One Birth of a Suboptimist The machine's danger to society is not from the machine itself but from what man makes of it. --Norbert Wiener When I told people I was writing a book about AI and automation, I got two types of reactions. My more tech-skeptical friends and colleagues generally approved. They'd been hearing the gloomy predictions about job-killing robots, and it worried them. They wanted me to confirm their fears of a looming automation crisis, and their suspicions that even if AI didn't cause mass unemployment, it would bring new harms--creepy surveillance, runaway self-driving cars, brain-melting social media apps--that would outweigh its benefits. Among people in Silicon Valley, though, more typical was the reaction I got from Aaron Levie, the CEO of the enterprise software company Box. "Oh God," he said. "Please tell me you're not writing one of those books that make everyone terrified and depressed." Levie is one of the optimists who believe we're all worrying over nothing. He's annoyed by what he sees as alarmist media coverage of AI, and he doesn't see automation as a major threat that requires an urgent solution. As I said, this is not a book about whether or not robots will destroy jobs. But that question is the place the AI discussion typically starts, and it's where people tend to get stuck. So I wanted to consider arguments like Levie's and develop my own mental model, grounded in evidence, that could help me figure out which AI fears were reasonable and which were probably overstated. First, I talked with a number of AI optimists and distilled what they told me into three big claims. 1. "We've been here before, and it turned out fine." First, the optimists argue, there are hundreds of years' worth of evidence that new technology helps us, rather than hurting us. We didn't suffer mass unemployment after the introduction of the assembly line, the invention of the lightbulb, or the creation of the internet because those technologies created new jobs to replace the ones they destroyed. And if you look at today's economic data, they say, there's no evidence that AI and automation are having a catastrophic effect. (In fact, the U.S. unemployment rate is near historic lows, and productivity growth has actually slowed over the past several decades, which isn't what you'd expect to see if mass automation were happening.) 2. "Automation doesn't destroy jobs. It makes them better." Second, the optimists say, automation doesn't replace workers; it assists them, by taking care of their mundane drudgery and allowing them to focus on more creative tasks. Nobody becomes a lawyer because they like doing legal research and logging their billable hours--and if those tasks can be done by AI, it will free lawyers up to spend more time doing the parts of their job they actually enjoy. The same goes for doctors who hate filling out medical history charts, real estate agents who hate creating listings, and teachers who hate grading papers. All of these people's lives improve when such tasks get automated. 3. "Robots won't cause mass unemployment because human needs are limitless. In the future, we'll come up with new jobs we can't even imagine today." The third argument the optimists make is that the pessimists are, essentially, failing to use their imaginations. Just a few decades ago, the vast majority of the biggest companies in the world--including Facebook, Google, and Amazon--didn't exist. Until recently, there was no such thing as a YouTube marketer, a VR film editor, or a professional esports coach. We've always been good at coming up with new, interesting tasks for ourselves as technology makes new things possible. And the optimists think our bottomless desires will keep us employed for a very long time. After months of research and reporting, the position I landed on was neither total optimism nor total skepticism. It's more like "suboptimism"--a word I made up to convey my belief that while our worst fears about AI and automation may not play out, there are real, urgent threats that require our attention. If I had to rank myself on a 1-to-10 worry scale, with 1 being "AI will cause no problems whatsoever" and 10 being "AI will destroy us and everything we hold dear," I'd probably hover around a 7. I am much more optimistic--maybe a 2 or a 3--when it comes to the technology itself. I think that well-designed AI and automation could radically improve many people's lives. Self-driving cars and trucks alone could save hundreds of thousands of lives a year from fatal accidents, which would be a good thing even if it resulted in truckers and taxi drivers losing their jobs. Precision medicine--an AI-assisted approach to disease treatment and prevention--could help us find new, life-saving treatments. There are a million more ways AI could improve our lives, from the mundane (more accurate credit card fraud detection) to the fun (new forms of adaptive, AI-powered videogames). I am much more worried, though--maybe an 8.5 or a 9--about the humans who are designing and implementing this technology. I've seen that AI is being eagerly embraced by profit-hungry executives and starry-eyed entrepreneurs, many of whom are deliberately underselling the risks of harm and displacement resulting from their products. I know that many bosses are using AI and automation to micromanage and surveil their employees and that, as a result, many workers' jobs are getting worse and more precarious instead of better and more secure. I worry that flawed and biased data sets will result in flawed and biased AI, and that the rush to adopt these systems will disproportionately hurt marginalized groups, including women and racial minorities. I think that AI is, and will continue to be, used by authoritarian regimes to commit human rights abuses. I don't have a ton of faith in the tech industry's ability to self-regulate, and I'm concerned that politicians won't show up until it's too late. I'll admit that part of my suboptimism is a gut reaction, informed by my years of covering the technology industry and watching it fall short of its ideals. But it's also based on what I found out about the case for AI optimism--and why it might be weaker than the optimists think. "We've been here before, and it turned out fine." The first thing I learned is that many optimists have not done their history homework. Because while many of them are eager to claim that the Fourth Industrial Revolution will be great for humankind, they rarely note that, for many people, the first three weren't that great. In the eighteenth and nineteenth centuries, as the United States and Britain industrialized, workers routinely faced brutal conditions in overcrowded and unsanitary factories, and were often subjected to long hours and horrendous exploitation. Some of the harshest conditions were faced by child laborers, who were paid pitifully small wages, packed into squalid boardinghouses, and abused when they failed to meet their bosses' standards. Local newspapers were filled with hellish descriptions of life in the working class, like this 1839 report from a Boston newspaper about a boardinghouse for girls who worked at a textile factory: The young girls are compelled to work in unhealthy confinement for too many hours everyday . . . their food is both unhealthy and scanty . . . they are not allowed sufficient time to eat . . . they are crowded together in ill-ventilated apartments in the boarding-houses of the corporations . . . in consequence they become pale, feeble, and finally broken in constitution. The Second and Third Industrial Revolutions went more smoothly for workers, in part because of the labor protections that emerged out of the First Industrial Revolution. But they still had plenty of problems. The Second Industrial Revolution created the Gilded Age, a period of American history during the late nineteenth century that was marked by staggering corruption, bloody labor clashes, bitter racial animus, and soaring income inequality. The advances in digital technology during the Third Industrial Revolution made it possible for multinational companies to offshore and outsource jobs to countries with lower labor costs--a great thing if you're an executive or a shareholder, but less helpful if you're a manufacturing worker in the Midwest. Capital owners generally do well during periods of technological change. But the benefits of progress often take longer to materialize for workers. Gregory Clark, an economic historian at the University of California, Davis, estimates that British workers' real wages didn't rise until more than fifty years after the beginning of industrialization--meaning that by the time the Industrial Revolution paid off for the workers who actually lived through it, most of them were already retired or dead. Excerpted from Futureproof: 9 Rules for Humans in the Age of Automation by Kevin Roose All rights reserved by the original copyright owners. Excerpts are provided for display purposes only and may not be reproduced, reprinted or distributed without the written permission of the publisher.
