The AI Revolution

This is a review of Kai-Fu Lee's book "AI Superpowers: China, Silicon Valley and the New World Order". 

Lee's predictions are unsettling at first. Being an American citizen I found myself getting defensive about how rosy the picture Lee paints looks for China. But the author patiently and fairly elaborates on why his forecast is what it is. A few chapters in I found myself gradually coming around to seeing the validity of his arguments. 

The China Advantage

Apart from Lee's fondness for the word "juggernaut", the book is well-written and informative. Having worked in China, I have first-hand experience of how they've modernized themselves almost overnight to become the envy of the world. I have had the pleasure of traveling on their fast trains, e.g. the one connecting Shanghai to Hangzhou (a technology hub and home to the Chinese e-Commerce "juggernaut" Alibaba), and they are second to none. After visiting China, it is embarrassing to return to Boston and ride the T, as the Greater Boston transit system is known.

Perhaps not unlike other developing nations, many in China went straight from cash to mobile payments, entirely skipping over credit/debit cards. By contrast, mobile payments are barely used in the US because everyone is already accustomed to credit cards which offer fraud protection. Eventually, the concept of fraud protection and fees is sure to reach the world of mobile payments. Until then, the huge influx of mobile payment data is giving China a heavy advantage on the AI front since AI gorges itself on data just as a Hummer guzzles gas.

Small businesses didn't have POS systems and avoided credit card fees, but they do have smartphones. Mobile payment apps move money directly between accounts without charging fees in most cases. That China has no qualms regarding individual privacy relative to the first-world comes up rather late in Lee's discussion and isn't addressed head-on. The autocratic system that allows China to freely experiment with technologies like facial recognition without as much as a murmur from their one billion-plus population is more or less given a free pass by Lee. 

I hate to be so blunt, but let us recall that China was the only country in the world that allowed the legal sale of ivory until they finally put an end to it in 2017, kicking and screaming some would say. Anything that makes money is generally allowed in China. Such a system cannot prosper in the long term. But before we adopt a holier than thou attitude towards China, we will do well to recall the history of the US and other so-called developed nations living in glass houses.

Big Government

As Lee points out, Obama was relentlessly criticized for one failed investment with Solyndra as part of his administration's attempt to support renewable energy projects. Chinese leaders don't have to worry about such political ramifications and can forge ahead with their plans unencumbered. That's what makes the moral argument against China's autocratic system such a non-starter. Perhaps that is why Lee never fully ventures there. 

A clear distinction between the entrepreneurial environments in the US and China is the incessant reminder by Lee that as far as China is concerned government support is a critically necessary pillar. That is obviously not the case in the US. That dependence implies that even a minor instability in government could derail China's innovation train. Lee credits the government for enabling entrepreneurs but downplays the negative aspects of doing business in an autocratic environment wherein criticizing the government can have grave consequences, as seen recently with the disappearance of Jack Ma.

But it's not all bad. As Lee explains, the Chinese political system involves competition among lower ranks for promotion. And the way to get promoted is to succeed at implementing the goals set by the central government. In some ways, it sounds more promising than the US system where often whoever can spend the most money on campaigning is almost guaranteed a win. Naysayers will point to the folly of having the public sector intervening in the private sector. But in China that intervention seems to have turbo-charged the migration from a manufacturing economy to an internet economy

Lee explains that whereas the Silicon Valley model is to build an ecosystem wherein the startup provides an internet-based layer on top of existing brick and mortar logistics, the Chinese model is more heavyweight, or what I prefer to call full-stack, wherein they reduce outsourcing as much as possible in order to control costs. However, as a result, there's no core-competency and every startup is eventually going to either die or become a monopolistic behemoth. There's no scope for many competent companies to collaborate. 

This is the reason why US companies like Google, Twitter, Yelp, Uber, Facebook, WhatsApp, and others haven't succeeded in getting a foothold in China. And although Alibaba, Tencent, WeChat, and others have failed even more miserably to make a dent in the US, they have spread their tentacles quite successfully in many other parts of the world, especially in Asia and Africa. 

The real point of inflection will be when a Chinese startup makes inroads into the US market. So far, TikTok's success in the US has been a rare example of this. However, there are many more like TikTok waiting in the wings. Chinese researchers are devouring the latest AI research publications and are willing to adapt their apps to each market. In contrast, US app makers prefer to go with a one-size-fits-all approach that doesn't always work in non-US markets.

Killing the Competition

The model in India has similarities with China in that a few companies like Tata, Birla, Mahindra, Reliance, etc. have been allowed to become too big. Airbnb's Chinese rival Tujia has gone beyond the Airbnb model and expanded into the rental properties business which blurs the line between a hotel and a room in a house. The charm of staying at a unique spot is lost as is the entire point of the business model. It's a sledgehammer approach to business that I find not very attractive. But I do thank Lee for laying it out so clearly for us to understand and appreciate. 

Overall, Lee is far too bullish on China's prospects. If you thought work-life balance is poor in the US, it's much worse in China and is reminiscent of Japan during the post-World War II rebuilding years. We all know how the Japan story developed. And by that measure, China should not be counting its chickens just yet. 

Epiphany

Chapter 7, where Lee talks about his cancer diagnosis and the resulting epiphany regarding the meaning of life, felt a bit over the top and hyperbolic. I waited patiently for him to return to his core competency of AI. The blueprint for co-existence with AI that follows from Lee's cancer-induced epiphany is to let AI do the thinking and let humans do the loving, with the goal of developing a society with more empathy. Not an entirely easy pill to swallow. 

Lee oversimplifies the value of human interaction and tries to boil it down to one word, i.e. love. But human interaction is about a lot more than just love. When I talk to friends and family I want them to be opinionated, challenge my viewpoint, teach me something new, make me laugh, gain my respect. I am not going to have such an interaction with a machine, at least not any time soon. Lee seems to be blind to all this.

However, I do agree with the author's vision that the material abundance that would result from harnessing AI/ML should be used to spread love, compassion, and a minimum standard of life for all inhabitants of this wonderful planet we call Earth. 

Kai-Fu Lee splits the current job market into four quadrants and labels them as safe, human veneer, slow creep, and the danger zone. In the "human veneer" quadrant, for example, he suggests placing doctors with the title "compassionate caregivers", since their future job will be to compassionately deliver the diagnosis produced by AI. I don't know about you, but I'm not paying extra for the compassionate delivery if I can save a few bucks and just grab the printout and proceed to my next appointment. 

The subsequent chapters (8 and 9) are just more of the same and no more discussion of AI per se. But chapters 1-6 make the book well worth reading.