Kai-Fu Lee: AI superpowers. China, Silicon Valley, and the new world order
Houghton Mifflin Harcourt, Boston, 2018
Artificial intelligence (AI) is everywhere at the moment. It is in our everyday applications and organization strategies, but also a frequent discussion topic at dinners and conferences. There is a myriad of texts published on the topic – many describing shorter term technical developments and business opportunities that these known technologies could provide us – but also a growing amount of texts describing the further future opportunities and challenges. I am definitely not an expert in this area, but even the non-expert can’t help bumping into these books everywhere. Many of them are very generic. I have for example read Byron Reese: The Fourth Age. Smart Robots, Concious Computers and the Future of the World. I know it has been hyped in all sorts of reviews and I can understand the need of this kind of books, but I wasn’t really blown away. There are also many best-selling books like Yuval Noah Harari’s Homo Deusor Steven Pinker’s Enlightment Now. The Case for Reason, Science, Humanism and Progressthat look at the larger societal implications of AI, at least in a chapter or two. Research in the area is flourishing (see for example fcai.fi) and on-line courses for the broader public are popular. One of these, the Elements of AI(elementsofai.com) has to date been taken by more than 140 000 people and is free, so basic knowledge is easily available. I can recommend taking a look at all of these.
However, there is one recent book on AI development I have personally enjoyed more than many others, and it is this one by Kai-Fu Lee. It’s hands-on and practical, and isn’t as grand in its approach as many of the ones mentioned above. It compares the development of AI in China and Silicon Valley, not only the technology but more importantly the applications created with it. In addition to the concrete description of what has happened in each of the stages of AI development in these countries, Lee also thinks a lot about what political circumstances or people’s mindsets are needed for this development to happen. I suppose I like the book rather than for a brilliant analysis just simply because I recognize so much of the events described in the book. I realize I spent much of the reading time nodding “yes, I remember that”, “yes, that’s how it happened”, “yes, I remember person X also describing that case”, and so on…. In the end of the book Lee does go beyond the current usage, and becomes less sharp than in the beginning of the book.
The utilisation of AI is often shown to happen in four waves; Internet AI, Business AI, Perception AI and Autonomous AI. The two first waves, Internet AI and Business AI, are already all around us, reshaping our digital and financial worlds in ways we can barely register. Internet AI has largely been about using the AI algorithm as a recommendation engine, a system that leans our personal preferences and then serves up content hand-picked for us. Data is labelled, and these labels -our purchases, likes, views etc- are then used to train algorithms to recommend more content for us. This helps Amazon to know what you might want to buy or Google what you might want to search for.
Business AI also utilises data created by traditional companies and organisations, which have been automatically labelling huge quantities of data for decades, and mining these databases for hidden correlations that a human might not notice. This is the case in applications such as medical diagnosis, where an algorithm can scan through an endless number of images of malign cancer, and learn to recognize it better than humans.
The third wave, perception AI, is now digitizing our physical world, recognizing our faces, understanding our requests and “seeing” the world around us. Algorithms can now group pixels from a photo into meaningful clusters and recognize objects, or understand words and meanings out of audio data. Autonomous AI will come last but have a huge impact on our lives with self-driving cars, autonomous drones and intelligent robots in factories.
This book aims to compare China to the US, to show which one is better on the practical usage of AI. Although I don’t typically like putting any two contesters against each other, to say if China is better than the US or vice versa, this book is in reality not so much a beauty contest. It is more of a very practical description of what has been happening in AI development in China lately, with the backdrop of similar development in Silicon Valley (which the author clearly thinks is the comparison point that everyone is aware of). Maybe with the exception of Business AI, where the US-based companies have traditionally been strong, much of the current developments are very much in favour of China. This relates to big shifts in current society.
According to Lee, we are currently undergoing a global shift that is the product of two transitions: we are moving from the age of discovery to the age of implementation, and from the age of expertise to the age of data.
The age of implementation is a way to say that with current AI development it might no longer be about who does the first and most ground-breaking inventions, but about who will implement it in the most usable and best scalable ways. Understanding this distinction between discovery and implementation is, according to Lee, core to understanding how AI will shape our lives and which country will drive this development.
China has a heritage of appreciating copying. In China, entry into the country’s imperial bureaucracy depended on word-for-word memorization of ancient texts and the ability to construct a perfect essay following rigid stylistic guidelines. Rigorous copying of perfection was seen as the route to true mastery. In addition to a cultural acceptance to copying the Chinese internet ecosystem has two more psychological foundations to its benefit; they have a scarcity mentality (not generations of wealth, but most people still have parents who have very little) and a willingness to drive any promising new industry. This has led to a very harsh entrepreneurial environment. For any idea, there has been an onslaught of thousands of mimicking companies competing on the domestic market. This has forced companies to innovate. Lee compares the Chinese market to a Colosseum where they have had to fight like gladiators for their survival; only the strongest can win, the fight is lethal, and the rest die. In the age of AI this cutthroat entrepreneurial environment will be one of China’s core assets.
The explosion of real-world internet services across China was dubbed the O2O Revolution, short for “online to offline”. The concept is simple: turn online actions info offline services. A brilliant example of this is the rise of China’s super-app, WeChat. In five years Tencent built WeChat to become the “remote control for life”. It was a result of tying together several online-to-offline services. With it you could pay at a restaurant or hairdresser, unlock a bike, manage your investments, book a doctor’s appointment and have your medicines delivered to your home. All blurring the lines between the online and the offline. The “Pearl Harbour” of mobile services and paying came with the Chinese New Year in 2014. In Chinese tradition, you give “red envelopes” to your friends and family during this occasion, small and decorative packages with cash in them. TenCent’s innovation was simple and fun, as WeChat made it possible to send out digital red envelopes with real money. The money in WeChat Wallet could be used to make purchases, transferred to other friends, or moved to the user’s bank account as long as they had linked their account with WeChat.
Currently much of the latest research is available instantly and online, and in transferring this knowledge two traits are crucial: openness and speed. There mere volume of people being able to do this also counts. As AI filters into the broader economy the quantity of AI engineers is more important than the quality of the elite researchers, says Lee. In the age of AI implementation an army of AI engineers is needed to team up with entrepreneurs to turn discoveries into game changing companies. China is currently training such an army.
So Chinese entrepreneurs live in a world were speed is essential, copying is an accepted practice, and competitors will stop at nothing to win a new market. The Chinese also work hard. Or as Lee puts it: “I can tell you that Silicon Valley looks downright sluggish compared to its competition across the Pacific. […] Its a cultural environment that inspires a truly maniac work ethic. Compared to China, Silicon Valley is lazy.” And compared to the Silicon Valley Europeans work a lot less…
They can also turn their ideas into products very quickly. Much of this has been made possible through the governments strong push to build an environment where this is possible, coined under the epithomy of “Made in Shenzhen”. The area gives an unparalleled flexibility of the supply chains on armies of skilled industrial engineers. In Shenzhen entrepreneurs have direct access to thousands of factories and hundreds of thousands of engineers who help them iterate faster and produce goods cheaper than anywhere else. Hardware entrepreneurs say that a week spent nothing in Shenzhen is equivalent to a month in the US.
The other transition that Lee refers to is the transition from the age of expertise to the age of data. In deep learning, there is no data like more data. The more examples the return is exposed to, the more accurately it can pick out patterns and identity things in the real world. This is turning China into the Saudi Arabia of data, a country that suddenly finds itself sitting atop stockpiles of the key resource that powers this technological era. Much of this data is also a result of the tight collaboration between the Chinese Government and its companies.
In addition to the abundance of data, there is also a cultural benefit to usage of this data that is different to many other countries. Lee says that the Chinese are less worried about their faces being recognised in apps, and appear to be more willing to give away personal data for the benefits that these apps bring into their everyday life.
According to Lee the Chinese users’ cultural nonchalance about data privacy and Shenzhen’s strength in hardware manufacturing give it a clear edge in implementing AI.
Today there are “seven giants” in corporate AI research: Google, Facebook, Amazon, Microsoft, Baidu, Alibaba, and TenCent. Four of these are from the US and three from China. It is Google (or its parent company Alphabet) that has the biggest budget. Googles budget for R&D is as big as the other six together, and the US government’s spending on math and computer science research is only about half of Google’s. But the governments also have a much bigger role than just their own research spending. The governments can form the ecosystems and mindsets in which this innovation occurs. China has excelled on this lately.
Lee concludes that China has four advantages when it comes to successful AI development and usage:
- the age of implementation
- the age of data
- world class entrepreneurs
- a proactive government
Both the US and China have AI strategies: the US released their long-term AI plan in 2016. Although it offered a decent summary of changes in the notion, prescriptions of adaption, and was issued by Obama’s White House, it had according to Lee little effect.
In 2017 the Chinese State Council published the “Development plan for a new generation of Artificial Intelligence”, similar to the White House plan. It also spelled out hundreds of industry-specific applications of AI. It included a roadmap of how China would become an AI superpower. It called for China to reach the top tier of AI economies by 2020, achieve major new breakthroughs by 2025, and become the global leader in AI by 2030.
Similar government-accelerated technological development has happened twice in China in the past decade. Between 2007 and 2017, China went from having zero high-speed rail lines to having more miles of high-speed rails operational than the rest of the world combined. During the “mass innovation and mass entrepreneurship” campaign that started in 2015, 6600 new startup incubators were created that shifted the national culture around technology startups.
Lee exemplifies this “mass innovation and mass entrepreneurship” era with the case of Guo Hong, a new kind of governmental official for a new era- focusing on China needing to both build things and create ideas. The country’s most important economic actor isn’t a single company but the Chinese Government. Guo Hong’s new venture, the Avenue of the Entrepreneurs, and the phrase “mass entrepreneurship and mass innovation”, became the slogan for a momentous government push to foster startup ecosystems and support technological innovation. The guidelines were set by the government but the execution was left to local officials. Start-up clusters flourished all around the country. The Chinese government wanted to engineer a fundamental shift in the Chinese economy, from manufacturing-Ied growth to innovation-Ied growth, and it wanted to do so in a hurry.
Now China’s strategy on becoming the world’s leading AI superpower, is creating similar effects.
Perception AI is blurring the lines between the online and offline worlds. You can now order a meal by talking on your couch and your refrigerator tells your shopping cart you need milk. These new blended environments are called OMO (online- merge-off line). Perception AI will turn shopping malls, grocery stores, city streets and our home to OMO environments. I remember talks and concepts of these services from at least a past couple of decades, but now they are happening in real life in China. As an example, Xiaomi, originally launched as a low-cost smartphone, is now building a network of Al-empowered home devices that turn our kitchens and living rooms into OMO environments.
OMO can also be used in education. AI powered education experiences include in-class teaching, homework and drills, tests and grading, and customized tutoring. All these create the student profile, and education becomes precisely tailored to the student’s abilities. A great example of this (which isn’t mentioned in Lee’s book) is SnapAsk, where part of the business innovation is to outsource high-school tutoring to a huge network of university students online. All the data is tracked and analysed, and we can only guess how much this profiling will be used in the future, and by whom.
Autonomous AI combines the three previous waves ability to optimize from complex data and sensory powers. The first autonomous AI will according to Lee surface in commercial settings, like a robot picking strawberries or beetles collecting merchandise in a warehouse.
As an example of this wave Lee mentions the Shenzhen-based DJI, the world’s premier drone maker and by tech journalist Chris Anderson called “the best company I have ever encountered”. DJI spends enormous resources on research and development, and is already deploying autonomous drones for industrial and personal use. Swarm technologies are still in their infancy, but when hooked into Shenzhen’s unmatched hardware ecosystem the results will be awe-inspiring, says Lee.
If we look into the future of what this development could bring us, Lee doesn’t really believe in a big change coming through Artificial General Intelligence (AGI) or any other scenario that is based further in time. He says that the real challenge is loss of work, and that this change is very close in time. The lead in AI productivity deployment will lead to productivity gains we haven’t seen since the Industrial Revolution. In fifteen years, AI will be able to replace 40-50% of jobs in the US. This will lead to a far more personal and human crisis, a psychological loss of one’s purpose.
According to Lee, 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 optimisation algorithm. Instead, we must move toward a new culture that values human love, service and compassion more than ever before. Or, as Lee puts it, to “use AI to build a more humanistic world”.
Building societies that thrive in the age of AI will require substantial changes to our economy but also a shift in culture and values. Meaning becomes increasingly important and inequality an overwhelming threat.
As part of these changes Lee presents the three R’s: Retraining workers (this is where life-long learning comes into the picture), reducing work hours, or redistributing income. According to Lee, all these approaches have their advocates. In addition, new service jobs are emerging we cannot even imagine today. Although Lee doesn’t really talk a lot about it, I think this is also a big opportunity. I understand that new technology will make many current jobs obsolete and it is important to think about how we can best support this shift. At the same time, this new(ish) quest for meaning and the opportunities that new technologies give us, provide possibilities for many new kinds of jobs. We need to think about how we want the world to unfold, what kind of services or approaches people want or need, and how do we create the kind of society we want. I struggle in agreeing with the pessimists of this being only a negative development – I think the scenario of people thinking of more and creative now jobs is just as possible. If we can move some of the more monotonous work to machines and focus on doing jobs we love and care about it can’t be all bad….
As Lee states in his introduction, the skilful application of AI will be China’s greatest opportunity to surpass the US. But more important, this shift will create an opportunity for all people to redeliver what it is that makes us human. I’ll look more into this question of what makes us human in my next book report, promise.
Lee continues that this will require that we reimagine and reorganize our society from the ground up. It will take societal unity, creative policies and human empathy, but if achieved it could turn a moment of outright crisis into an unparalleled opportunity.
How do we make the best out of these opportunities? In a human way? And what should we do in Europe, which somehow isn’t even existing in this discussion?