AI and automation: should we be concerned?

Over the past couple of decades, many countries have undergone what many call a fourth industrial revolution. New technologies across fields ranging from music to transportation have changed our world forever. Specifically, Artificial Intelligence has seeped into every aspect of our lives. Whether it be the YouTube algorithm deciding which video to recommend next or Google Maps finding the fastest route from home to work, Artificial Intelligence has made many parts of our lives better and more convenient.

But there has been a cost to these new developments. Across the nation, hardworking men and women have lost their jobs and been replaced by the very technologies that make our lives so much more luxurious. These changes in our economy and society have lead many to speculate that future job loss will only increase, and that we should prepare for these changes. But exactly how much of an effect will Artificial Intelligence have on the job market in the coming years, and should we be concerned?

In 2013, two Oxford researchers, Carl Frey and and Michael Osborne, attempted to answer this very question. Frey and Osborne looked at 70 different jobs and classified them as either automatable or not automatable, before feeding these classifications to a basic logistic regression model which — with the help of O*NET data — then predicted the probability of 630 other jobs being automated. Frey and Osborne’s final conclusion was that 47% of all jobs were at risk of being automated away.

While 47% seems like an almost apocalyptically high number, there are many factors Frey and Osborne failed to consider. First of all, their ‘training data’ (the 70 jobs they labelled themselves) classified jobs in a binary fashion, either automatable or not, with no room for any middle ground. The researchers did this for jobs in which they were entirely confident about the automatability, as to reduce bias in the training set. However, by doing so, they almost did the opposite, as only their most strong opinions were included in the training data.

Secondly, the study neglects the possibility of the creation of more jobs. This has been one of the largest critiques against the argument that we should prepare for coming automation, as throughout history, multiple industrial revolutions have occurred, with none of them resulting in mass unemployment. And some say that this time is no different; for example, economist James Bessen recounts a case study in which automation actually increased the demand for jobs:

“The average bank branch in an urban area required about 21 tellers. That was cut because of the ATM machine to about 13 tellers. But that meant it was cheaper to operate a branch… And when it became cheaper to do so, demand for branch offices increased. And as a result, demand for bank tellers increased.”

Clearly an or industry being automated does not mean it’s jobs will be lost, as Bessen explains.

And this isn’t all that Frey and Osborne forget to account for. What they forget to account for is the fact that many jobs only have specific tasks within them get automated away, instead of the entire job. For example, new developments in computers’ abilities to read MRIs certainly doesn’t make the job of doctor obsolete, but the specific task of predicting pneumonia from an MRI might shift to computers. Another study tried to account for this more nuanced view of automation, done by economists Melanie Arntz, Terry Gregory, and Ulrich Zierahn. By focusing on the tasks within certain jobs, they arrived at the conclusion that only 9% of jobs were in danger of being lost to automation — significantly less than the 47% figure reached by Frey and Osborne.

Arntz, Gregory, and Zierahn weren’t the only ones to reach their own number, however. PwC, an accounting firm, found that 38% of jobs were susceptible to automation, and McKinsey, a managing consultant company, used its own method to conclude that between 16 million and 54 million jobs will be lost between 2016 — quite the range. The OECD decided to join in on the fun and found that the median job has a 48% chance of being automated.

If you think these estimates seem to be a little all over the place, you’d be correct. There are simply too many factors to accurately predict what the effect of Artificial Intelligence will be on the future job market. But the fact we don’t know the extent to which Artificial Intelligence will affect the market is precisely why we should be scared. Yes, it is possible that in 20 or 30 years we will look back and realize that we were being overly cautious, but the possibility that millions will find themselves without a proper way to support themselves and their families is a risk too great to ignore. We as a society must take immediate steps to ensure that we can maintain a functioning society and market that works for everyone as our world undergoes unprecedented changes in the next decade.



A senior at Garfield High School in Seattle, Washington.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store