The Engineer’s View on the Robot Apocalypse

In Sarah Kessler’s The Optimist’s Guide to the Robot Apocalypse, she evaluates the question of whether robots will take all our jobs. Ms. Kessler’s take is that we shouldn’t worry, this won’t happen, just like it didn’t happen all the times in the past that automation threatened to take our jobs.

Kessler summons some charts and factoids to make her case. That would be a great argument if the charts and facts actually supported her case, which they don’t. Lets have a look.

Kessler’s primary historical example is the automated weaving frame. Queen Elizabeth I refused to grant a patent for an ancestral device because she feared it would put too many subjects out of work.

By the early nineteenth century, these weaving frames had put tens of thousands of Englishmen-and-women out of business, destroying what had been a successful cottage industry, reducing wages by 90 per cent. This disaster gave rise to the Luddite movement, who were not opposed to technology per se, but merely against starving because of it. The Luddites ransacked factories, destroying the automated equipment.

The English government’s response to the Luddites’ popular uprising was instructive. They sent in troops to protect the automated weaving equipment. In 1812, the English parliament made destroying a weaving frame a capital offense, while declining to legislate against starvation.

Kessler says “…weaving technology ended up creating more jobs for weavers. By the end of the 19th century, there were four times as many factory weavers as there had been in 1830”. This statistic is disingenuous. Employment of weavers dropped precipitously from about 1800 to 1830. The population who had been cottage weavers had to physically move to other cities with other industries, and wait an entire human lifetime before workers displaced by automation were employed again by the fruits of automation.

The new jobs created by automated weaving mills were of an entirely different character to the lost jobs. Weaving mills of the late nineteenth century were legendary for monotonous, dangerous, low-wage work, child labor, industrial accidents, and all the ills that led to the rise of the labor movement. This is hardly an endorsement of the benefits of automation.

The weaving industry is, in a nutshell, a perfect example of the robot apocalypse. There was no new, higher paid work to replace the jobs lost to automation, only marginal employment.

Kessler summarizes her history lesson, saying, “…widespread unemployment due to technology has never materialized before.” I would say her own evidence refutes this claim. I might mention also the mass migration of population from farms to cities in the twentieth century, the hollowing out of Rust-belt manufacturing jobs in the 1980s, and the jobless recovery of the last 5 years. We put a positive spin on these losses, talking about the “service economy”. But that is a code name for low-paying retail jobs replacing well-paying factory jobs.

Kessler shows a graph of Amazon.com jobs, showing that while Amazon has exponentially increased the use of robots, the use of human labor has also risen. That’s great, but what would be Amazon’s use of human labor if there had not been robots. This graphic proves nothing at all, other than that robots are taking over.

Kessler cites the usual arguments that reduced costs due to automation permit either lower prices or more profit, which might lead to more employment. Or it might lead to more robots, which Kessler calls “increased investment”. It is my personal opinion that relying on a second-order effect of reduced costs to maybe create increased employment is entirely too slender a thread from which to hang society’s whole future, when robots are clearly doing things that humans used to do. We have, after all, the example of the automation of weaving to inform us.

Kessler cites studies showing a correlation between automation, particularly in the form of increased computer use, and increased employment. It’s not immediately clear whether these studies consider network effects like decreased employment at competing firms. Some of these studies cover industries where computerization has made new kinds of products (flat screen TVs, smart phones) possible. But when automation is used to decrease costs, and the chief cost is labor, automation will devastate employment in these industries.

For instance, there are about three and a half million truck drivers in the United States. These jobs currently pay a relatively high wage and require minimal training. According to a report issued by the White House last December, most of these jobs will eventually be automated.

It’s interesting that technology has eliminated jobs from the cotton fields, to the weaving mills, to the truckers who bring the goods to market, to department stores disintermediated by the internet. As whole networks of jobs are automated, displaced workers have fewer and fewer places they can turn for work.

And the work they can get is of lower quality. They are competing with robots to drive trucks, sell air duct cleaning over the phone, or scan medical slides for signs of cancer. The tirelessness and consistency of robots colors the jobs that humans do, making them more mechanical, more subject to quotas and measures. Humans are not accustomed to being inside the automation. They rebel against it, but it’s no use. They are just carbon-based pick and place robots now, doing the routine jobs that are just slightly too complex to fully automate today. Kessler acknowledges this fact, but it doesn’t seem to bother her.

Automation Makes Humans Obsolete

Like Kessler, many people have been smug about the loss of low-skilled jobs. Whole categories of jobs have disappeared; telephone operators, typists, drafters, typesetters all rendered irrelevant by computers in the 1960s and 1970s. Then assemblers and manufacturing workers in the 1980s.

But information workers believed they were the future. Their jobs were creative, and machines could never do that. Only they didn’t have to. They only had to debase the value of these professions. Computerization bit into line managers in the 1990s. Photographers, reporters, and journalists felt the cold hand of the robot apocalypse in the 2000s. Now lawyers, radiologists, pathologists, and university professors are endangered. All of a sudden it’s not just low-skilled people losing their jobs. It’s pretty much everyone.

Displaced workers have fewer and fewer options if the remaining jobs require high I.Q.s and a college education, or offer subsistence wages and no benefits. At best, these workers have to retrain for a period of several years, then move to places where the high tech jobs are, all while avoiding starvation. At worst, it means underbidding a robot for a subsistence wage. Eventually even these roads may not be passable.

Kessler says we may need automation to counteract declining birthrates. It may also be that birthrates are declining because there is nothing for us to do. The world is full, so people stop making babies.

Greed is the Key to the Robot Apocalypse

Philosophers and religious scholars tend to think there is something unique about human consciousness, some ineffable spark that cannot be captured. Engineers are more sanguine. From an engineering standpoint, there is no reason to believe that robots cannot eventually be made to do absolutely everything humans can do, and perhaps more. After all, to an engineer, a human being is just a meat robot, controlled by a computer made of grey meat.

Kessler argues that as aspects of each job are automated, that enables workers to add more value and earn a higher wage. If 40% of a job is automated, it frees up time in which workers can earn more value. That’s one outcome. Another is that 40% of workers are let go, and the remaining workers do the remaining work. Any result other than the optimistic extreme makes jobs go away. Helping workers earn more value requires an investment in training. Cutting the workforce makes fewer demands on the captains of industry.

There is tremendous profit motive in making better robots, at least initially. Thousands of very bright engineers are beavering away today trying to make us obsolete. The first ones to do it will be wealthy beyond imagining. But how will mere humans compete with the superior mechanisms they build? We may think, like Dr. Frankenstein, that we will control the monster, that it will do our bidding. However, a fully autonomous robot is a better robot, more flexible and versatile. Someone will build one simply for short-term profit. There is absolutely no reason to believe that this story will end well for us humans.

I can visualize the last Neandertal, admiring those tall, gracile, homo sapiens, wondering what the future will be like without Neandertals. Maybe our robot overlords will be gentle with the last of us, keep us as pets, or as a reserve of intelligence in case of unforeseen need. But if they are careless and we die out like the passenger pigeon, it will be no great loss to them.

I consider an outcome as optimistic as that described by Ms. Kessler as so unlikely, that I wonder who is paying to have such material presented over and over. When I hear other engineering minds sounding notes of extreme caution, it makes me hope that I am old enough to die peacefully in my sleep before the turds hit this particular turbine.