There is growing fear, or at least suspicion, that smart machines will soon lead to unemployment as they eliminate more jobs than they produce. A pew survey of “experts” in the fields of artificial intelligence and economics showed an even split about whether or not this is an inevitable outcome of technological progress. Oxford philosopher Nick Bostrom said that wage labor will plummet if sophisticated AI is created, while others such as Mike Collins, writing for Forbes, and MIT’s David Autor are more optimistic, at least for now because artificial intelligence falls qualitatively short of what human minds can do. Is this digital or cognitronic revolution economically different from the industrial revolution? Nobody today thinks the cotton gin or railroads were a bad idea. Some argue this is different and this post is about those arguments.
Two kinds of fears: short and long term
Innovations can be disruptive to economies in the short-run. The industrial revolution and assembly-line production laid to waste entire industries of handmade products in the world. There are no more milkmen, log drivers, radio actors, switchboard operators, or ice cutters. When entire occupations vanish due to technological obsolescence, unemployment can rise and many lives can be adversely affected. But nobody thinks we would be better off picking cotton by hand just to make jobs, or that we should not use email (or blogs!) because it reduces the need for letter carriers. Short term disruption is just a price we must pay for progress sometimes. What about the not-so-short term?
Now we get theoretical. There has never been any long-term loss of jobs or economic malady caused by technological innovation. New technology that creates more of something without using more resources causes the economy to grow and, usually, product costs to fall. When products are cheaper, then new businesses can emerge and hire more people because the cost of doing things has decreased. The argument by those such as Nick Bostrom seems to be that there is a finite list of possible jobs and that machines that can replace or do human-type thinking will take more and more of these jobs, which then cannot be replaced as these are finite in kind.
In the following video “Humans Need Not Apply”, popular YouTube channel author C.G.P. Grey explains what is different now is that the machines replace brains, whereas during most of human history machines simply replaced human muscle.
He says that formerly human-only feats like composing music, writing news articles, and making medical diagnoses show that most jobs will become technologically obsolete, not just drudgery or simple tasks.
This sounds a bit scary. Should we fear automation intruding into think-y and creative jobspaces? Is there a coming economic collapse due to technological unemployment? No. There is no reason whatever to think any of this and I will explain why.
Economists have long pointed this out. In fact it has a name, the Luddite fallacy. New technology replaces jobs when it cuts costs. This means goods and services are cheaper and the disposable income of the average citizen increases. They will then spend that money elsewhere; often the shift spurs the growth or creation of new goods or services, but one way or another, it ends up reinvested in industry.
Or, look at it this way. Some magic robot is invented tomorrow. It drives, does surgery, your taxes, makes a great tofu burger, and a hundred other jobs for pennies on the dollar versus the cost if a human did the job. Half of the population is now unemployed and unemployable. Now, question: who is buying those goods and services? The people who don’t have jobs or money? Few, if any, businesses could survive a sudden loss of half their consumer base. This is an extreme example, but it illustrates the point: it isn’t economically possible for a corporation to be profitable by using any technology that reduces costs while removing customers from its market. In other words…
The economy serves us and is us
I like C.G.P. Grey’s videos. They’re sharp and informative, but in the above video he draws a senseless comparison between the obsolete horses replaced by cars and not given new jobs, and humans who are soon to be similarly replaced. The comparison is daft because cars were invented and produced to forward the interests of people. If they didn’t, they would not have been. If horses had a say in the matter, history would have unfolded differently.
There is no evidence
In the long-term, there has never been any evidence that technological advances have increased the overall unemployment rate. —Tejvan Pettinger
Right now, the US has a low unemployment rate, 5.5% and it has been dropping consistently for a year. This is at or near what most economists call “full employment” and there is no recent trend toward unemployment.
The paranoid claims of technological unemployment are that it is “coming” and always “just around the corner”, but it never seems to get here. This is true in spite of the fact that we have now had machines replacing scores of service and professional (read: not mere labor) jobs for years. I am open to revising my expectations about the future. Technology can indeed re-write what we think of as possible. But some evidence must be provided to support these claims. As of now, all evidence is to the contrary.
Why so grim?
We fear unemployment and replacement by robots because robots really do replace our jobs and adversely affect our lives. But this is a problem of social and economic policy, not the basic nature of economic systems. History is very clear that being at or near 100% employment is no insurance against millions of people starving to death. We should have safeguards against anyone lacking food and shelter, regardless of the details of our economy.
The word “Luddite” famously refers to early 19th century laborers who smashed machines such as weaving looms and threshers which had reduced demand for certain jobs. Most people think the Luddites were technophobic or afraid of technological innovations putting them out of work. But this is probably wrong. In fact, they targeted 200-year old inventions (among others) and were protesting against unfair labor practices, according to Smithsonian Magazine‘s Richard Conniff. This truth about the Luddites is grounds for real worry. In their time, industry changed faster than regulatory policy and public understanding of fair economic policy. This could happen again. Not because a machine shows up to do your job, but because of the inhumanity of the CEOs and traders that may seek to abuse both.