I spent some time over the past week to read what’s now one of my favorite books. Written by Kai-Fu Lee, who was the former head of Google China and is currently CEO of Sinovation Ventures, AI Superpowers: China, Silicon Valley, and the New World Order offers a glimpse into the various scenarios that could unfold as AI technology develops. Lee’s unique perspective from being a top AI executive in both the U.S. and China allows him to straddle two worlds that will ultimately decide the fate of how our world adapts to the AI revolution. His key message is a powerful one, essentially summarized by his ending words:
“AI’s greatest potential to disrupt and destroy lies not in international military contests but in what it will do to our labor markets and social systems. Appreciating the momentous social and economic turbulence that is on our horizon should humble us. It should also turn our competitive instincts into a search for cooperative solutions to the common challenges we all face as human beings”
Lee’s perspective on the growth of the Chinese tech ecosystem
Lee starts off by examining the Chinese tech ecosystem from its early stages in the 2000s and early 2010s. Originally, Chinese tech companies were known to be “copycats” of their American counterparts. After all, top American startup ideas were the gold standard. To the outsider looking in, this ecosystem seemed immature and lacked innovation. However, Lee paints a different picture of this breed of entrepreneurs who successfully fended off extreme competition to emerge victorious. When one operates in an environment where it’s not uncommon for rival startups to make false claims to the police in hopes of arresting competition, copying is widely practiced, and pricing would be fought to the last penny, successfully emerging from this environment required the strategic acumen, innovation and grit that is rarely found anywhere else in the world. It’s truly a modern age coliseum, with the gladiators being entrepreneurs.
These entrepreneurs would go on to masterfully integrate consumers into their products. Unlike American consumers who treat the internet like White Pages (i.e. the main goal is to look up specific information), research showed Chinese consumers treated the entire internet like a search engine where they would browse multiple options no matter the domain before narrowing down their focus. In addition, the heavy adoption of mobile payments allowed Baidu, Alibaba, Tencent and other Chinese tech giants to collect a massive amount of data, which China is just now seeing the true benefits from. Tencent’s WeChat in particular, became a Swiss army knife of apps that was arguably the biggest component of accelerating this data generation.
China’s Sputnik moment for AI
The turning point for the Chinese ecosystem came in the form of China’s “Sputnik” moment as Lee calls it, when AlphaGo defeated the top ranked human Go champion. To those who don’t known what Go is, it’s a Chinese board game believed to have been developed over 2,500 years ago and holds significant cultural importance as an ancient art in China. What’s relevant from an AI perspective is there’s more possible moves in Go than the total number of atoms in the universe. Therefore, it was widely expected that AI needed more time to successfully develop before beating a top human player.
After AlphaGo not only defeated, but dismantled Lee Sedol, the Chinese government went into an AI frenzy. To someone outside of Chinese culture who didn’t realize the cultural significance of the game, it would have been easy to overlook AlphaGo’s victory in relation to China’s interest in AI. Few realize that over 280 million Chinese viewers tuned into the AlphaGo matches against Lee Sedol.
AI In 2030 initiative
Lee argues that China could have taken the American approach to new markets (which he said worked amazingly well for America) and let private capital naturally flow into AI startups and research. But, China needed to play catch up. At least 10 years behind the U.S., Canada and other developed nations in AI advancement, China decided to turn on the fire hose of investment and force progress. The AI in 2030 plan calls for China to be the premier world leader in AI and sets ambitious intermediary milestones to achieve this.
Local governments raced like they were let off the starting blocks to throw their weight behind the central government’s initiative. Unlike American culture, Chinese culture leans heavily towards conformity and deference towards authority figures. This flurry of investment, support, and focus further highlighted one of the main advantages of China’s technology ecosystem, it’s data. Lee believes that we’re currently in the implementation phase of AI after the emergence of deep learning, meaning that a lot of the major advancements needed to commercialize the technology and apply them in a business environment have already been made. Therefore, while the AI talent remains highest in the U.S., there are diminishing returns to the quality of your engineers and AI researchers after hitting a certain threshold. The big differentiating factor at this point is the amount of quality data you can use to train algorithms.
Lee does acknowledge that if one of the seven tech giants (Google, Facebook, Apple, Microsoft, Baidu, Tencent, Alibaba) finds a new AI technology breakthrough, then the balance could go back to the discovery phase where innovation of new techniques trumps the implementation of existing technologies. He also notes that whoever discovers this new breakthrough stands to gain monopoly power over any other tech company in the world. For what it’s worth, he believes his former employer (Google) has the best shot.
The Kai-Fu Lee scorecard: The U.S. vs. China in AI
Lee believes in 4 waves of AI that he also uses to compare where the U.S. and China stack up against each other.
(1) Internet AI – Leverages the fact the internet users are automatically labeling data as they browse
(2) Business AI – Leverages data from businesses to find hidden correlations that can train algorithms
(3) Perception AI – Digitizing the world around us (images, audio, and video) through smart sensors and devices
(4) Autonomous AI – Leveraging the developments from the three previous waves to create a world in which autonomous agents are fully integrated into every aspect of our lives
His scorecard below highlights why his book is called AI Superpowers (emphasis on the plural nature of the title). Both the U.S. and China have their respective strong points.
The path forward
Lee faces a true crisis late in his career with a cancer diagnosis in 2013 that dramatically changes his outlook on life and AI (he has since recovered to beat his cancer). As a self-described “iron-man of achievement and work” before his diagnosis, Lee realizes how important the concepts of love and social connectivity are to the human experience. In this entire AI revolution, he argues that the distinguishing factor that gets lost in all of the hype is that humans will write their own AI story.
How we choose to utilize AI will make or break our future. The potential for super profits for a select few, and the resulting dramatic income inequality as AI develops is all too real of a threat. In addition to the monetary results of AI, humans for so long have defined their place in the world by their “work”. The loss of this identity for many people will have profound mental effects that cannot be underestimated.
Kai-Fu Lee (and many others in the world) are already thinking of solutions to these issues. Lee’s overall thought is the fundamental way we think of Universal Basic Income (UBI) as a solution ignores the root cause of the real problems that people will face. In essence, UBI is an easy solution the winners of AI can implement to avoid responsibility for the inequitable world they will create. Instead, he proposes that we create a social investment stipend that can be implemented globally.
This stipend will be a decent/above average government salary given to those who invest their time and energy in those activities that promote a kind, compassionate, and creative society. These activities would be included in three broad categories:
- Care work
- Community service
- Education
This approach would serve as a replacement for how we reward economically productive activities, and the funding for this stipend can be in the form of super-taxes on the wealthiest benefactors of AI driven profits. Let’s be real, if you’re making $100bn a year in pure profit due to your monopoly on AI, what’s the big deal if that’s further cut down to $75bn? The world as a whole should be benefiting from the progress of AI, so the need for high incomes to offset education, healthcare, social services, and other needs should theoretically be decreased.
The purpose of the stipend would not be to replace the traditional welfare system, but rather offer respectable income to talented individuals who are displaced by AI in the three main areas above. Each area will involve various levels of rank/compensation (to incentive top performers), in addition to part-time and full-time opportunities. Lee believes that there is enough diversity in potential jobs within each of the three categories, that every person can find a niche to work in. The larger message this type of program sends is that we’re all in this journey together, and we’ve only progressed this far as a society by working together to build each other up.
For those who have made it to the end of this post, I thank you. Hopefully you leave encouraged to help contribute to making the world a better place as AI slowly takes over our lives. I didn’t capture everything in Lee’s book, so make sure to give it a read if you find this post interesting.