The Robots Are Coming!

by Andres Oppenheimer

  So what are my conclusions after interviewing some of the world’s leading futurologists? One of the main ones—in addition to others that I’ll address in the final chapter—is that, yes, some jobs will cease to exist, but the vast majority won’t disappear. Instead, they’ll be transformed. Much of what we are doing today will be done by intelligent machines, and that will require us to constantly update our skills. In many cases, we will have to reinvent ourselves. There’s no doubt about it: many of us will lose our jobs. The big question for each of us, and for society in general, is how traumatic the transition to an automated world will be.

  I am a techno-optimist, at least in the long term. Automating the labor force will bring about a tremendous increase in productivity, drive down costs, and grow the economy to the benefit of all. But when it comes to the short term, until the increased productivity is translated into massive national revenues, and until we are able to agree on how to distribute these revenues, many workers will be unemployed and marginalized. As much as the techno-optimists are correct in saying that new technologies will create many indirect jobs, there is one incontrovertible fact: the manufacturing companies of the past employed many more people than the data companies of today, and they offered many more social benefits than current service jobs do. And it will take us quite some time to change our current culture that glorifies work, as Bostrom proposes, so those who lose their jobs don’t also lose their self-esteem and sense of purpose. Can we really be happy if we’re unemployed or if we are all shining one another’s shoes? Not at the moment.

  Transitioning into a more automated world will be harsh, and it will create even more social turbulence than what we have recently seen in many industrialized countries. The new populist, nationalist, and sometimes racist movements that won elections in the United States and Europe erroneously blame job losses and wage depreciation on immigration. But these are false claims: jobs aren’t being threatened by immigrants, but by automation. In fact, illegal immigration in the United States has fallen dramatically in recent years. Apprehensions of undocumented immigrants along the Mexico border totaled 310,000 in 2017, down from 1.7 million in 2000, according to U.S. Border Patrol statistics. As professors Daron Acemoglu of MIT and Pascual Restrepo of Boston University have shown, President Trump won the 2016 presidential election thanks in no small part to the growing sense of discontent with automation in some northern states. “The swing to Republicans between 2008 and 2016 is quite a bit stronger in commuting zones most affected by industrial robots,” Acemoglu noted.

  We will have to come up with educational and social solutions as soon as possible to prevent short-term technological unemployment from becoming a long-term social disaster. If we don’t, we will see a growing backlash against automation, and against technology in general, that will adversely affect our economies. Just as we saw an anti-globalization movement in the 1990s and 2000s, we are likely to see an anti-robotization movement in the 2020s and 2030s. The protests against Uber by taxi drivers in some of the world’s big capitals, or against Facebook by Americans concerned about what they see as the company’s failure to protect their privacy, are just a prologue of what we may see happening in coming decades.

  We will have to place the social impact of technology at the center of our political agenda, and we will have to be prepared to adapt as quickly as possible to the coming transformations, lest they catch us by surprise. And we will have to find new solutions, such as a universal basic income in exchange for community work, in order to avoid massive social conflicts. In the chapters that follow, we will look at some of the changes that will take place in a number of different occupations—ranging from law to medicine, finance, commerce, manufacturing, cultural industries, and journalism—and discuss how we can adapt to, and perhaps even improve our lives in, an increasingly automated world.

  * “It worked great; everyone had a wonderful time,” Osborne assured me with pride. “We asked each of the wedding guests to fill out an online questionnaire asking them about their personality and interests, and we asked the algorithm to seat those who had similar interests together and to keep them separated from people they already knew.”

  2

  THEY’RE COMING FOR JOURNALISTS!

  THE FUTURE OF THE MEDIA

  MIAMI, FLORIDA

  Journalism is far from being one of the industries that employs the most workers, but I’m going to start with it because it’s among those that are being most affected by automation, and it also happens to be the one I know best. The birth of the Internet, when people began to read the news online for free, struck journalism like a tsunami and wiped away tens of thousands of jobs. In the United States alone, according to the Department of Labor, the number of reporters, correspondents, and editors across the print, radio, and television platforms has fallen by 38 percent in the last decade, from 66,000 to 41,000 employees. In other words, 25,000 journalists have lost their jobs in the past ten years.

  But what’s even more worrisome is that, during this same period, the increase in jobs among exclusively digital journalists wasn’t even close to offsetting the loss of jobs in print magazines and newspapers. According to a study published by the Columbia Journalism Review based on data from the U.S. Bureau of Labor Statistics’ Occupational Employment Statistics program, “in 2005, for every one digital-only journalist, there were 20 newspaper journalists. In 2015, for every one digital-only journalist, there were four newspaper journalists.”

  Exclusively digital media outlets had their boom in the early 2000s, when many venture capitalists bet large sums of money on Internet news websites. But most of these websites failed to find a business model for making money off their massive audiences. In addition to growing competition from platforms such as Twitter, Facebook, and other social media networks that were increasingly monopolizing both readers’ attention and advertisers’ dollars, digital newspapers and magazines were faced with the challenge that a good number of their readers were spread not only across the country but across the globe. Traditional advertisers didn’t care about these websites’ far-flung readers. Retail stores, supermarkets, and car dealerships, for example, were only interested in reaching the local market. A grocery store in Brooklyn didn’t care if a news website had readers in South Korea: it wanted to reach Brooklyn residents, or—even better—people living within a few blocks from its location. After the novelty wore off, many exclusively online websites went bankrupt.

  And in 2018, when newspapers and magazines were once again beginning to balance their books by spreading the news through Facebook to that social network’s 2 billion people, they woke up to several new, completely unexpected blows. Facebook cofounder and chairman and CEO Mark Zuckerberg had announced that the company’s new algorithm would be giving higher priority to messages between family and friends about childbirths, marriages, and other family events than to news. Headlines and video footage from journalistic sources would be relegated to a second-tier category. In other words, Facebook was getting back to its roots: a social network devoted to exchanging messages between friends and family members. The company had most likely noticed that young people were increasingly switching to other social networks that didn’t bombard them with what could be seen as boring news, or perhaps headlines didn’t keep them on the site as long as conversations with friends about more trivial topics.

  Zuckerberg may have decided to downgrade the flow of news on Facebook because of the widespread criticism of his company after U.S. intelligence agencies disclosed that Russian hackers linked to the Kremlin had used the social network to spread fake news and help Trump win the 2016 presidential elections. There was also speculation that Facebook’s new strategy was part of the company’s attempt to enter China, a key market that Zuckerberg’s company had not yet been able to approach because of China’s censorship laws. Whatever the reason, the fact is that Facebook’s decision shook the journalistic world to the cor
e because many media outlets were relying on that platform for nearly 40 percent of their readership. As The New York Times noted, “The algorithm changes will almost certainly affect ad-supported media companies like BuzzFeed and Bustle, which depend in part on Facebook for eyeballs.” In addition, it said, both The New York Times and The Washington Post “will also have to confront likely declines in traffic.” The entire media industry was hit hard by the change in the Facebook algorithm, and all projections seemed to indicate that jobs in journalism would continue to fall.

  THE DISAPPEARANCE OF DESIGNERS, EDITORS, AND TRANSLATORS

  I’ve witnessed with my own eyes the shrinkage of American newsrooms, as many of my colleagues’ jobs were terminated in recent years. Google’s search engine eliminated most of the library researchers who previously helped reporters find or double-check their data. Electronic pagination has replaced most newspaper layout artists, who used to be an ever-present feature in newsrooms with their large drawing tables and T-squares where they used to design each page. The spell-check programs we all have on our computers have done away with many copy editors. And lately, Google Translate has begun to eliminate the jobs of translators, who occupy many seats on the international desks of newsrooms.

  The accuracy of automated translations has improved dramatically, to the point that machine translation programs can now translate documents in less than a second and require relatively little human editing. For decades, at the Miami Herald, I used to use several outside translators. I would write my columns in English, send them out to be translated into Spanish, and would then check the Spanish version before publication. But one day in late 2016, something unexpected happened that led me to discover the new automated translation programs: the translators weren’t available—they were on vacation or too busy with other assignments—so a colleague in the newsroom told me she would run my column through Google Translate. My first reaction was to laugh; I had tried Google Translate in the past, and it was a disaster. But when I received the text this time, I was amazed by how good it had become. While it had some mistakes, the translation was fairly well done, and editing it took me roughly the same time it used to take me to edit a human translator.

  When I asked what had happened, a Google engineer told me that they had actually just started using artificial intelligence for its automated translations between English and Spanish, French, German, Chinese, Japanese, Korean, and Turkish, with spectacular results. She added that the program had improved more in recent weeks than it had since its beginnings more than a decade earlier. And, she assured me, thanks to artificial intelligence, which allows the program to constantly learn from its mistakes, the translations would rapidly become increasingly better.

  A few weeks later, The New York Times ran an extensive story on the remarkable improvements made by machine translation programs thanks to artificial intelligence. “The A.I. system had demonstrated overnight improvements roughly equal to the total gains the old one had accrued over its entire lifetime,” the article read. Shortly thereafter, I found myself using Google Translate rather than outside freelance translators. I had mixed emotions about it. On the one hand, I was amazed by the technological advance, which made my life easier. I no longer had to wait long hours for a human translator to get my columns and return them to me in Spanish. Now it could be done in a matter of seconds. But on the other hand, what was the human cost of this new technological marvel? What had happened to the person or people who used to translate my columns?

  THE INTERVIEWS IN THIS BOOK WERE TRANSCRIBED BY A MACHINE

  Several other parts of my job have been automated as well over the past two years. When I started doing research for this book five years ago, I did things the same way I’d been doing them for decades. I’d record my interviews and then transcribe them myself. This process was extremely tedious and often lasted for hours on end, but it was faster, more effective, and cheaper than hiring a freelancer to transcribe the tapes or digital recordings. But in 2016 I discovered online transcription services like Rev.com—which primarily use a combination of automated programs and freelance editors from around the world—and I started using them. At first they charged between eighty cents and a dollar per minute of audio, but they simplified my life immensely.

  Since then, I haven’t transcribed a single recorded interview. In 2017, I started using Trint.com, an automated transcription service that worked even faster: it could have a transcript back to me in less than an hour, at a cost of just twenty cents per minute of audio, though not quite as precise as those edited by humans. A year later, Temi.com offered transcriptions “in five minutes” at a cost of only ten cents per minute. These programs are often cheaper because instead of using human editors, they place the burden of correcting and polishing the text on you yourself. Many of them allow you to mark with your cursor the words or phrases that don’t seem correct, and listen to that precise part of the original audio, which makes the process of correction relatively simple.

  Now, whenever I do a face-to-face or a telephone interview, I record it on my smartphone and email it directly to a transcription service. Then I’ll go to the gym or the supermarket, and when I get home two hours later, the transcription is waiting for me in my inbox, ready to be used. This leaves me much more free time and mental energy for doing further research, conducting other interviews, or just running errands. But on the other hand, this new technology undoubtedly has a human cost. What will become of the transcriptionists who do this work for a living? Much like many translators, they may find themselves looking for a new line of work.

  “COMPUTER KEYBOARDS WILL DISAPPEAR WITHIN THE NEXT FIVE TO TEN YEARS”

  Journalists and almost everyone else will be using new automated writing programs going forward. The very act of writing itself will become automated as people begin using speech-to-text conversion programs instead of typing on a keyboard as we do now. I admit that I don’t see myself dictating my books or articles into a machine any time soon, but if you look at any American child under the age of thirteen, you’ll realize the direction in which things are going: they have spent their entire lives asking questions orally of virtual assistants like Siri or Alexa, as opposed to typing them into Google’s search engine.

  Technology companies are already anticipating that writing, as it’s done today, will soon be considered old-fashioned. As Claudio Muruzábal, president for SAP Latin America and the Caribbean, told me, “computer keyboards will disappear within the next five to ten years.” He added, “Just like you are now used to asking Alexa to play a song, you will get used to dictating your articles and emails.”

  When I told him that while I don’t see myself giving up my keyboard any time soon, I was already using automated services to transcribe my interviews, he explained that very soon the whole process of what I’m doing will become much easier. “The different systems will be more integrated,” he said. “Now you interview someone, you record it, you email it from your phone to your computer, and from there you send it to a transcription service. After you get the text back, you send it to another website to translate it, and maybe to another one to edit it. The real revolution is going to be when you can do all of this at once, in one place, with one single piece of integrated software.” In other words, soon enough I’ll be able to conduct an interview, hit a button on my cell phone, and say, “Transcribe, translate, and edit this.” “The entire process is going to be much friendlier,” Muruzábal told me. “You won’t need to jump from one system to another. The phone will do it all by itself.”

  Technology buffs say this whole process of automation will substantially improve the quality of journalism because—just as today’s word processors suggest synonyms—soon-to-appear software programs will be able to offer ideas for enriching our articles. For example, they’ll be suggesting historical references, comparisons with other people and countries, and even sources. If I’m writing an article about the
future of computer keyboards, the software itself will take note and a list of experts I may want to interview will pop up on the screen, along with their respective contact information. A number of technology companies are already working on augmentation journalism, which will give more tools to journalists to do their job. Much like programs for transcription, translation, editing, and synonym search engines before it, the new augmentation journalism software programs will undoubtedly allow reporters to devote more time to research and analysis. But the question remains: What will happen when ever more intelligent computers start doing investigative reporting and write opinion pieces as well?

  WHO SAID THAT POLITICAL AND ECONOMIC ANALYSTS ARE SAFE?

  Although tech company spokespeople claim that intelligent machines will never replace opinion writers or investigative reporters because those jobs require a sixth sense that computers lack, there are important voices that are beginning to contradict that assertion. Andrew McAfee and Erik Brynjolfsson, the MIT professors who coauthored The Second Machine Age, have concluded that the so-called experts—whether prominent journalists, economists, or politicians—are much more prone to mistakes than computers. “We need to rely less on expert judgments and predictions,” they wrote, adding that a well-designed, tried-and-true computer program “tends to perform as well, or better than, human experts making similar decisions. Too often, we continue to rely on human judgment when machines can do better.”

  The authors give a number of examples of algorithms that are proving to be more accurate—and more judicious—than human beings, including the case of the Israeli traffic court judges discussed in the previous chapter. Unlike those judges, they point out, algorithms don’t get upset when they’re hungry or get more punitive as the day goes on. A similar study, conducted in the United States, showed that judges who graduated from a well-known state university issue far more severe sentences the morning after their football team suffered an unexpected loss.