The Robots Are Coming!

by Andres Oppenheimer

  TECHNOLOGICAL UNEMPLOYMENT

  Before Frey and Osborne, a number of economists had already been warning about the potentially disruptive effects of robots and artificial intelligence. Their argument was that unlike in the past, technology today is growing exponentially, or at an increasingly accelerated rate. According to what’s known as Moore’s law—which is based on an article written by Gordon Moore, a cofounder of Intel, in 1965—the processing power of computers increases by 100 percent every eighteen months. A decade from now, computers will be 10,000 percent more powerful than they are today. And according to Lin Wells, in his November 1, 2014, essay “Better Outcomes Through Radical Inclusion,” this will change the world much more significantly than it has in the past, because technological acceleration will be taking place not only in the field of computing but also in robotics, biotechnology, and nanotechnology.

  By 2003, economists like Maarten Goos and Alan Manning of the London School of Economics were already warning that the exponential advancement of technology was beginning to polarize the labor force, and that the only workers whose jobs would be safe would be the better and the lesser educated. The majority of those in the middle, they argued, would be left without work. In their paper “Lousy and Lovely Jobs: The Rising Polarization of Work in Britain,” Goos and Manning predicted that smart machines would soon be able to react to unexpected situations, allowing them to replace not only those who do routine work but also those who perform complex tasks. And in 2011, MIT economists Erik Brynjolfsson and Andrew McAfee had written a book titled Race Against the Machine, in which they also warned about the growing advantage that intelligent machines had over humans.

  However, Frey wanted to dig a bit deeper and proposed a study of the 702 jobs listed by the U.S. Department of Labor that would rank them in accordance with their relative risk of being eliminated by new technologies. He contacted Michael Osborne, a young engineer specializing in robotics and artificial intelligence who taught just a few blocks away at the Department of Engineering Science, and asked him for help creating an algorithm that could classify jobs according to their risk of disappearing in the near future.

  Osborne, who had yet to turn forty when I interviewed him, is a tall, thin man who seemed to have permanently disheveled hair. He has the classic look of an absent-minded genius. Not only does he teach his students at Oxford how to create algorithms, he also uses algorithms to guide his private life. When I asked him, only half-jokingly, if he would describe himself as a nerd, or as somebody who used technology to manage the most trivial details of his personal life, he smiled affirmatively. He told me, as an example, that he had even created an algorithm to plan his wedding party, organizing seating arrangements for the guests according to their ages, occupations, and interests.*

  For their study on the future of jobs, Frey and Osborne created an algorithm into which they fed examples of what kinds of jobs were already being displaced by automation, and asked it to compare them with the 702 jobs on the U.S. Department of Labor list. The results surprised even themselves, as they told me in separate interviews.

  THE RANKING OF THE MOST THREATENED JOBS

  Frey and Osborne’s algorithm produced a ranking that begins with jobs that have a 99 percent chance of being replaced by robots, drones, autonomous vehicles, and other intelligent machines. According to their study, this most-at-risk category includes telemarketers (who in many countries have already been replaced by robocalls), insurance underwriters, new accounts clerks, library technicians, and cargo and freight agents, all of whom are at risk of being replaced by computer programs capable of accumulating and processing data and making future projections much more accurately than humans.

  The jobs with a 98 percent chance of being replaced in the next fifteen to twenty years include administrative employees, bank loan officers, and many insurance agency workers, whose routine tasks can be easily replaced by artificial intelligence. Interestingly enough, also included in this category are sports umpires, whose calls on controversial plays will be increasingly replaced by drones and video replay systems that can make much more precise judgments than humans.

  Jobs with a 97 percent risk of being killed in the near future include phone operators and store salespeople, who in many cases are already being replaced by e-commerce or humanoid robots that can answer questions from customers in much the same way that Siri, Alexa, Google Assistant, Cortana, and other virtual assistants can respond to whatever we want to ask them. Other jobs that, according to this same study, run a 97 percent risk of disappearing include real estate brokers, who like travel agents are being replaced by Internet websites that allow us to virtually visit homes for sale, and cashiers, who are already being replaced by self-checkout stations at many stores and supermarkets.

  To my surprise, another kind of job that is on the way out is that of restaurant waiters. How could that be? I asked the study’s authors. Well, it’s already happening, they replied. In many U.S. fast food restaurants, including McDonald’s, Chili’s, Applebee’s, and Panera, it is already common to order and pay for your food with a tablet or on your own smartphone. Soon enough, robots won’t just be replacing the waiters, but will be replacing many of the chefs as well.

  “We were surprised to see that servers were among the most likely people to have their jobs automated,” Osborne told me. “When we designed the algorithm, which learns from examples just like children do, we ran thirty-five examples of jobs we thought weren’t very susceptible to automation, followed by another thirty-five we considered to be more at risk. We included servers with the latter group, because we thought that personal engagement in restaurants was an essential part of their work and couldn’t be replicated by a machine. But when the results were calculated, it turns out we were wrong, and that waiting tables is a routine job that could easily be automated.”

  Among the jobs listed as having a 96 percent risk of disappearing are chefs—who are already being replaced by robots, as I witnessed in Japan’s sushi chains—secretaries and administrative assistants, receptionists, information clerks, and bank or department store tellers, all of whom are being increasingly replaced by tablet computers or humanoid robots. According to the study, other jobs at high risk of replacement include taxi drivers, truck drivers, and messengers—who will be displaced by autonomous vehicles—tourist guides, dental technicians, pharmacy technicians, butchers, legal clerks, and accountants.

  THE JOBS THAT WILL SURVIVE ARE THE ONES THAT ARE HARD TO EXPLAIN

  According to Frey, “Anything that relies on storing and processing information, something that computers are relatively good at, is highly automatable. That includes anything from basic administrative back-office work to basic assembly production jobs, to even some types of sales and service jobs. The list is endless.” And according to Osborne, “The new thing about technology is that it’s able to substitute routine clerical work the way it used to be able to replace routine mechanical work.” When I asked him if he could give me a general rule about who is most at risk of losing a job due to automation, he responded that “the probability of automating an occupation is closely related to the level of skill or education….Highly skilled people tend to be sufficiently well-equipped to keep pace with technological change, to be able to move to whatever new occupations emerge, and to use new technologies, whereas at the bottom of the spectrum, we see the kind of skills that can be replaced entirely by technology.”

  Other futurologists I interviewed agreed that academic training and skills like creativity, originality, and social and emotional intelligence—which should also be taught at colleges and universities—will be key to the jobs of the future. And academic training will have to go far beyond current one-dimensional careers like law, medicine, and business administration. New careers will be increasingly interdisciplinary and will include technological, critical thinking, and problem-solving capabilities, as well as social skills. Further
more, tertiary education will be intermittent, in the sense that it will be constantly updated over time. For example, today’s dermatologists study medicine, specialize in dermatology, and spend a significant part of their time examining brown spots on their patients’ skin to determine whether they could potentially be cancerous. But in the future, growing numbers of patients will be using applications on their smartphones that can take a picture of a dark spot on the skin and instantly tell them whether it’s dangerous or not.

  Doctors who want to work in dermatology will have to specialize in skin cancer treatments that will increasingly rely on the diagnostic of algorithms and robots. And to do this, they will have to study previously unrelated subjects like statistics and perhaps even robotics. We will always need good doctors who provide humane treatments, who can empathize with their patients, and who can explain a diagnosis made by an intelligent machine, but those who best understand the new technologies will be the most successful. If a cardiologist, for example, also studied engineering and can manufacture pacemakers using a 3-D printer, tailoring them to each individual patient, he or she will be in much greater demand than their average colleagues.

  Anders Sandberg, a Ph.D. in computational neurosciences who was also involved in the Oxford study on the future of jobs, summed it up for me this way: “It’s quite simple: if your job can be easily explained, it can be automated. If it can’t, it won’t.” In other words, algorithms and robots are better than humans at doing repetitive, predictable tasks. An algorithm learns the same way a child does, based on examples or behaviors that are taught. If you can show another person a detailed list of the tasks he or she performs and if most of these tasks are relatively predictable, sooner or later that job will be replaced by an algorithm or a robot. And to take it one step further, just about every job has at least a few chores that will end up being automated.

  RECENT HISTORY SHOWS US THAT THE IMPOSSIBLE IS NOW POSSIBLE

  At first glance, the Oxford researchers’ algorithm seemed somewhat ruthless. Is there really a greater than 90 percent chance that the jobs of waiters, receptionists, truck drivers, tour guides, and real estate agents won’t exist in the near future? I asked them. Don’t waiters do a job that, in addition to providing a human touch, requires a high degree of improvisation and creativity when it comes to responding to unexpected situations? How would a humanoid waiter deal with a customer who wants to exchange one dish for another? And when it comes to truck drivers, will an autonomous truck be able to react quickly enough to unexpected events like a dog running across a road?

  They both responded with resigned smiles. It might be hard to believe, but this won’t be the first time in history that common jobs disappear only to be replaced by others, they said. Until recently, in many countries around the globe, agriculture employed more people than any other industry. Now it’s far from the main source of work. For example, the number of people working in agriculture in the United States fell from 60 percent of the population in 1850 to less than 2 percent in the early twenty-first century. There were only 700,000 farmers in the United States in 2010, down from 12 million a century earlier in 1910, despite the fact that the overall population had tripled during that period. But the world didn’t come to an end. On the contrary, average Americans saw their standard of living rise. If agriculture could be revolutionized by tractors and computers, thus forcing millions of people to seek new occupations, they argued, why not expect something similar from the current revolution in robotics and artificial intelligence?

  It’s a fair question. Plus, recent history shows us that skepticism is a bad method of predicting the future. Many ideas that seemed preposterous just a decade ago are now part of our daily lives, which we now accept as completely normal. After all, if someone had told me ten years ago that I would have a virtual assistant named Siri on my cell phone and that I could ask her to tell me the capital of Afghanistan and receive the correct answer, I would have considered the whole idea to be insane. The same thing would have happened if someone had told me an application called Waze could tell me, loudly and in the language of my choice, the shortest and quickest route to my destination, or that I could buy plane tickets online, or that supermarket cashiers would be replaced by self-checkout stations. But all that—in fact, much more—is already happening.

  Technological advances happen so quickly that they numb our capacity for amazement. New innovations might stun us for a few minutes, but soon enough we incorporate them into our lives as if they’ve always been with us. Not many people would remember that the iPhone as we know it today didn’t exist before 2007, or that Waze—the app that calculates traffic and driving time to a destination—went into worldwide use only after the company was bought by Google in 2013. But today many of us wonder how we had ever lived without them.

  IN JAPAN, SOME HOTELS ARE ALREADY BEING RUN BY ROBOTS

  When I traveled to Japan to conduct interviews for this book, the first thing I did was book a room at the Henn na Hotel, which had attracted a great deal of attention for becoming the first hotel in the world operated by robots. I had read that the concierge, receptionists, bellhops, and even the waiters at the hotel are fully automated, and I wanted to experience firsthand how that worked. When I arrived at the hotel, located in a suburb near Tokyo’s Disney Resort, I found that, yes, several of the employees had been replaced by robots and intelligent machines…although, as you will see, some of them were not terribly smart, or at least were in the process of learning.

  Behind the reception desk were two giant robotic dinosaurs that looked quite real and were wearing green hats like those used by concierges at traditional Japanese hotels. Apparently excited to see me enter, they started making guttural grunts and swinging their heads from side to side. Since there were no human beings behind the counter, I had no choice but to deal with them. As I later learned, other branches of the hotel chain have robotic receptionists who look like men and women, but because of its proximity to Disney, the Tokyo branch had opted for dinosaurs to delight its younger guests. At the other end of the lobby was an aquarium about six feet across in which three large goldfish were swimming in circles, rising and diving in their irregular paths. It was only when I approached it that I realized they were also robotic, driven by some sort of internal motor.

  “Welcome to the Henn na Hotel,” the reptilian concierge said in a raspy voice, which is what you might expect from a dinosaur, though it was pleasant enough not to frighten anyone. Using his claws to point to a tablet on the counter in front of him, the dinosaur continued: “Please select your language of preference.” Once I did that, his head continuing to sway from side to side, he added, “Please place the first side of your passport on the screen and press the start button.”

  Then I asked the robotic dinosaur, “What’s your name?” But the robot didn’t seem prepared to answer that question. Shaking his head more excitedly than before, he began to growl. “Grrrrr­rrrrr­! Grrrrrrrrr! Grrrrrrr!” “You have a name, don’t you?” I pressed, rather enjoying the exchange. The giant robotic reptile became even more agitated, swaying from side to side as if desperately seeking an answer, before finally releasing a prolonged “Grrrrr­rrrrr­rrrrr­rr!” Obviously, this particular robot didn’t have much of a sense of humor. Either that, or he didn’t like being taken out of his preprogrammed dialogue routine.

  “Are you angry?” I asked him. Then the reptile with the concierge hat seemed to lose patience and said, “One moment, please. An assistant will be here shortly to help you.” A few seconds later, a young lady in the hotel’s uniform emerged from a little door behind the counter and, after greeting me with the customary Japanese reverence, explained to me with the utmost seriousness that the dinosaurs were new at their jobs and were currently only able to help guests check in and out of their rooms. In other words, these were not fully conversational dinosaurs.

  “Okay,” I said, a bit disappointed. I
placed my passport on the little screen, hit the start button, and followed the reptile’s instructions to the letter. I filled out the form on the tablet with my name, address, and signature. Then, gesturing with his claws, the dinosaur asked me to move over to the next machine “to finish the registration process.” This small tower resembling an ATM asked for my credit card and in return gave me a plastic key and a receipt with my room number on it. As soon as I took the key, the dinosaur—now visibly happy—proclaimed, “The registration process is complete.” He raised his arms, and with what looked like a smile on his face, he concluded with “Please proceed to the elevator, and enjoy your stay at our hotel.”

  I WILL BE YOUR PERSONAL ASSISTANT. LET ME KNOW WHAT YOU NEED.

  When I entered my room, number 611, I was greeted by the voice of another robot. This one looked like a big egg, was the size of a soccer ball, and was perched on top of one of the dressers. It was painted with colorful dots and had an electronic tablet on its front that showed two cartoonish eyes that looked from side to side when the machine talked. “Welcome to our hotel, Andrés-san. My name is Tapia,” it said, introducing itself and talking in a girl’s voice. “Let me show you all the things I can do for you.” Immediately, a menu popped up on the screen displaying a variety of options, including games, news outlets, weather reports, music stations, and commands for controlling the TV, lights, and air conditioning, all of which could be activated vocally. I set out to test Tapia, and indeed, when I asked her to turn off the lights and turn on the television, she did so immediately.