The idea of technological unemployment goes back to the Luddites. The Luddites were a group of 19th-century textile artisans that feared the technological development of automated looms because it would cause massive unemployment. Both Karl Marx and John Maynard Keynes picked up on the idea of technological unemployment. The Government Accountability Office (GAO) wrote a report on it back in 1982. Books and articles have been written about how we'll lose our jobs to machines. Yet we've had disruptive technology before. The cotton gin, the printing press, the automobile, the dishwasher, the Internet: these are all examples of technologies that disrupted labor markets in some way, shape, or form. How did the Luddites end up faring? How does anyone whose job is potentially or actually affected by technological development?
They were somewhat right, but it's a similar phenomenon we see with immigration or outsourcing because looking at the layoffs and resulting short-term job displacement (read: short-term pain for longer-term gain) is only part of the equation when considering the effects on labor markets. Not only did new jobs develop in the textile industry as a result of automated looms, but society as a whole prospered by having access to better, cheaper clothing. This is the sort of trend that has historically taken place with new technology. While there were some individuals in the horse-and-buggy industry lost their jobs in the short-run because of the automobile industry, you're going to find very few individuals who are going to honestly say that the invention of the automobile has been an overall loss. Plus, if you want to decry unemployment, look at the past century of unemployment history. We don't see any significant changes in the unemployment rate that would suggest massive, long-term, structural unemployment. There is no evidence that technology has caused massive unemployment or even the shift in income inequality (also read this report). Any massive shifts in unemployment in the past century-plus have either been caused by financial crises and/or government intervention. If history is an indicator of anything, it's that the amount of people employed overall will increase because the short-term job displacement only leads to labor market pattern shifts in the long-run.
But maybe this time, the effects of automation will be different. Maybe there's too much of a good thing. Perhaps we are in a stage in which massive acceleration in technological development makes the past a useless indicator for how labor markets will fare. It's not the first time this argument has been used, but the New York Times actually brings up a good point as to why it might be more difficult this time around:
Through much of the 20th century, workers moved out of agriculture and into manufacturing jobs. A high school diploma and a basic willingness to work were often enough, at least for white men, because the technologies of those times often relied on accompanying manual labor.
We cannot deny the appeal of behind robots. Faster, cheaper, and better inputs translate into overall improved quality of production, all of which are good for the consumer. This technology creates for a better customer experience, which creates new demand. There are certain advantages that come for the employer, as well. Robots don't need to be trained. There are no issues of overtime, on-the-job theft, or health concerns with a robot. The closest a robot gets to being fired is being decommissioned. Plus, robots are a good way of avoiding issues with minimum wage, liability (read: expensive litigation costs), and anti-discrimination laws. If you want human labor to exist with the labor participation rate that currently exists (or even push for a higher rate), we need policies that encourage the hiring of human labor.
This round of automation is scarier than past rounds because the decoupling between productivity and employment is not just going to affect low-skilled labor; it's going to affect everyone. An Oxford study (Frey and Osborne, 2013) estimated that as much as 47 percent of U.S. jobs could potentially be at high risk of automation in the the next decade or two. Frey and Osborne also wrote a more recent report for Citi to the same effect. However, not everyone is convinced. Some think that the upcoming automation won't be any more disruptive than past technologies (Gordon, 2014; Autor, 2014; Miller and Atkinson, 2013) due to complementarity.
Similar to economic modeling or climate change modeling, certain prognosticators saying that "we are at risk" isn't the same thing as "it's going to happen." Robots and computers will undoubtedly play a part in the future of labor markets. The acceleration also presents unique challenges that our ancestors did not have to contend with when it comes to technological change. If I had to make an educated guess, I would opine that there are going to be some sectors more heavily affected by others. As research from the Brookings Institution points out (Kearney et al., p. 4), being able to automate routine acts will be easy. Jobs with non-routine physical movement or abstract tasks, on the other hand, will be more difficult for robots to perform. Language recognition and in-person interaction will also prove to be elusive for robots. Empathy, communication, problem-solving, persuasion, and creativity are all skills that very well might never be mastered by robots (Autor, 2014). The New York Federal Reserve shows how manual and routine jobs have already been on the decline since 1975, whereas cognitive and non-routine jobs have been on the uprise.
The question here is not about whether certain sectors will experience automation [because it's already happening], but how employees, employers, and governments respond and adapt to the change that will affect the course of the future. If you want to push for a $15 per hour minimum wage at fast food restaurants, it's only going to incentivize McDonald's to automate sooner. If you're worried about corporations making wicked profits off of robots, then we need to take another look at intellectual property rights and patents because it's just a tool for crony capitalism to flourish. We need an education system that doesn't just teach critical thinking and interpersonal skills; we need to teach skills that people can use in the workforce instead of peddling the idea that "everyone should pursue their dream." What's more is that we need a workforce that can continuously develop skills for an increasingly dynamic job market (Kearney et al., p. 5). We also need to focus on policies that will encourage entrepreneurship and business dynamism (ibid., p. 6). To accomplish such dynamism, we need tax reform to encourage competition (e.g., removing the corporate tax, employer-sponsored health insurance), as well as deregulation in areas that don't need regulation (e.g., occupational licensing).
Yes, there are going to need to be some drastic changes as to how we think about job creation and job training. At the end of the day, the fear of losing your job to a machine could very well be nothing more than a case of the Frankenstein complex.
2-27-2016 Addendum: Here is a well-thought out list of five reasons we should not be worried about artificial intelligence.
6-6-2016 Addendum: The Organization for Economic Cooperation and Development (OECD) published a report last month on automation. While most people will experience the automizing of certain facets of their jobs, only 9 percent of U.S. jobs will be fully automized. Rather than worry about robots taking over in the labor market, it's more a matter of how to use robots to help us create jobs that require less automation. Consulting firm McKinsey also estimated last November that only 5 percent of jobs will be fully automated.
12-17-2017 Addendum: This article covers two myths in the automation debate: 1) Production is nowhere high enough to make automation problematic, and 2) job destruction by automation is not high enough because automation will more likely change the nature of a job instead of downright eliminating it.