What Billy Joel Can Teach Us About AI
It’s too simplistic to view advances in automation as inspiring either sadness or euphoria.
At first blush, the aging pop star Billy Joel wouldn’t seem to have much to tell us about automation and the future of work. His main contribution to forecasting was the dystopian “Miami 2017 (Seen the Lights Go Out on Broadway),” a 1976 song that ends with the island of Manhattan being sunk at sea. (Spoiler: It’s 2017 and the island of Manhattan is very much buoyant. I was there today.)
However, in sifting through the constant barrage of headlines, reports, projections, and warnings about robots and automation, the hook of another Billy Joel song, “Summer, Highland Falls,” comes to mind: “It’s either sadness or euphoria.”
Those seem to be the polar reactions to the advent of the next wave of disruptive technology. Automation and artificial intelligence, we are told, have the capacity to replace all manner of jobs quite soon. And the result will likely be sadness on a massive scale for many workers. Each day brings news of investments and progress in autonomous driving. Well, it turns out that one of the most common jobs done by people in the U.S. without a college education is driving — trucks, taxis, buses. And if the avatars of autonomous driving are to be believed, virtually all of those people could be at risk. According to a U.S. government report issued last year, some 3.1 million driving positions — about 3 percent of our current workforce — could be vulnerable.
With the growth of e-commerce, warehouse jobs have been another bright spot for less-skilled workers. But on Monday, the Wall Street Journal reported on a new breakthrough that could let robots put individual toys into boxes and then close them. According to the Journal, several companies, including big department stores, “have recently begun testing robotic ‘pickers’ in their distribution centers. Some robotics companies say their machines can move gadgets, toys and consumer products 50% faster than human workers.” That would be very sad news indeed for the 950,000 Americans who work in the warehouse sector.
Euphoria, of course, is the flip side of sadness. As sports fans know intimately, one person’s devastating loss is another person’s source of joy. In economic terms, the beneficiaries of automation are legion. It starts with those who benefit from owning the robots and software or from deploying them. As bad as automatic pickers will be for warehouse workers, it will be great for the high-level managers and shareholders of Amazon. But automation can also bring euphoria to the hundreds of millions of American consumers. By taking costs out of the system, automation and artificial intelligence will allow people to access more goods and services at lower prices.
There’s another strain of euphoria. Take humans out of the equation, the reasoning goes, and you remove costly human failings. Robot pickers would be less likely than people to pack the wrong item, thus ensuring that children get the exact presents ordered for them every time. In theory, letting machines do more of our work should lead to healthy benefits for everyone. One 2015 study, for example, argued that autonomous driving would reduce traffic fatalities by 90 percent.
But here’s the thing. Just as much of human experience exists in the vast terrain between sadness and euphoria, history tells us that technology and automation unfold in the broad spectrum between disaster and nirvana. Behind both the sadness and euphoria lies a kind of technological determinism: Just because a technology can do a task in some instances, it will do so in all instances. And, further, that technology will be welcomed and enabled by society and policy. Oh, and once it is up and running, it will be flawless in its daily performance.
According to a U.S. government report, some 3.1 million driving positions — about 3 percent of our workforce — could be vulnerable.
One can understand why computer scientists and engineers subscribe to this vision of determinism: They are enamored of the algorithms they write in their labs. But the world in which code goes to work is significantly less elegant, and much messier. It turns out that, as Vinnie Mirchandani argued in s+b, society has built up powerful and important circuit breakers that often inhibit or delay a new technology’s rollout. I have no doubt that machines will do more package sorting in the future (just as they do more mail sorting today than they did 10 years ago). But I sincerely doubt they will do all of it in the next decade. I have no doubt that computers and software will play a greater role in helping drivers move their vehicles around safely (and they already are). But I also seriously doubt that, in 10 years’ time, the roads will be populated entirely by autonomous vehicles. What’s more, given the frequency with which my phone drops calls, my computer crashes, and the airlines mix up my flight reservations, I’m skeptical that these systems will function entirely smoothly.
Yes, many employees will be displaced by smart machines in the coming decades, just as millions have been displaced in the last several decades. Rather than sitting back and being either sad or euphoric that machines are doing all our work, people will find an experience somewhere in between: Humans working with machines, humans overseeing machines, humans taking directions from machines.
Just because a machine can do a person’s work doesn’t mean that our economic system will give it a reason to displace its human analog. A programmed player piano can do excellent renditions of Billy Joel songs. But nobody will pay to listen to it. Meanwhile, the Piano Man, now 68 years old, sells out Madison Square Garden every month.