “just in case.” Some workers, like food servers, bartenders and black jack dealers, may be preserved by cultural norms.
Doom But even if only half the jobs they believe are likely to be automatable are actually automated, that’s still about 23.5% of all types of jobs. That means things will get worse for everyone as A) more people will be competing for the jobs that are left and B) unemployed people will spend less money, reducing the demand for the products and services provided by the non-automated professions. And while the industrial revolution created many new types of jobs to help replace those displaced by machinery, there’s no guarantee that will happen again. Even if it does, it could take years for enough new jobs to emerge to replace the old ones.