Half a dozen New York Times, Washington Post, and Wall Street Journal articles on robots at work have appeared over the past two-plus months. I’m glad to see this issue getting attention. However, despite a general improvement in the facts and assumptions in such stories, there are still some misconceptions, which continue to be destructive. For example, it has been a long time – over 70 years in fact – since manufacturing peaked as a share of American jobs, and so the issue now is not how we will deal with its decline, but how we will manage the slowdown of its successor, services. I was glad to see, though, a refutation in two of those pieces of the obsolete idea that robotics is affecting only the most menial positions, and the articles had other valuable insights as well. So what is happening, overall, in this area?
First, the key to understanding which jobs are vulnerable to takeover by robots or other automation technology is the presence or absence of underlying algorithms. Any work function dependent on implementing a linear sequence is in eventual danger of being mechanized away. If your job consists of following rules, it will someday not need to be done by a human. The good news is that is robots, computers, other machines, and even artificial intelligence all require algorithms, which is why, for example, janitors are much more resistant to being replaced by technology than are insurance underwriters.
Second, often robotics will not take over entire jobs, but will supplement existing ones, with workers then doing only some of their duties. As Ana Swanson put it in her Washington Post article, “emergency room doctors might spend more time on the most serious or unusual cases, while robots do the triage and diagnose routine illnesses.” However, in most of those circumstances, positions will still be lost, as, in many settings, with the time saved fewer human employees will be needed.
Third, in remarkable contrast to the past, robots are rapidly becoming more common and more sophisticated. Per the International Federation of Robotics, shipments of industrial robots in the entire world added up to 55,000 in 1992, and it was not until 19 years later, in 2011, that that number more than doubled. Since then it has climbed steadily, with about 185,000 in 2014 and a projected 260,000 in 2015. Researchers have also redoubled their efforts to increase robots’ scopes, with, as one example, SoftWear Automation’s expectation that their mechanical devices will be able to sew complete garments, which since it involves soft cloth has been a classically hard type of task for them to complete, by the end of this year.
Fourth, in what may well be the most common urban-ride system of the 2020s, General Motors is combining with amateur cab company Lyft to develop driverless taxis. That could rescue both companies, if regulation catches up with the latter and, as some people think, the proliferation of self-driving cars slashes sales of privately owned ones. It is a visionary step, well worth the $500 million GM is investing, as it shows that both organizations are placing themselves more centrally in the auto-transportation business. They are avoiding a mistake that has happened to entire industries; if American railroad companies had not fallen into the trap of being committed to trains, we might be flying Southern Pacific and Milwaukee Road instead of United and Southwest.
Fifth, even with robots aplenty, United States manufacturing will hardly disappear. In 2013, the last year for which full statistics are available, only China added more value that way. American labor being more costly is nothing new, but there remain many applications, such as aircraft, where technical knowledge is critical, and many others where the convenience of being close to the market outweighs foreigners working for less. As more and more jobs get automated away, cheaper-labor countries lose much of their advantage, as robots cost much the same to operate anywhere. Although there will never be anything like the armies of people going to factories we had in the past, there will still, indefinitely, be a lot of things made here.
Sixth, we have not resolved a concern I raised over four years ago. The final Work’s New Age Principle states “Factory jobs replaced farm jobs. Service jobs replaced factory jobs. No paid work that we know of now will replace service jobs.” This situation explains why we cannot count on new positions to replace ones we lose. It is true that, as researcher Michael Chui said in the Washington Post piece above, “we have been automating work for hundreds of years, and we’ve always been able to create new jobs,” and that, although “a century ago, most of our ancestors were farmers,” “today, it’s not as if we have 80 percent unemployment.” Likewise, Ji Shisan, writing in The New York Times, was correct to say that “we don’t blame the steam engine or tractors or sewing machines for unemployment now.” Yet we know what replaced the numerically devastated extraction and manufacturing jobs of the past, and we still have nothing more to cover lost service ones than classic futurist Herman Kahn’s Quaternary phase, in which people do things for themselves or their communities without pay. With robots here to stay, if we can’t get out of this bind, we may, someday, see that 80%.