AI Superpowers: China, Silicon Valley, and the New World Order

by Kai-Fu Lee

WHAT THE STUDIES SAY Predicting the scale of AI-induced job losses has become a cottage industry for economists and consulting firms the world over. Depending on which model one uses, estimates range from terrifying to totally not a problem. Here I give a brief overview of the literature and the methods, highlighting the studies that have shaped the debate. Few good studies have been done for the Chinese market, so I largely stick to studies estimating automation potential in the United States and then extrapolate those results to China. A pair of researchers at Oxford University kicked things off in 2013 with a paper making a dire prediction: 47 percent of U.S. jobs could be automated within the next decade or two. The paper’s authors, Carl Benedikt Frey and Michael A. Osborne, began by asking machine-learning experts to evaluate the likelihood that seventy occupations could be automated in the coming years. Combining that data with a list of the main “engineering bottlenecks” in machine learning (similar to the characteristics denoting the “Safe Zone” in the graphs on pages 155 and 156), Frey and Osborne used a probability model to project how susceptible an additional 632 occ...

In predicting what jobs were at risk of automation, economists looked at what tasks a person completed while going about their job and asked whether a machine would be able to complete those same tasks. In other words, the task-based approach asked how possible it was to do a one-to-one replacement of a machine for a human worker. My background trains me to approach the problem differently. Early in my career, I worked on turning cutting-edge AI technologies into useful products, and as a venture capitalist I fund and help build new startups. That work helps me see AI as forming two distinct threats to jobs: one-to-one replacements and ground-up disruptions. Many of the AI companies I’ve invested in are looking to build a single AI-driven product that can replace a specific kind of worker—for instance, a robot that can do the lifting and carrying of a warehouse employee or an autonomous-vehicle algorithm that can complete the core tasks of a taxi driver. If successful, these companies will end up selling their products to companies, many of whom may lay off redundant workers as a result. These types of one-to-one replacements are exactly the job losses captured by economists using the task-based approach, and I take PwC’s 38 percent estimate as a reasonable guess for this category. But then there exists a completely different breed of AI startups: those that reimagine an industry from the ground up. These companies don’t look to replace one human worker with one tailor-mad...

THE BOTTOM LINE Putting together percentages for the two types of automatability—38 percent from one-to-one replacements and about 10 percent from ground-up disruption—we are faced with a monumental challenge. Within ten to twenty years, I estimate we will be technically capable of automating 40 to 50 percent of jobs in the United States. For employees who are not outright replaced, increasing automation of their workload will continue to cut into their value-add for the company, reducing their bargaining power on wages and potentially leading to layoffs in the long term. We’ll see a larger pool of unemployed workers competing for an even smaller pool of jobs, driving down wages and forcing many into part-time or “gig economy” work that lacks benefits. This—and I cannot stress this enough—does not mean the country will be facing a 40 to 50 percent unemployment rate. Social frictions, regulatory restrictions, and plain old inertia will greatly slow down the actual rate of job losses. Plus, there will also be new jobs created along the way, positions that can offset a portion of these AI-induced losses, something that I explore in coming chapters. These could cut actual AI-induced net unemployment in half, to between 20 and 25 percent, or drive it even lower, down to just 10 to 20 percent. These estimates are in line with those from the most recent research (as of this writing) that attempted to put a number on actual job losses, a February 2018 study by the consulting firm ...

While China will face a wrenching labor-market transition due to automation, large segments of that transition may arrive later or move slower than the job losses wracking the American economy. While the simplest and most routine factory jobs—quality control and simple assembly-line tasks—will likely be automated in the coming years, the remainder of these manual labor tasks will be tougher for robots to take over. This is because the intelligent automation of the twenty-first century operates differently than the physical automation of the twentieth century. Put simply, it’s far easier to build AI algorithms than to build intelligent robots. Core to this logic is a tenet of artificial intelligence known as Moravec’s Paradox. Hans Moravec was a professor of mine at Carnegie Mellon University, and his work on artificial intelligence and robotics led him to a fundamental truth about combining the two: contrary to popular assumptions, it is relatively easy for AI to mimic the high-level intellectual or computational abilities of an adult, but it’s far harder to give a robot the perception and sensorimotor skills of a toddler. Algorithms can blow humans out of the water when it comes to making predictions based on data, but robots still can’t perform the cleaning duties of a hotel maid. In essence, AI is great at thinking, but robots are bad at moving their fingers. Moravec’s Paradox was articulated in the 1980s, and some things have changed since then. The arrival of deep lea...

THE ASCENT OF THE ALGORITHMS AND RISE OF THE ROBOTS

The physical automation of the past century largely hurt blue-collar workers, but the coming decades of intelligent automation will hit white-collar workers first. The truth is that these workers have far more to fear from the algorithms that exist today than from the robots that still need to be invented. In short, AI algorithms will be to many white-collar workers what tractors were to farmhands: a tool that dramatically increases the productivity of each worker and thus shrinks the total number of employees required. And unlike tractors, algorithms can be shipped instantly around the world at no additional cost to their creator. Once that software has been sent out to its millions of users—tax-preparation companies, climate-change labs, law firms—it can be constantly updated and improved with no need to create a new physical product. Robotics, however, is much more difficult. It requires a delicate interplay of mechanical engineering, perception AI, and fine-motor manipulation. These are all solvable problems, but not at nearly the speed at which pure software is being built to handle white-collar cognitive tasks. Once that robot is built, it must also be tested, sold, shipped, installed, and maintained on-site. Adjustments to the robot’s underlying algorithms can sometimes be made remotely, but any mechanical hiccups require hands-on work with the machine. All these frictions will slow down the pace of robotic automation.

AI will bring that same monopolistic tendency to dozens of industries, eroding the competitive mechanisms of markets in the process. We could see the rapid emergence of a new corporate oligarchy, a class of AI-powered industry champions whose data edge over the competition feeds on itself until they are entirely untouchable. American antitrust laws are often difficult to enforce in this situation, because of the requirement in U.S. law that plaintiffs prove the monopoly is actually harming consumers. AI monopolists, by contrast, would likely be delivering better and better services at cheaper prices to consumers, a move made possible by the incredible productivity and efficiency gains of the technology. But while these AI monopolies drive down prices, they will also drive up inequality. Corporate profits will explode, showering wealth on the elite executives and engineers lucky enough to get in on the action. Just imagine: How profitable would Uber be if it had no drivers? Or Apple if it didn’t need factory workers to make iPhones? Or Walmart if it paid no cashiers, warehouse employees, and truck drivers? Driving income inequality will be the emergence of an increasingly bifurcated labor market. The jobs that do remain will tend to be either lucrative work for top performers or low-paying jobs in tough industries. The risk of replacement cited in the earlier figures reflects this.

The reality is that it will not be long until AI algorithms can perform many of the diagnostic functions of medical professionals. Those algorithms will pinpoint illness and prescribe treatments more effectively than any single human can. In some cases, doctors will use these equations as a tool. In some cases, the algorithms may replace the doctor entirely. But the truth is, there exists no algorithm that could replace the role of my family in my healing process. What they shared with me is far simpler—and yet so much more profound—than anything AI will ever produce. For all of AI’s astounding capabilities, the one thing that only humans can provide turns out to also be exactly what is most needed in our lives: love. It’s that moment when we see our newborn babies, the feeling of love at first sight, the warm feeling from friends who listen to us empathetically, or the feeling of self-actualization when we help someone in need. We are far from understanding the human heart, let alone replicating it. But we do know that humans are uniquely able to love and be loved, that humans want to love and be loved, and that loving and being loved are what makes our lives worthwhile. This is the synthesis on which I believe we must build our shared future: on AI’s ability to think but coupled with human beings’ ability to love. If we can create this synergy, it will let us harness the undeniable power of artificial intelligence to generate prosperity while also embracing our essential...

Building societies that thrive in the age of AI will require substantial changes to our economy but also a shift in culture and values. Centuries of living within the industrial economy have conditioned many of us to believe that our primary role in society (and even our identity) is found in productive, wage-earning work. Take that away and you have broken one of the strongest bonds between a person and his or her community. As we transition from the industrial age to the AI age, we will need to move away from a mindset that equates work with life or treats humans as variables in a grand productivity optimization algorithm. Instead, we must move toward a new culture that values human love, service, and compassion more than ever before. No economic or social policy can “brute force” a change in our hearts. But in choosing different policies, we can reward different behaviors and start to nudge our culture in different directions. We can choose a purely technocratic approach—one that sees each of us as a set of financial and material needs to be satisfied—and simply transfer enough cash to all people so that they don’t starve or go homeless. In fact, this notion of universal basic income seems to be becoming more and more popular these days. But in making that choice I believe we would both devalue our own humanity and miss out on an unparalleled opportunity. Instead, I want to lay out proposals for how we can use the economic bounty created by AI to double-down on what mak...

THE THREE R’S: REDUCE, RETRAIN, AND REDISTRIBUTE Many of the proposed technical solutions for AI-induced job losses coming out of Silicon Valley fall into three buckets: retraining workers, reducing work hours, or redistributing income. Each of these approaches aims to augment a different variable within the labor markets (skills, time, compensation) and also embodies different assumption about the speed and severity of job losses. Those advocating the retraining of workers tend to believe that AI will slowly shift what skills are in demand, but if workers can adapt their abilities and training, then there will be no decrease in the need for labor. Those advocates of reducing work hours believe that AI will reduce the demand for human labor and feel that this impact could be absorbed by moving to a three- or four-day work week, spreading the jobs that do remain over more workers. The redistribution camp tends to be the most dire in their predictions of AI-induced job losses. Many of them predict that as AI advances, it will so thoroughly displace or dislodge workers that no amount of training or tweaking hours will be sufficient. Instead, we will have to adopt more radical redistribution schemes to support unemployed workers and spread the wealth created by AI.

Lifelong learning via online platforms is a nice idea, and I believe retraining workers will be an important piece of the puzzle. It can particularly help those individuals within the bottom-right quadrant of our risk-of-replacement charts from chapter 6 (the “Slow Creep” zone) stay ahead of AI’s ability to think creatively or work in unstructured environments. I also like that this method can give these workers a sense of personal accomplishment and agency in their own lives. But given the depth and breadth of AI’s impact on jobs, I fear this approach will be far from enough to solve the problem. As AI steadily conquers new professions, workers will be forced to change occupations every few years, rapidly trying to acquire skills that it took others an entire lifetime to build up. Uncertainty over the pace and path of automation makes things even more difficult. Even AI experts have difficulty predicting exactly which jobs will be subject to automation in the coming years. Can we really expect a typical worker choosing a retraining program to accurately predict which jobs will be safe a few years from now? I fear workers will find themselves in a state of constant retreat, like animals fleeing relentlessly rising flood waters, anxiously hopping from one rock to another in search of higher ground. Retraining will help many people find their place in the AI economy, and we must experiment with ways to scale this up and make it widely available. But I believe we cannot count...

SILICON VALLEY’S “MAGIC WAND” MENTALITY In observing Silicon Valley’s surge in interest around UBI, I believe some of that advocacy has emerged from a place of true and genuine concern for those who will be displaced by new technologies. But I worry that there’s also a more self-interested component: Silicon Valley entrepreneurs know that their billions in riches and their role in instigating these disruptions make them an obvious target of mob anger if things ever spin out of control. With that fear fresh in their minds, I wonder if this group has begun casting about for a quick fix to problems ahead. The mixed motivations of these people shouldn’t lead us to outright dismiss the solutions they put forth. This group, after all, includes some of the most creative business and engineering minds in the world today. Silicon Valley’s tendency to dream big, experiment, and iterate will all be helpful as we navigate these uncharted waters. But an awareness of these motivations should sharpen our critical engagement with proposals like UBI. We should be aware of the cultural biases that engineers and investors bring with them when tackling a new problem, particularly one with profound social and human dimensions. Most of all, when evaluating these proposed solutions, we must ask what exactly they’re trying to achieve. Are they seeking to ensure that this technology genuinely and truly benefits all people across society? Or are they looking only to avert a worst-case scenario of s...

I fear that many of those in Silicon Valley are firmly in the latter camp. They see UBI as a “magic wand” that can make disappear the myriad economic, social, and psychological downsides of their exploits in the AI age. UBI is the epitome of the “light” approach to problem-solving so popular in the valley: stick to the purely digital sphere and avoid the messy details of taking action in the real world. It tends to envision that all problems can be solved through a tweaking of incentives or a shuffling of money between digital bank accounts. Best of all, it doesn’t place any further burden on researchers to think critically about the societal impacts of the technologies they build; as long as everyone gets that monthly dose of UBI, all is well. The tech elite can go on doing exactly what they planned to do in the first place: building innovative companies and reaping massive financial rewards. Sure, higher taxes required to fund a UBI will cut into those profits to a certain degree, but the vast majority of the financial benefits from AI will still accrue to this elite group. Seen in this manner, UBI isn’t a constructive solution that leverages AI to build a better world. It’s a painkiller, something to numb and sedate the people who have been hurt by the adoption of AI. And that numbing effect goes both ways: not only does it ease the pain for those displaced by technology; it also assuages the conscience of those who do the displacing.

As I’ve said before, some form of guaranteed income may be necessary to put an economic floor under everyone in society. But if we allow this to be the endgame, we miss out on the great opportunity presented to us by this technology. Instead of simply falling back on a painkiller like a UBI, we must proactively seek and find ways of utilizing AI to double-down on that which separates us from machines: love. Admittedly, this won’t be easy. It will require creative and different approaches. Executing on these approaches will take a lot of legwork and “heavy” solutions, reaching beyond the digital sphere and into the not-so-neat details of the real world. But if we commit to doing th...

MARKET SYMBIOSIS: OPTIMIZATION TASKS AND HUMAN TOUCH The private sector is leading the AI revolution, and, in my mind, it must also take the lead in creating the new, more humanistic jobs that power it. Some of these will emerge through the natural functioning of the free market, while others will require conscious efforts by those motivated to make a difference. Many of the jobs created by the free market will grow out of a natural symbiosis between humans and machines. While AI handles the routine optimization tasks, human beings will bring the personal, creative, and compassionate touch. This will involve the redefinition of existing occupations or the creation of entirely new professions in which people team up with machines to deliver services that are both highly efficient and eminently human. In the risk-of-replacement graphs from chapter 6, we expect to see the upper-left quadrant (“Human Veneer”) offer the greatest opportunity for human-AI symbiosis: AI will do the analytical thinking, while humans will wrap that analysis in warmth and compassion. In that same chart, the two quadrants on the...

Instead, patients will desire—and I believe the market will create—a more humanistic approach to medicine. Traditional doctors could instead evolve into a new profession, one that I’ll call a “compassionate caregiver.” These medical professionals would combine the skills of a nurse, medical technician, social worker, and even psychologist. Compassionate caregivers would be trained not just in operating and understanding the diagnostic tools but also in communicating with patients, consoling them in times of trauma, and emotionally supporting them throughout their treatment. Instead of simply informing patients of their objectively optimized chances of survival, they could share encouraging stories, saying “Kai-Fu had the same lymphoma as you and he survived, so I believe you can too.” These compassionate caregivers wouldn’t compete with machines in their ability to memorize facts or optimize treatment regimens. In the long run, that’s a losing battle. Compassionate caregivers would be well trained, but in activities requiring more emotional intelligence, not as mere vessels for the canon of medical knowledge. They would form a perfect complement to the machine, giving patients unparalleled accuracy in their diagnoses as well as the human touch that is so often missing from our hospitals today. In this human-machine symbiosis created by the free market, we would inch our society ahead in a direction of being a little kinder and a little more loving. Best of all, the emergen...

Finally, the internet-enabled sharing economy will contribute significantly to alleviating job losses and redefining work for the AI age. We’ll see more people step out of traditional careers that are being taken over by algorithms, instead using new platforms that apply the “Uber model” to a variety of services. We see this already in Care.com, an online platform for connecting caregivers and customers, and I believe we will see a blossoming of analogous models in education and other fields. Many mass-market goods and services will be captured by data and optimized by algorithms, but some of the more piecemeal or personalized work within the sharing economy will remain the exclusive domain of humans. In the past, this type of work was constrained by the bureaucratic costs of running a vertical company that attracted customers, dispatched workers, and kept everyone on the payroll even when there wasn’t work to be done. The platformatization of these industries dramatically increases their efficiency, increasing total demand and take-home pay for the service workers themselves. Adding AI to the equation—as ride-hailing companies like Didi and Uber have already done—will only further boost efficiency and attract more workers.

But despite all these new possibilities created by profit-seeking businesses, I’m afraid the operations of the free market alone will not be enough to offset the massive job losses and gaping inequality on the horizon. Private companies already create plenty of human-centered service jobs—they just don’t pay well. Economic incentives, public policies, and cultural dispositions have meant that many of the most compassion-filled professions existing today often lack job security or basic dignity. The U.S. Bureau of Labor Statistics has found that home health aides and personal care aides are the two fastest growing professions in the country, with an expected growth of 1.2 million jobs by 2026. But annual income in these professions averages just over $20,000. Other humanistic labors of love—stay-at-home parenting, caring for aging or disabled relatives—aren’t even considered a “job” and receive no formal compensation. These are exactly the kinds of loving and compassionate activities that we should embrace in the AI economy, but the private sector has proven inadequate so far at fostering them. There may come a day when we enjoy such material abundance that economic incentives are no longer needed. But in our present economic and cultural moment, money still talks. Orchestrating a true shift in culture will require not just creating these jobs but turning them into true careers with respectable pay and greater dignity. Encouraging and rewarding these prosocial activities me...