The robots are coming for our jobs. No, seriously.
There’s no need to take up arms — they come in peace — but you may want to dust off your resume and make sure you have the skills that our future robot overlords will be looking for.
As many as half of U.S. jobs will replaceable by machines in the next decade or two, according to a recent Oxford study explored last week by Bloomberg’s Aki Ito. Want to know the future? Talk to the robot lawyers at Stinson Leonard Street LLP, who reduced the human hours spent researching case documents by as much as 95 percent. Or ask Aethon Inc.’s TUG robots transporting food, medicine and soiled linens in a hospital near you.
(Actually, those robots aren’t yet trained to answer questions about the future of humanity. For that, you’d be better off talking to Apple Inc.’s Siri, who might have a snarky answer, or IBM’s Watson, who might just have the right one.)
The shift to automated workers isn’t new, but it’s speeding up. The jobs most immediately at risk aren’t necessarily low-skill jobs like janitorial services, table bussers or hot dog vendors. Those positions are varied enough and the employees paid poorly enough that it doesn’t make sense to replace them with expensive automated systems. Similarly safe (for now) are some of the highest-skill jobs: CEOs, judges, surgeons. For the rest of us in the middle: Get ready.
MIT economists Daron Acemoglu and David Autor describe the thinning in the middle as job polarization. Here’s a graph from their 2010 paper that outlines the changing workforce over three decades. Each line represents a decade of jobs, from low skill required (left) to high skill (right). The general “U” shape of the chart shows growth of low-skill and high-skill jobs. Note the wasteland of steady job attrition in the middle. This trend also accounts for some of America’s sharp rise in income inequality in the Internet age.
The polarization study was preceded by a prescient 2003 MIT paper. It suggests that instead of thinking about jobs as low-skill versus high-skill, we should really be thinking more in terms of routine (assembly work, bank telling, data entry) versus non-routine (scientific discovery, managing people, political persuasion). In today’s workforce, anything routine is replaceable.
The dividing line between what’s routine and what’s not is shifting. When the 2003 study was published, truck driving and medical diagnosis fell clearly into the non-routine category. That’s no longer the case. According to the Oxford study, taxi drivers and chauffeurs are now among the most at-risk positions, facing an 89 percent probability of becoming automatable. (Considering that Google’s fully autonomous cars have already clocked more than half a million test miles, 89 percent may be lowballing it.) Similarly, in health care, IBM’s powerful artificial intelligence machine, dubbed Watson, is already getting paid to dispense advice on complex medical treatment decisions.
As a journalist, I see this shift in the algorithmic tools available to help determine where stories appear on a webpage — once the exclusive domain of top newspaper editors. More threatening is the emergence of computer-generated news, like the earthquake story in yesterday’s L.A. Times or the Narrative Science earnings previews that Forbes publishes. Compared with some of my early fumblings as a business journalist, the robots win.
Agriculture: how about zapping individual weeds in a field of cabbage? There’s a robot for that. Telemarketing? Talk to the cyborg. Stock trading algorithms are nothing new, but automated loan officers are. The Oxford study gives a 98 percent probability that human loan officers will be obsolete. Car dealers, we can hope, will soon follow.
To beat the machines, we’ll have to get creative. Choreographers, curators and art directors are relatively safe, according to the Oxford study. So are jobs that require social perception: counselors, clergy, nurses (though digital nurses like this one are coming soon). Combine creativity and social perception, and you’ve an invincible job description: poet.
What’s more human than poetry? It’s also an extremely lucrative profession, unless you compare it with anything else. There have been attempts at robot poetry, but the results are impenetrable.
Oops, that was a poem by celebrated E. E. Cummings. Here’s a sampling produced by an actual machine, the New York Time haiku generator
Like algorithmic news and robot companions, robot poetry is imperfect. But that’s how these things start. Technology that first comes across as a gimmick — early brick-size cell phones, 1980s talking computers — in a few years evolves into a trusty sidekick. Give it a few years more, and it will stab you in the back and take your job.
“Technological unemployment,” warns one economist, could result from “our discovery of means of economizing the use of labor outrunning the pace at which we can find new uses for labor.” That economist was John Maynard Keynes, speaking at the start of the Great Depression. Keynes also predicted, correctly, that the “phase of maladjustment” from technological unemployment would be temporary, that as machines relieved the toils that tax our bodies, they would free us up to stretch our minds.
For the last century, that’s been true. But what happens when the brain jobs are lost to machines too? We will either need to augment our physical and intellectual selves with our machines or stand idly by as our inventions exceed our aptitudes. As with countless generations before us, we are headed toward a future where much of what we consider to be “work” will be done by machines. Perhaps, as before, we will create new jobs and new pursuits to keep us busy. We can hope.
In the meantime, avoid technological unemployment by doing what the robots can’t. Break routine. Do something unexpected. Figure out what makes you uniquely human and nourish it.