The author, thinker, innovator, and investor popped by AI2 (virtually) to discuss AI’s ability to cure rare diseases, cleanse the earth, even end poverty—as well as its dark sides, too.
If you work in the AI space—even as a tinkerer—you’ve stumbled upon the work of computer science powerhouse Kai-Fu Lee. The Taiwan native moved to the United States to attend high school in Tennessee, went on to be a classmate of Barack Obama’s at Columbia University (where he graduated summa cum laude), and published his PhD dissertation on automatic speech recognition as a monograph. Lee eventually landed top-level jobs at Apple, Microsoft, and Google, then started his own VC, Sinovation Ventures, which currently has some $2 billion in assets under management across 300 companies, primarily in China.
This month, however, the author is focused on his latest book, AI 2041: Ten Visions for Our Future, a follow-on to his New York Times bestselling 2018 tome, AI Superpowers, which described how artificial intelligence will reshape geopolitics and business practices around the globe. Released in mid-September, the new book is, quite unusually, co-authored with Chen Qiufan, a renowned Shanghai-based Sci-Fi writer who once worked for Lee at Google. Composed of 10 essays written by Lee on a specific subset of technology and its predicted impact on the next two decades — each followed by a fictional take on the same tech — the book provides an entertaining read: part HBR and part William Gibson.
Lee joined us virtually from Beijing to discuss some of his predictions for AI in the near future and why this technology can elevate all of humanity and give us back the one commodity that cannot be purchased: free time.
AI Will Revolutionize Healthcare
“This is the most compelling application for AI and will have the biggest impact in the next 10 to 20 years. Ninety-nine percent of genome sequencing is not looked at by scientists, but is surely useful. Given such large amounts of data over large populations measured longitudinally, AI is poised to come up with new ways of diagnosis and treatment. That said, I want to be cautious, because IBM Watson and Google Health faced major challenges. Established experts are naturally repelled by the thinking that a lot of data will lead to statistically better outcomes, because doctors are trained to view every life as sacred. They don’t look at numbers as statistics as many of us do, and the two cultures will clash. A gentle way to introduce what I believe are revolutionizing innovations to the medical community is to offer them as tools that experts may choose to use, just as they may choose to use blood tests. If we are proven correct, the accuracy will be compelling and doctors will naturally defer to AI.”
“I am particularly fond of drug discovery, because it is using AI as a way to accelerate new remedies. It doesn’t change the process or replace humans. We invested in a company called Insilico Medicine that has two drugs in clinical trial for pulmonary and kidney fibrosis. Basically, AI found the target within the pathogen and proposed a small molecule solution that a human chose. When it’s all rolled out and proven, we think it can reduce the cost by a factor of 10 and the time by a factor of three. Pharmaceuticals can use this tool to treat rare diseases, and we can tackle a lot of rare diseases.”
AI and Longevity
“I’m a big fan of working in areas where there are no incumbents. Very few people, David Sinclair at Harvard being one exception, work on human longevity. It’s really looking at the world where all this data can be fed into an AI that compares you to other people and gives you advice on what parts of your lifestyle you might tweak in order to perform better. The AI gathers all my data every quarter through a blood test, and it compares me with myself three months ago and tells me whether my new exercises or sleep changes have been effective. Being on this regiment for the last year, the appearance of my blood has been reduced by six years! So I’m excited by an AI that is just me and my data, and doesn’t impact the world.”
AI Will Accelerate Transportation
“I believe there will be a need to have smart infrastructure, like highways that have sensors that talk to cars. Why should cars have to look at signs designed for humans? I also can imagine smart downtowns that will separate pedestrians from cars. That requires a redesign of downtown, or even building a new city. In China that is now part of urban planning, and we are seeing promise in autonomous vehicles, although that is a decades-long endeavor. I also see autonomous vehicles as becoming operating systems for robots. They will have deep-learning, instantaneous response, and the ability to see, hear, move, and manipulate. The cost of sensors will come down, so there is a potential for a giant company to emerge that will challenge the dominance of Windows and Android.”
Automated Food Production
“Automation is the only path forward for China, because people are making more money and the cost of labor is twice that of Vietnam. In terms of automation, we are seeing development, with computer vision as the low-hanging fruit; next is manipulation, then grasping. And, of course, drones are part of the equation. So we are very heavily invested in automation…autonomous harvesters and pesticide delivery, self-replication…the automation of vertical farming and even ethical, 3D-printed meat using cow molecules…[will] dramatically change the future factory.
AI Will Create Clean, New, Smart Materials
“We can see that clean materials can be created one molecule at a time, and there are databases being created with these molecules. No longer will we have to go to take oil and make plastic out of that toxic fossil fuel. We can go directly to build these materials one molecule at a time — materials that simulate rubber, plastic, fragrances, cosmetics, etc. So the cost will come down considerably.”
These advances, and others, Lee argues, will result in an Age of Plenitude. We, as a society, can surely provide a decent living to everyone and eradicate poverty and hunger, probably in 20 years — so long as richer countries help poorer countries, big companies reduce their costs when the price of R&D and manufacturing plummets, and wealthy individuals consider the plight of poorer folks. “AI can possibly eradicate poverty and hunger, and AI and automation can replace routine work,” says Lee with enthusiasm. “We’ll have an opportunity to evolve our work ethic, to pursue things we love, to do things we are good at. We will be liberated from doing repetitive work.”
Of Course, Not All AI Will Be Used to Benefit Humanity
Though he remains an AI optimist (“about 66.6% optimist,” he insists), Lee’s book discusses some ethical quandaries, such as AI’s use in creating autonomous weapons and eradicating privacy. He mentioned how all new big technological discoveries were used for ill at their birth, from the electric chair to viruses and disinformation on the internet, and that some institutions should be studying Doomsday scenarios. Mostly, however, he believes that big, terrible events have caused the whole world to wake up and understand how to respond to new types of weaponry.
When asked specifically about the use of AI in China to create a surveillance state to target minorities, however, the author was more coy. “I think, actually, the Xinjiang situation is too complex for me to describe in a credible way, and there is an explanation from the state in which I can see their point of view—though we may not all agree. So I’ll pick a different angle: Covid.”
Lee then went on to explain how facial recognition, phone scans to prove a person is negative, and full contact tracing are in effect in China, which he admits would be perceived as giving up privacy by Americans. “But I think the justification is that China is more of a collectivist society, where the safety of all is far more important than the personal freedom of any one individual. That is our social contract.”
China is on par with the United States, he says, with about a dozen AI companies up for IPO in the next year, mostly in automation. He is surprised to see that academic communities in the two countries continue to work together, despite the adversarial relationship between their governments, but remains cautiously optimistic about the future of AI collaboration between China and the United States. “I didn’t expect it could get this bad, but it’s a world we have to accept and figure out how to work within.”