One of the several jobs-related topics getting gobs of press this year has been robots. I could write a post every week on only the stories released, even if I just hit their high and low points. However, in the interest of writing on other topics as well I have restrained myself to two, on January 15th and May 2nd, unless you count my March 25 artificial intelligence issue.
On May 2 I named nine points for universal agreement, on robots’ lack of actual intelligence, their need for controlled environments, their one-way progress, their potential to do any algorithmic task, the value of projecting where they are going, their productivity gains always involving job takeovers, their uncertain acceptance, and the dangers of their being programmed amorally. These aren’t really controversial, and everyone from Luddites to Ray Kurzweil should buy into them. However, too many otherwise good recent articles have included clearly incorrect statements. What are they, and why are they wrong?
“March of the machines” (The Economist, June 25th) cited artificial intelligence (AI) researcher Andrew Ng as saying that worrying about a general AI breaking free of human controls “is like worrying about overpopulation on Mars before colonists have even set foot there.” While I have not taken what is now called AI as seriously as most, since to me it seems only incrementally more advanced than simple if-then computer programming, if we can teach systems how to learn and how to set their own goals we must consider what they might do, as their actions could be far less controllable than regulating the number of people moving to or even living on an extraterrestrial base.
In Harvard Business Review’s June 23rd “Knowledge Jobs Most Likely to Be Automated,” authors Julia Kirby and Thomas H. Davenport noted the trend for robots and other automating features to take over parts of jobs, which they say prevents naming areas where people should avoid jobs and careers. That is false, since as I documented in 2013’s Choosing a Lasting Career, jobs in the 25 U.S. Department of Labor career groups vary vastly in how susceptible they are to not only automation, but globalization, efficiency, and 18 other current or anticipatable trends. Whether you agree or disagree with my findings, the mass of which are still current, there is no reason for employers to minimize automation gains by maintaining the same numbers of pertinent workers. As well, the authors’ statement that machines will be “freeing up humans to work on more value-creating projects” is erroneous, at least in the huge majority of fields with good job candidates not being hired as it is.
In “Industrial robot sales hit record” (Financial Times, June 22), Michael Pooler quoted a company robotics head as saying that “people today don’t want to do dull, dirty, dangerous and delicate jobs any more.” I could introduce this man to tens of thousands of Americans in my state alone who would gladly do all of those things – if the pay were right. Almost all skills shortages are really training and pay shortages; while those are the free choices of the companies involved, and they are entitled to position them as workers’ shortcomings instead, we have the final option in what we will believe.
Eduardo Porter’s June 7 New York Times piece, “Jobs Threatened by Machines: A Once ‘Stupid’ Concern Gains Respect,” looked at some invalid thinking on work and automation across the years, in response to Wassily Leontief’s pithy and prescient response to those doubting machines could take over jobs, “They replaced horses, didn’t they?” It included another economist, Kenneth S. Rogoff, saying that “it seems unlikely that millions of workers are headed to the glue factory like discarded horses” (why not, and why doesn’t that simile fit?), and a comment from a young Lawrence H. Summers that productivity gains from technology would create so many jobs from people having more money to spend that employment would stay at least the same (sure, if there were no products like software where cost per item drops almost as fast as the number of items sold goes up, and if companies paid workers according to productivity instead of on supply and demand). It is common in economics to see even eminent practitioners saying things that, on further analysis, simply don’t make sense – I suspect political biases for some, but don’t understand its other causes.
On June 1, The Washington Post published Robert J. Samuelson’s “The robot invasion that isn’t yet here.” Samuelson, who has written much fine material, took a remarkably anti-job-loss stance here, saying that since our country had added 14 million jobs in six-plus years “there’s little evidence that robots have yet had much effect on job creation in the current recovery.” That does not follow to me. He also wrote that according to a University of Michigan study “only 16 percent of respondents wanted self-driving vehicles” (there were 260 million registered in the U.S. in 2014, which works out to “only” 41.6 million). At the end of the piece, he fell into the common trap of saying that “workers may become scarce” when “baby boomers age and retire” (as I have documented in no fewer than three books, most in the baby boom generation will not voluntarily retire, and despite 10,000 Americans turning 65 every day, latent demand for work is stronger, in relation to official unemployment, than it has ever been.)
Lilah Raptopoulos’s April 8 Financial Times compilation “Robots: friend or foe? You told us” presented readers’ comments on robotics, with most chosen ones being excellent. One responder said that he asked his 17-year-old son if he thought his “future profession already exists.” That is not an error, but brings up an important distinction. Even though many specific job tasks are completely new from five or ten years before, the overall positions change remarkably little. Martin Ford, in The Lights in the Tunnel, documented how, in 2006, not a single American job area, except for the subset of restaurant work labeled “fast food,” with at least one million workers was nonexistent in 1930. Obviously that could change, but we know of no reason why it should.
Finally, an old area of automata reappeared, by Maija Palmer on May 4th, also in Financial Times: sexbots. Maybe my recollection is wrong, but it seemed a foregone conclusion two decades ago that not-quite-human sex partners would be common by mid-century if not sooner, and I have read almost nothing since. This story, “Prospect of sexual relationships with robots poses moral dilemmas,” started almost from the beginning, with cinematic references, currently available sex dolls with semi-robotic features, and one commentator’s thought that “female sex robots will dehumanise women” since “creating objects that closely resemble human females… leads men to regard women as objects.” Well, if they don’t resemble human females, they won’t be too popular, so I expect the mainstream view on such things to take two parts: one, and more than anything else, apathy; and two, seeing such things as economic inferior goods like bottom-grade beer.
On that not really titillating note, I wrap up. Be assured, no matter what, that robots will take over plenty more jobs somewhere and somehow – maybe not huge amounts today or tomorrow, but during, if you are lucky, your lifetime. Let’s think straight about them.