Kai-Fu Lee was born in Taiwan but grew up in Tennessee, which is nothing — nothing — like Taiwan or China. His PhD is from Carnegie-Mellon and for the first half of his career Lee was “that voice recognition guy” first at Apple, then Microsoft, then Google. Lee took Google to China the first time (a new Google China effort is starting just now). Today Lee is an Artificial Intelligence expert who runs a $1 billion venture fund with offices in Taipei and Beijing and, according to Anina (the pretty girl in the picture with me who has lived in Beijing for most of the last decade) he’s “an absolute technology rock star — everyone in China knows Kai-Fu Lee.”

Lee is also a prolific author and in his latest book, coming out in September, he wants to explain to the world how Artificial Intelligence will be dominated by Silicon Valley (mainly Google) and China, with China having somewhat of an edge, how half of all jobs in the world are going to disappear because of AI, but how that only sounds like a bad thing.

Well maybe it is a bad thing but there’s no way around it and things could all turn out better in the end.


One important point here is an idea I don’t really state very well below, that AI is going to take away a lot of jobs and throw society into a crisis of how to deal with that. Kai-Fu says the USA is approaching this crisis through what he calls the “Three R’s” — Reduce (number of workers), Retrain (workers to prepare them for new careers) and Redistribute (income, through giving everyone a Universal Basic Income – money for nothing). The Three R’s won’t cut it according to Kai-Fu Lee who, sadly, doesn’t give a fully-explained alternative strategy, though he is optimistic.

AI Super-Powers: China, Silicon Valley, and the New World Order isn’t out yet, though you can pre-order it on Amazon. I have read the book and followed up with a pretty extensive e-mail exchange with the author, who is allowing me to publish that exchange below. Reading this post is in no way comparable to reading the book, which has way more information than we discuss here. I found the book interesting and worth the read. I think it is an important first strike at a positive way to see the jarring changes that are coming with AI. The book isn’t perfect and I’m not sure it’s even entirely correct. But it certainly made me think.

Here is our exchange:

BOB: I’m sorry this has taken so long. I’ve just been so busy which is, of course, the best way to handle the impending challenge of Artificial Intelligence. If you are busy enough, AI doesn’t matter.

I want to give you a broad spectrum of my thoughts along with a few questions. I found your book very interesting but couldn’t decide whether it was too alarmist or not alarmist enough. Clearly AI will change everything, but then doesn’t something come along every couple of generations to change everything? How are the industrial revolution, the rise and fall of slavery, the era of the railroads, the car, WW1, the great depression, WW2, the Marshall Plan, the Internet Highway System, color TV, the Internet, Vietnam, and cellphones each in their own way any different from Artificial Intelligence? They aren’t. There are always going to be shocking changes and we’ll always — somehow — adapt.  This is not to say that we’ll always thrive. I don’t feel we are thriving now, do you?

KAI-FU: I think AI is different from the earlier revolutions you mentioned because it is a black box for creating value, improving efficiency, and doing better than what we do in a broad sense.  Also it is faster because it doesn’t require building up an electrical grid, or Internet connectivity, or cell towers — it just runs on the cloud.  It’s as though Internet created the infrastructure and data for it, cloud computing and GPUs created the backend for it, and it’s much more plug-and-play than any previous revolution.

There are those who see even grander aspirations, but I’m just staying within the already proven technologies plus their natural extensions, the inevitable platformization, and the domain-specific tunings.

BOB: Yet in my experience Artificial Intelligence has taken as much — if not more — time to come to market. Ignoring the AI frenzy of the 1980s, which I lived through, the current AI systems can only grow as fast as the cloud and the cloud seems to me to have grown darned slowly. I just finished six years of helping a cloud startup that was just sold and its success — while dramatic — took WAY longer than any of us expected. And isn’t it always so? Yes, there will come a time when AI advances may well create themselves, but how far away is that, another 10, 15, 20 years? Ray Kurzweil and I are in total disagreement about this and have been for years. So far I am right and Ray is wrong. 

KAI-FU: I am definitely not in Ray’s camp.  I have known him for 30 years.  He is in the romantic and optimistic camp.  I am in the pragmatic engineering camp.

Cloud scaling is one way — they’re all adding AI functions.  But home-made AI is happening too. Did you see that farm boy who used TensorFlow to build a device to sort cucumbers?

With the cost of computing coming down, the ease of the kit getting better, and platform providers not charging for the little guys, there will be a gold rush to using AI. 

BOB: One problem I had with the first half of your book was how proud you are of those brave Chinese entrepreneurs who succeed by breaking rules including patents and copyrights. They emerged in a controlled and isolated economy where they were allowed to copy at will and this is something of which to be proud? I don’t get it.

KAI-FU: I am not proud of the rule breakers.  For that part, I played the role of a historian — telling it like it is.  They broke rules, copied innovations, but then they grew up, and became business innovators themselves.

It is a wake-up call for Silicon Valley, which has rested in its laurels.  History is behind us.  Today, the Chinese companies have demonstrated a totally different way of creating companies — entrepreneurs would be well-advised to study them — the current practices/competition/approach are definitely worthy of many Harvard Business cases.

BOB: Alas, Harvard frequently gets it wrong. I wrote a book in 2014, The Decline & Fall of IBM, and found HBS falling over itself to praise IBM at exactly the points where the company was being most stupid. Watson? Let’s call a failure a failure. 

And I find it difficult to see China as being any more of a development crucible than is Sand Hill Road, where startups have even less time to succeed and therefore more pressure to evolve. Explain to me how this is incorrect, because the numbers are published and are real. In comparison to other adolescent startup cultures, yes China is and will be successful and they are (finally) getting it, but Silicon Valley still has more of everything including investor impatience. So you are right to see the ultimate show-down as being China versus Silicon Valley, but China still has a long way to come.

KAI-FU: I will tell you that five years ago, if you put a Chinese and a Silicon Valley CEO (in a similar area) on a panel, and the Chinese CEO stutters and has a big accent.  Now, most of the time, the Chinese CEO eats the SV CEO for lunch (due to the power of the argument, the magnitude of the success).

It’s kind of like you put Silicon Valley CEO with a London CEO.  While the latter is perfectly smart, he/she lacks the scale and ambition to build something great.

The first skirmish will be Uber vs. Didi.  We’ll see how that works out.

BOB: I spent 20 years running a consulting business in Japan and the USA. My partner was in Yokohama and I was in Palo Alto and our primary function was to help giant Japanese companies avoid embarrassment. Here’s what would happen. A revered engineer or manager would reach retirement age and typically he (always he) would be sent off with the gift of a special project for the company — something to keep him busy, keep him paid, yet often not of strategic value to the company. The example I like to use is time travel. “Sato-san and his team will be conquering time travel during the next five years.” Lunches would be consumed, trips to Italian, Franch and German labs would happen, but no real work would be done. Usually the project would be forgotten. Occasionally the failure would be noticed and an underling sacrificed. And very rarely my partner would be called-in just in case I could find in Silicon Valley some garage in which time travel had already been conquered. I was usually able to come through and occasionally the big company even came out of the experience looking brilliant.

In this era I mainly worked with Japanese companies but occasionally with Chinese and Korean. I met a Korean company called Daou Tech, which had an interesting function — keeping Korean companies honest. Daou was — and is — like Switzerland. Their value added was honesty. They were not the last honest Korean company, they were in my experience the only honest Korean company. Even Samsung couldn’t keep itself from stealing IP whenever possible — unless  that IP was somehow run through Daou Tech, which had some iron grip I could never understand. So often in these kind of affairs, though, that grip probably came down to being a necessity: without Daou Tech, Samsung and the others would eventually have failed. It wasn’t in their nature to be honest; on some level they were too proud to be honest; but without honesty the larger system fails, so what to do? Daou Tech.

Same with China, I am sure. So you can be proud of their dynamism, but if those entrepreneurs are going to steal from me they will suffer for it, which doesn’t help anyone. So I don’t see it as something of which to be proud.

KAI-FU: Again, the copycat behavior was a thing of the past.

BOB: No it isn’t. The fact that something is illegal doesn’t mean it doesn’t happen.

I remember back in the era of the Clipper Chip, when the NSA wanted a backdoor into all new computers, I interviewed the director of the FBI. “Won’t this monitoring capability lead to agents reading the e-mail of their lovers?” I asked.

“Nope,” said the head of the FBI, “that would be illegal.”

In fact it turned out that the FBI, CIA, NSA and other such agencies have all been guilty of breaking the law exactly as I described. So are Chinese companies. And so are some American companies, but I’m pretty sure it’s a much bigger problem in China. 

KAI-FU: Well Youtube started with all illegal videos, and Google Video insisted on legality.  But Youtube got so much share, Google had to buy them.

So as you said, it happens in both countries.  And as you said it is worse in China.  But I’m saying it is improving dramatically.  So I am awaiting this to sort itself out in a few years.

The China environment today clearly does not have the level of integrity found in the US, but it has gone a long way.  Eight years ago I struggled at Google as everyone pirated music, and Baidu produced a MP3 search for pirated music.  But today, every Chinese giant licenses music properly.

BOB: Not according to a friend of mine who is the financial manager for many iconic rock bands. One of his jobs is auditing royalties and there’s about a 40 percent revenue differential per track played in China compared to the U.S. — a differential that can be traced straight to piracy. It’s not a matter of lower pricing — 40 percent of Chinese plays generate no revenue at all.  

KAI-FU: The government hasn’t cracked down on all the pirated music sites thoroughly, just the big ones.  But if you look at the market share now, all the market leaders and market gainers pay.  So the legalization is inevitable.

BOB: I think you miss a few points that could have been made about population transitions in an AI-type crisis. Des Morris, who wrote The Naked Ape back in the 1960s wrote a follow-up book called The Human Zoo in which he noted that zoo over-crowding of monkeys always seemed to lead to a rise in monkey colony homosexuality — natural birth control. I think we see the same thing today in the rise of the Millennials with their disdain for things like cars. As an American teenage male in the 1960s, having or getting a car was the single most important life passage. Not anymore. That’s an enormous change that lowers income requirements and makes it easier to function in an AI-inspired culture of economic limits. 

There are ways to go about the three Rs that are less jarring. Adopt European vacation, retirement, and higher education policies. If college was suddenly free, who would turn it down? “Sorry, I insist on paying that $370K for a Stanford education.” Yet in Europe, tuition fees are disappearing.

KAI-FU: Interesting ideas.  Very creative.  Thanks!  That’s the kind of creativity we need — that I brought up in Chapter 8.  Left alone to figure out the right policy, U.S. or any other country will most likely mess up.

I think there is still the “work makes me what I am” issue that these do not solve.  But then, maybe “work makes me what I am” is more Chinese, less American, and not at all European?

BOB: Don’t we all long to be European? 

Remember stories of how all those Japanese guys were working themselves to death? If you’ve ever spent time inside a Japanese company they are incredibly inefficient and a lot of that extra time is just spent drinking with the boss (at the boss’s expense). Still, those salarymen had something to prove so they’d sleep at their desks and never take vacation. What to do? Increase the number of national holidays. In the USA we have nine days of national holidays each year. In Japan they have 23 days — days during which the buildings are closed and nobody is allowed to go to work. The difference between 23 and 9 happens to be about the average number of vacation days taken by American workers. So those Japanese folks who are working themselves to death are still taking just as many “vacation” days (or more) than their American counterparts. 

KAI-FU: Yes these are like four-day work weeks. Should help, but the question remains: will you pay a routine worker 100% or 80% of the 80% workload?  If we agree that compassionate/service jobs like elderly care need to become more attractive, yet they’re already underpaid, making it 80% only hurts.

BOB: My point is Japan responded successfully to a national work crisis based primarily on neuroses and nobody noticed. They didn’t notice because the solution allowed Japanese folks to proudly overwork without actually overworking. There have to be other ways to effect change in the workplace without causing major pain. 

If we’re going to fault the USA (and I have no problem doing that), middle class wage stagnation since 1980 has hurt the ability of those workers to consume, hurting in turn the U.S. economy.  If we looked at these measures (the Three Rs) as helping the economy, suddenly those fat-cat CEOs might see things differently.

KAI-FU: U.S. wealth gap is the single item all people can agree on — it has gotten a lot worse, and will get much worse with AI.   With all this talk, all they can come up is UBI?

BOB: Couldn’t we do the same thing — just increase the number of holidays?

The ultimate economic argument that folks like Trump refuse to get is that the work force is also the consumer force. Export economies like Germany, China, and Japan see that less so, but China is right now going into a transition where its domestic consumption is going to be more and more important. Henry Ford raised his workers’ pay to $5 per day in 1908 and became the dominant car company as a result. He got it. Trump doesn’t.

I hope altruism does rise in the AI age, but I don’t see any way to MAKE that happen. Do you?

KAI-FU: I think we are on a path of the worst collision between nationalism, AI monopolies, lack of social responsibility, growing inequality, and joblessness.

But I think the future is going to depend on how we choose to face it, and our collective consciousness will determine our fate.  So I feel I might play a small role to take a more positive outlook and hope it might spread to a small number of people.  

BOB: We are definitely going into a period of enormous flux. I have three sons (16, 14, 12) and can’t imagine what any of them will do for a career. They have dreams and aspirations, of course, but the playing field is shifting so rapidly that it is hard to predict the future with any confidence. But what I CAN predict is that we won’t go into a dystopian future because my kids wouldn’t stand for that. There is a huge correction coming and the only thing my kids believe is that they will control that correction. Who am I to say they are wrong?

KAI-FU: My kids are both in the arts : )

My old friend Adam Dorrell’s company, CustomerGauge, is having an event on September 14th at the Computer History Museum in Mountain View, CA. I’ll be there as host, moderator and valet parking attendant. CustomerGauge is a Software-as-a-Service (SaaS) platform that measures and reports on customer feedback in real time.