The Great AI Duopoly


Nathan Gardels is the editor-in-chief of Noema Magazine.

Kai-Fu Lee is the chairman of Sinovation Ventures and the president of its Artificial Intelligence Institute. He was the founding president of Google China. He recently spoke about his new book, “AI Superpowers: China, Silicon Valley and the New World Order,” with The WorldPost’s editor in chief, Nathan Gardels.

WorldPost: Artificial intelligence is surely the most consequential technological development of the 21st century. In your book “AI Superpowers,” you’ve written the most comprehensive global account of artificial intelligence to date. What are the central themes of your book?

Kai Fu Lee: The benefits, as well as the potential disruptive effects of AI, are well known. To give only a few obvious examples, AI has already enabled us to amplify our knowledge through search engines, connect people who speak different languages through machine translation and save us billions of dollars from credit card theft and fraud.

In the future, AI holds the promise of significantly improving medical diagnosis and lowering the cost of health care. AI can personalize education and improve the speed of learning. It will enable automated stores and factories, thus significantly reducing the cost of goods and services. Finally, intelligent machines will take care of our chores, such as driving, cooking, cleaning, dishwashing and laundry, giving us back our most valuable resource — our time. PricewaterhouseCoopers estimates AI will add $15.7 trillion to the world economy by 2030.

Taking care of our chores and automating manufacturing, of course, will mean a total disruption of patterns of work and employment. Some estimate as many as 40 percent of current jobs will be lost to intelligent machines.

Most reports and books on the subject have been written either from a technologist/researcher perspective or from a business, history or economics perspective. I come at it through all of these lenses and as someone who has extensive experience in both China and the United States — the great AI duopoly. My aim is to acknowledge and understand the critical lessons of China’s rise as a tech power, not based on misconceptions in the West about intellectual property theft or government protection. Of course, Chinese entrepreneurs had their scrappy, questionable methods in the early days. But its tech industries have evolved into a completely legitimate method of starting world class Internet and AI businesses.

The Chinese have developed a valuable methodology of tech business innovation that involves an ultra-fast iterating product design based on instantaneous feedback from massive market data.

In the United States, tech companies don’t compete very hard, so when you build out a franchise such as Instagram, competitors say: “O.K., I’m going to do something else. I don’t want to do what you do.” In China’s super-competitive environment, when people see an idea, they jump in with a copycat mentality — it is not about copying Silicon Valley but about copying any features developed in any country, if it is useful.

So to prevent the copycat who has more money or better engineers from taking over your franchise, you need to erect extremely high walls to protect yourself. This leads to a completely different business model from Silicon Valley. It forces start-ups to take extraordinary risks to build extremely complex “moats” to maintain a competitive advantage that will protect profits and market share over the long term. It is this combination of a competitive, fight-or-die mentality married to abundant capital and a gigantic networked consumer base that is fueling China’s ascent in commercial technology. This is worth studying by American businesses.

The AI revolution will have two engines — China and the United States — pushing its progress swiftly forward. It is unlike any previous technological revolution that emerged from a singular cultural setting. Having two engines will further accelerate the pace of technology.

WorldPost: In your book, you talk about the “data gap” between these two engines. What do you mean by that?

Lee: Data is the raw material on which AI runs. It is like the role of oil in powering an industrial economy. As an AI algorithm is fed more examples of the phenomenon you want the algorithm to understand, it gains greater and greater accuracy. The more faces you show a facial recognition algorithm, the fewer mistakes it will make in recognizing your face. The more medical records you show to a diagnostic algorithm, the more accurate its diagnoses will be.

All data is not the same, however. China and the United States have different strengths when it comes to data. The gap emerges when you consider the breadth, quality and depth of the data. Breadth means the number of users, the population whose actions are captured in data. Quality means how well-structured and well-labeled the data is. Depth means how many different data points are generated about the activities of each user.

Chinese and American companies are on relatively even footing when it comes to breadth. Though American Internet companies have a smaller domestic user base than China, which has over a billion users on 4G devices, the best American companies can also draw in users from around the globe, bringing their total user base to over a billion.

Americans, so far, enjoy a distinct advantage when it comes to quality. Companies and public institutions in the United States are much more likely to use software that structures their data for immediate use. Chinese corporations and public entities are moving in this direction, in part due to increased bureaucratic incentives for utilizing data. Still, they lag substantially behind U.S. organizations in terms of accumulation of AI-ready data.

But when it comes to depth of data, China has the upper hand. Chinese Internet users channel a much larger portion of their daily activities, transactions and interactions through their smartphones. They use their smartphones for managing their daily lives, from buying groceries at the market to paying their utility bills, booking train or bus tickets and to take out loans, among other things.

Weaving together data from mobile payments, public services, financial management and shared mobility gives Chinese companies a deep and more multi-dimensional picture of their users. That allows their AI algorithms to precisely tailor product offerings to each individual. In the current age of AI implementation, this will likely lead to a substantial acceleration and deepening of AI’s impact across China’s economy. That is where the “data gap” appears.

WorldPost: This data gap, combined with China’s distinct business model that you’ve described, increasingly puts tech industries on entirely different footings. You say a “parallel universe” already exists there. What are the implications?

Lee: The radically different business model in China, married to Chinese user habits, creates indigenous branding and monetization strategies as well as an entirely alternative infrastructure for apps and content. It is therefore very difficult, if not impossible, for any American company to try to enter China’s market or vice versa. It’s like two different jigsaw puzzles. You can’t take a piece from one and try to fit it into the other — everything is different.

At the same time, companies in both countries are pursuing their own form of international expansion. The United States uses a “full platform” approach — all Google, all Facebook. Essentially Australia, North America and Europe completely accept the American methodology. That technical empire is likely to continue.

The Chinese have realized that the U.S. empire is too difficult to penetrate, so they are looking elsewhere. They are trying, and generally succeeding, in Southeast Asia, the Middle East and Africa. Those regions and countries have not been a focus of U.S. tech, so their products are not built with the cultures of those countries in mind. And since their demographics are closer to China’s — lower income and lots of people, including youth — the Chinese products are a better fit.

There are also local entrepreneurs in countries such as India, Indonesia and Brazil building their own products, who are open to partnerships with Chinese companies. Chinese companies share in the upside of the investment but do not own the company outright, which is the American approach. In these less-developed countries, Chinese companies are making significant investments, injecting not only money but technical know-how, including AI. For example, the Chinese rideshare company Didi Chuxing acquired Uber’s competitors in Brazil and Indonesia.

If you were to draw a map a decade from now, you would see China’s tech zone — built not on ownership but partnerships — stretching across Southeast Asia, Indonesia, Africa and to some extent South America. The U.S. zone would entail North America, Australia and Europe. Over time, the “parallel universes” already extant in the United States and China will grow to cover the whole world.

WorldPost:  Europe seems pretty much out of the picture as any kind of independent player. Is it more or less a tech colony of the Americans?

Lee: Yes, it is. The United Kingdom and France have large aspirations, and Russia is a wild card. As you say, Europe is pretty much an American colony. There is, however, a sensibility in Europe distinct from America, which we see in privacy-oriented laws such as the General Data Protection Regulation. So some differentiation is appearing.

Policy-wise, we are seeing three approaches. The Chinese have unleashed entrepreneurs with a utilitarian passion to commercialize technology. The Americans are similarly pro-entrepreneur, but the government takes a laissez-faire attitude and the entrepreneurs carry out more moonshots. And Europe is more consumer-oriented, trying to give ownership and control of data back to the individual.

WorldPost: Given the parallel universes you describe, aren’t you concerned that the AI superpowers will end up in an arms race? After all, China and the United States are in the midst of a mounting trade war. The hawks in the Trump administration have targeted the “Made in China 2025” industrial policy in order to prevent China’s dominance in AI and robotics.

Lee: An AI arms race would be a grave mistake. The AI boom is more akin to the spread of electricity in the early Industrial Revolution than nuclear weapons during the Cold War. Those who take the arms-race view are more interested in political posturing than the flourishing of humanity. The value of AI as an omni-use technology rests in its creative, not destructive, potential.

While there is competition in commercial applications, those of us engaged in AI tend to be open about sharing basic research since the goal is advancing the field. Unlike many other sciences, AI experiments can be replicated and shared so each builds on the other’s knowledge. In a way, having parallel universes should diminish conflict. They can coexist while each can learn from the other. It is not a zero-sum game of winners and losers.

WorldPost: At the outset, you raised the biggest challenge presented by AI: the loss of jobs. How do you see that unfolding?

Lee: The prevalent and imminent challenge over the next 15 years is dealing with job displacement. We will see a massive migration from one kind of employment to another, not unlike during the transition from agriculture to manufacturing. It will largely be the lower-wage jobs in routine work that will be eliminated, while the ultra-rich will stand to make a lot of money from AI. Social inequality will thus widen.

The jobs that AI cannot do are those of creators, or what I call “empathetic jobs” in services, which will be the largest category that can absorb those displaced from routine jobs. Many jobs will become available in this sector, from teaching to elderly care and nursing. A great effort must be made not only to increase the number of those jobs and create a career path for them but to increase their social status, which also means increasing the pay of these jobs.

There are also issues related to poorer countries who have relied on either following the old China model of low-wage manufacturing jobs or of India’s call centers. AI will replace those jobs that were created by outsourcing from the West. They will be the first to go in the next 10 years. So, underdeveloped countries will also have to look to jobs for creators and in services.

WorldPost: Presumably, the wealth being created by the robots will have to be shared with those working in services of the future?

Lee: There should be some subsidy, yes. But I am opposed to the idea of universal basic income because it provides money both to those who don’t need it as well as those who do. And it doesn’t stimulate people’s desire to work. It puts them into a kind of “useless class” category with the terrible consequence of a resentful class without dignity or status.

To reinvigorate people’s desire to work with dignity, some subsidy can help offset the costs of critical needs that only humans can provide. That would be a much better use of the distribution of income than giving it to every person whether they need it or not. A far better idea would be for workers of the future to have an equity share in owning the robots — universal basic capital instead of universal basic income.

This was produced by The WorldPost, a partnership of the Berggruen Institute and The Washington Post.