Will Robots Steal our Jobs?

In recent years, the idea has spread that a jobless future is lurking just around the corner. Here on Youtube, the famous Youtuber CGP Grey has a good video summarizing the matter called ‘Humans Need not Apply,’ which we’ll link to in the video description below.

The argument is that the combination of new machinery and robots, as well as the possible advert of Artificial Intelligence will result in mass automation, which in turn will lead to a global dearth of jobs. With self-driving cars and automated software writing our newspapers, grading papers, and processing forms, this all seems very plausible. But the matter isn’t as clear-cut as a quick glance might have one believe.

First, it is nothing new that the invention of new technologies leads to the destruction of jobs. When cotton came onto the European markets, replacing clothes made of more labour-intensive materials such as flax, people angsted in the same way. And when, later, the spinning wheel was invented, people angsted again. When computers became widespread in the workplace, people too were told stories about work-less futures. Stories that seemed plausible at the time.

One could name several more examples to amplify this point. In most European nations, roughly 50% of the labor force was engaged with farming until just a few hundred years ago. Now, most modern countries have just 1-3% of the labor force engaged in farming. But losing roughly 50% of the continent’s jobs in just a few decades didn’t cause mass unemployment – instead, it opened the door to industrialization.

Every time in the history of the world that jobs have been destroyed by innovation, we have not been met with a lack of jobs that is constantly prophesied, but with the freeing-up of labor to pursue more productive tasks. But in almost many cases, even the smartest people of the time could not have predicted how the economy would adapt to the new conditions.

One could summarize the situation as a thought experiment: A man can’t see around the corner, but he knows that he has to turn it. Every time he is about to turn the corner, he hears the roar of a fierce lion, and he is sure that the end of his life is near. You observe the man repeating this experiment 300 times, but the lion never shows up. Every time he turns the corner he finds that – in spite of the fearsome roar – it’s safe to do so.

Now, if you had to bet your entire life savings on what would happen the 301st time he turns the corner, would you bet on him being eaten or safe? You might object that this time is different, but remember, that’s what everyone said the other 300 times. Every single time, it was impossible to predict what was waiting around the corner. People were just as prohibited from looking around the corner then as they are now – all they could take stock of is the scary lion’s roar – as is all you can do now. So which option will you bet on?

Of course, abstract frequentist arguments are rarely as grabbing as the concrete threat of lions and robots. That’s part of the reason why every time this happens smart people remain convinced that this time is different. However, there are a few other metrics we could look at.

One is economic growth rates relative to the pace of innovation. To economists, it is paradoxical that we are seeing predictions of a flood of innovation and paired with low growth rates. The opposite should be the case, since innovation boosts productivity, which in turn boosts growth. This might point to there being too little innovation going on these years, as opposed to too much. In fact, the pace of top-level invention has slowed considerably since the 70ies. True, we get new apps and smartphones, but broadly speaking, most of the fundamental technologies driving the economy today are simple extrapolations of technologies that are in many cases 50 or even 100 years old. Global power generation is still very much achieved with fossil fuels and today commercial flights are actually slower than they were 50 years ago.

Another is that regardless of how the actual numbers turned out historically, predictions of jobless futures have always had an especially easy time attracting attention in periods of low growth and insecurity. Which is just what the West is characterized by these years.

Of course, one may argue that observations made of the past are no guarantee of what the future will be like. This is true. But all in all, there is good reason to stop and pause, before rushing in to proclaim that this time is different.