The New Yorker recently published an article by Elizabeth Kolbert on automation, entitled “Our Automated Future.” The subhead of Kolbert’s article is “How long will it be before you lose your job to a robot?” Her depressing answer is “Not long.” She cites Jeff Bezos, founder of Amazon, saying, “It’s probably hard to overstate how big of an impact it’s going to have on society over the next twenty years.”
Popular press about automation has tended to focus on manual, blue collar, relatively unskilled work. Self-driving trucks (the first of which was tested by Uber a few months ago) threaten the jobs of 1.7 million truckers. Burger-flipping robots, being developed by startups such as “Momentum Machines”, will make tens of thousands of fast-food workers obsolete. (Its cofounder seems positively giddy about the development, saying “Our device isn’t meant to make employees more efficient. It’s meant to completely obviate them.”) And similar developments threaten workers in blue collar fields of janitorial services, construction, manufacturing, and warehouse work. In fact, I got most of these links simply by typing “[blue collar field] + robots” into Google.
Arguably, though, white collar work is even more vulnerable. One key reason is that replacing blue collar workers requires hardware (i.e., physical robots/machines), while replacing white collar workers often requires only software (i.e., a computer program). One “expert” quoted in a CNBC report stated, “There was a very well publicized report by Oxford University that ranks your job relative to its likelihood to be automated by 2020…Top of the list of those jobs was highly repetitive clerical-type activities…Anything that is highly repetitive, that follows a repetitive role and uses some kind of technology, can be automated.”
The Oxford University report that he refers to is fascinating. (It’s also 73 pages long, which makes it not-so-light reading.) I chuckled at the dry academic writing that could pass for black humor, such as, “Computerisation of cognitive tasks is also aided by another core comparative advantage of algorithms: their absence of some human biases. An algorithm can be designed to ruthlessly satisfy the small range of tasks it is given. Humans, in contrast, must fulfill a range of tasks unrelated to their occupation, such as sleeping, necessitating occasional sacrifices in their occupational performance (Kahneman, et al., 1982).” In other words, automating white collar, “cognitive” work is made easier by the fact that humans can’t do it 24 hours per day.
The key section of the report is where the authors develop a model for classifying which jobs are most likely to be automated. (In the future, I’m sure, a machine learning algorithm will be able to make this classification much more accurately.) Thankfully for us humans, “Yet some inhibiting engineering bottlenecks to computerisation persist.” What are these obstacles? They prevent automation from extending into three categories of tasks/jobs: 1) “perception and manipulation tasks” (e.g., surgery); 2) “creative intelligence tasks” (e.g., art or literature); and 3) “social intelligence tasks” (e.g., public relations or nursing). They estimate that 47% (!) of jobs don’t have significant bottlenecks preventing their automation. In other words, the same percentage of people who Mitt Romney thought were layabout losers are about to become so.
This taxonomy reminds me of a comment that Marco Rubio made during one of the presidential primary debates. He argued for more vocational training, saying “Welders make more money than philosophers. We need more welders and less [sic] philosophers.” Rubio’s remark was widely panned, mostly for being untrue, but the context in which he uttered it is also interesting:
“If you raise the minimum wage, you’re going to make people more expensive than a machine. And that means all this automation that’s replacing jobs and people right now is only going to be accelerated.
Here’s the best way to raise wages. Make America the best place in the world to start a business or expand an existing business, tax reform and regulatory reform, bring our debt under control, fully utilize our energy resources so we can reinvigorate manufacturing, repeal and replace Obamacare, and make higher education faster and easier to access, especially vocational training. For the life of me, I don’t know why we have stigmatized vocational education. Welders make more money than philosophers. We need more welders and less philosophers.
If we do that — and if we do this — if we do this, we will be able to increase wages for millions of Americans and we will be able to leave everyone better off without making anyone worse off.”
Here’s the strange thing to me. Rubio seemed to understand that automation is a grave threat to American jobs and wages. But he didn’t seem to understand how automation actually works. (Perhaps not surprising, given that Rubio might actually be a robot himself.) There are minimal bottlenecks to automation of welding. It is not a “creative intelligence task” nor a “social intelligence task”. It requires some amount of perceptual and motor skills, but nowhere near the level of, say, a surgeon. Unsurprisingly, welding has been undergoing automation/semi-automation for decades now. If I had to choose a job for the future, and my choices were philosophy and welding, the former would win, no contest.
(Rubio also failed to grasp that “repealing Obamacare” and “bring[ing] our debt under control” have literally nothing to do with preventing jobs from being automated, but, to be fair, neither did any of the other bozos on stage.)
Politicians on the left are, for the most part, not blithering idiots like Marco Rubio. But they too seem to fail to grasp the extent of the problem of automation and the speed at which it will arrive. In his State of the Union address, Obama talked about automation, saying “We live in a time of extraordinary change, change that’s affecting the way we live and the way we work. New technology replaces any job where work can be automated. Workers need more skills to get ahead. These changes aren’t new, and they’re only going to accelerate. So the question we have to ask ourselves is, ‘How can we make sure everyone has a fair shot at success in this new economy?’” The answer is apparently teaching kids new skills and more skills, such as “how to code.”
Democratic politicians and supposedly non-partisan experts have been promulgating this idea for decades. Bill Clinton, for instance, talked about job retraining as a bridge between the “Old Economy” and the “New Economy”. Bill Gates says that people need backgrounds in science, engineering, and economics to be successful in the economy of the future. And here’s a similar-minded excerpt from the CNBC article I referenced earlier,
“Kinson dismissed fears that every white collar worker is in danger of losing their jobs to robots.
“There’s been a lot of talk in the marketplace about this impending doom and collapse associated with the rise of this robotic technology, but this is an opportunity to enrich the activities of the people that remain.”
Kinson also suggested that automation was an alternative to companies attempting to cut production costs by moving operations offshore.
“If I can produce a robotic labor force, maybe I don’t need to look to ever more rural parts of India or further and further undeveloped economies to find that lower cost labor force,” he said.
“Actually, what I can do and would otherwise seem impossible is retain the high value work which involves thought and judgement and eliminate what is really highly mundane and repetitive.”
Setting aside the thinly-veiled threat to outsource jobs if we don’t let automation prevail, what Kinson is saying is that jobs of the future will involve “thought and judgment”, unlike the jobs of today that are being automated. (In other words, more like philosophy and less like welding.)
But where does that leave the people who don’t have such thought or judgment?
I am (thankfully) employed in one of the industries that will benefit from automation. I also devoted 10 years of my post-high school career to education. I am one of the workers that Obama or Gates or Clinton or perhaps even Rubio would tout as an exemplar, as a model for what workers should be doing to ready themselves for the future economy. But can we reasonably expect everyone to follow my path?
Take a look at any high school class in the country. What percentage of students have the aptitude to be a “knowledge worker”? Hell, take a look any college science or engineering class, of the type that Bill Gates is enamored with. Many of these students struggle with basic algebra, let alone the math required to design the robots of the future.
What I’m trying to say is that an economy based on knowledge alone is infeasible. An economy in which everyone designs robots to do the jobs that they used to do is unworkable. We can’t even figure out how to bring industry back to Appalachia or the Rust Belt, let alone 47% of the country. And no number of out-of-touch CNBC experts or clueless politicians will convince me otherwise.
Consider this, too. Even if everyone had the requisite math and engineering and coding skills, designing robots simply doesn’t employ that many people. As Kolbert writes, “Google also illustrates how, in the age of automation, new wealth can be created without creating new jobs. Google employs about sixty thousand workers. General Motors, which has a tenth of the market capitalization, employs two hundred and fifteen thousand people. And this is G.M. post-Watson. In the late nineteen-seventies, the carmaker’s workforce numbered more than eight hundred thousand.”
Politics is a slow-moving enterprise that often concerns itself with fighting the battles of the past. The Republican base in 2016 was still tilting against Roe v. Wade (which happened over 40 years ago), NAFTA (which happened over 20 years ago), and illegal immigration (which peaked 10 years ago). So I wonder: how will politics cope with problems such as global warming and automation? By the time we get around to addressing them, it might already be too late.