Robots Aren’t About to Steal Your Job Just Yet

This piece originally appeared on Monster.

The robots are coming for your job. They’re coming in droves. And they’re not going to stop until they are sitting at your desk, performing your tasks, and drinking from your coffee mug.

This is what we’ve been hearing for the last few years, anyway—and all sorts of studies seem to back up the robot labor revolution as an inevitability.

Among the more notable reports: the Oxford-Martin study in 2013, which enumerated in graphic detail all the jobs that would be handled by the ‘bots in the near future. (Apologies to telemarketers, mathematical technicians and tax preparers.) And more recently, the World Economic Forum reported January that the world’s economy would lose a net totally of 5 million jobs to automation by the year 2020. (Apologies this time to office managers and admin workers.)

Job seekers don’t seem to be freaking out—in a poll conducted in 2015 by Monster, 63% of respondents said they weren’t concerned about a robot taking their job. But should they be, what with this growing body of evidence?

We decided to put the question to one of the top American experts on the topic: Massachusetts Institute of Technology professor emeritus Frank S. Levy, a labor economist, who has been studying the effects of technology on jobs for two decades. Levy is the author of the 2004 book The New Division of Labor: How Computers Are Creating the Next Job Market. More recently, he’s the co-author of a working paper published last December entitled, Can Robots be Lawyers? which examined the ability of artificial intelligence to do various lawyer-ly tasks based on actual lawyers’ billable hours.

What he told us surprised us: For the most part, he sides with the poll respondents.

R2-D2, J.D.? T.B.D.

Machines, software and robots can do a variety of tasks, even complicated ones, he says. But humans have one big advantage: They can roll with the punches, improvising in situations that are not utterly constant.

“If you start throwing curve balls to a computer it can get pretty hairy, pretty quickly,” he says.

While researching the work of lawyers, Levy and his co-authors determined that computers could make a dent when it came to classifying documents in the discovery process—when lawyers have to turn documents over to the opposition in a case.

This is because of the rote, redundant nature of the paperwork typically involved. Once a human can figure out the pattern of the words in the documents, you can write a machine-learning algorithm to do the rest, he says. So, as long as the right people are setting up the right algorithm, they can sit back and let the software do the rest. Bye-bye billable hours.

“Once you’ve trained the algorithm, you can turn it loose on all the rest of the documents,” he says.

But Levy said that artificial intelligence would not be particularly savvy when it comes to doing some of the other things lawyers routinely do. As underscored in a recent post in The New York Times Bits blog, lawyers are far from simple paper pushers. They have to think on their feet, provide counsel and perform many other tasks, often having to do so with wide-ranging emotion. These are things humans are pretty darn good at.

So in essence, he found that a machine could help a lawyer do his or her job better, more efficiently by removing some of the busywork. But a computer still can’t litigate in court.

Humans 1, Cyborgs 0

So lawyers are safe, but what about these 5 million jobs the WEF says will be lost to the machines?

Levy says high numbers get headlines, which can explain away some of the gaudy stats. But throughout our conversation, Levy continued to come back to the idea that beyond being able to decipher and repeat patterns, machines can actually do very little to compete with humans as it stands today.

He uses the example of the job of a concierge: That person is paid to be nimble, to be able to answer questions on the fly using knowledge of the area, factoring in the inquirer’s tastes and possibly any promotional opportunities as well—all while flashing a sincere smile.

“If you have a job that requires a lot of thinking on your feet, where things are different than what you’re used to [each day], that’s something that’s going to be around for quite a while,” Levy says.

Consider also self-driving cars. All the talk is that the technology exists already—but without digitally mapped roads, the self-driving cars of today are pretty useless, Levy says.

“If you were to wake up one morning and the Cambridge, Massachusetts roads department decided to remove the Mass. Ave. Bridge, but they didn’t tell the car and they didn’t update the map, you don’t know what the car is going to do,” he says. “The truth is we don’t have self-driving cars at all.”

Levy says jobs that involve a lot of human contact, a lot of working with other people in situations where you have to be spontaneous, are the most difficult ones for a machine algorithm to replace.

So a person in a job that involves technical skills and higher-level thought processing or interpersonal communications-based skills should feel pretty safe for a while.

Because computers don’t know how to be human…yet.