Tools and Weapons with Brad Smith

Kai-Fu Lee: How AI teaches us what it means to be human

Episode Summary

In 2017, leading AI expert Kai-Fu Lee shared a dire prediction: half of all jobs – both blue collar and white collar – could be automated within ten years, replacing the workforce with solutions built on artificial intelligence. Brad and Kai-Fu discuss what this coming change means for national economies and for people who care about their work. Kai-Fu lays out practical steps policy makers can take today to prepare, the three areas he believes human intelligence will continue to lead, and why he remains an AI optimist.

Episode Notes

In 2017, leading AI expert Kai-Fu Lee shared a dire prediction: half of all jobs – both blue collar and white collar – could be automated within ten years, replacing the workforce with solutions built on artificial intelligence. Brad and Kai-Fu discuss what this coming change means for national economies and for people who care about their work. Kai-Fu lays out practical steps policy makers can take today to prepare, the three areas he believes human intelligence will continue to lead, and why he remains an AI optimist.

Dr. Kai-Fu Lee has driven innovation in AI research and development for over three decades. He is the Chairman and CEO of Sinovation Ventures and President of Sinovation Venture’s Artificial Intelligence Institute. Prior to founding Sinovation in 2009, Dr. Lee was the President of Google China, and a senior executive at Microsoft, SGI, and Apple. In the field of AI, Dr. Lee built one of the first game playing programs to defeat a world champion, as well as the world’s first large-vocabulary, speaker-independent continuous speech recognition system. His bestselling book AI Superpowers: China, Silicon Valley, and the New World Order discusses US-China co-leadership in the age of AI, as well as the greater societal impacts wrought by the AI technology revolution. His new co-authored book AI 2041 explores how AI will change our world over the next 20 years.

Click here for the episode transcript.

Episode Transcription

Brad Smith: I'm Brad Smith, and this is Tools and Weapons. On this podcast, I'm sharing conversations with leaders who are at the intersection of the promise and the peril of the digital age. We'll explore technology's role in the world as we look for new solutions for society's biggest challenges. Today, we're surveying the state of artificial intelligence, tracing its origins and discovering where it's headed.

Kai-Fu Lee: I am definitely an AI optimist, but I hope I'm a realistic optimist who also looks at the downside and tries to find solutions.

Brad Smith: That's Kai-Fu Lee. And there are few people in the world who have done more with artificial intelligence. His work at Apple, Microsoft, and Google really helped shape the entire field. Now he's in China leading important venture capital work that's funding future innovation on the other side of the planet. Today, Kai-Fu joins me to discuss how AI is revolutionizing the way we live and work. We'll also hear how a personal crisis transformed his life and his perspective on AI's role in the world. My conversation with Kai-Fu Lee up next on Tools and Weapons. Well, Kai-Fu Lee, thank you for joining me from Beijing today.

Kai-Fu Lee: Thanks for having me.

Brad Smith: I want to start with something that you've done recently that I think is both interesting and different. You haven't just done what so many in the tech sector do, write a book about your experiences or the technology, you've actually gotten into the world of science fiction. You've written a new book, AI 2041. It imagines and tells stories about the kind of future we may all be living in 20 years from now. What led you to choose this approach to think and talk about the future of artificial intelligence?

Kai-Fu Lee: I've personally been inspired by science fiction my whole life. When I worked on speech recognition, it was really the Star Trek Holodeck that I thought about and many other AI technologists like Rodney Brooks at MIT got into AI because of inspiration from science fiction. So it's been huge for us in AI because sometimes technologists and engineers, we lack the imagination of storytelling. AI is such an important technology area. I wanted to help as many people understand it as possible. And yet AI is perceived to be a rocket science and maybe intimidating to many people. So I thought science fiction can play two roles. One is to inspire the people working in AI and see the wonderful future they could help create. And the other is to explain AI to many people by telling stories in ways that are not only not intimidating, but engaging and perhaps even entertaining.

Kai-Fu Lee: So it was with this thought in mind that I approached a top science fiction writer, Chen Qiufan, to co-write the book with me using technologies that I believe are feasible in 20 years, but told through his pen in stories that are entertaining. And after each of his stories, I would write a commentary of how does the technology work, what good can it do, what problems might it bring up and how might we solve those problems?

Brad Smith: Well, I smile as I listen to you say this. It reminds me of a day I think it was probably when you and I were both colleagues at Microsoft together. I came home and that evening my son, who was probably about 10 years old at the time, said, "Dad, I figured out what Microsoft should do. Just get a group of people together, watch every episode ever made for any of the Star Trek series, make a list of all of the inventions that they have and go create them. And Microsoft will do very well." So I've always had this sense that in so many ways our collective future is something that can be imagined. And in some ways it is so inspiring as your book is. But before we get to the stories in AI 2041, I also want to go a different direction as well, and that's part of your journey. Your life stands for the proposition that sometimes it's our ability to imagine great things and achieve them and other days it's the ability to respond to what we hoped would be something we wouldn't experience. And you've done both.

Brad Smith: So let me turn to a day that you experienced in 2013. I know the way that you look at the world changed, including the role of AI. Can you start us with that day in your journey?

Kai-Fu Lee: Yeah. It was in the fall of 2013 when, after I had several health exams and they found possible tumors. And then I went in for a PET scan and I remembered that I went in and asked for the results and the radiologist was just unsure if he would be authorized to show me what he actually saw. And I made him show me. My whole stomach was full of tumors. There were some 25 or more tumors, and that totally shook my whole life in terms of my priorities, in terms of the possibility that my life may end, and then it really led to a lot of revelation and even enlightenment in thinking about the meaning of life, the meaning of artificial intelligence, meaning of my family, and so on.

Brad Smith: I think for many people who listen to this, it's of course a literally gut wrenching day to have to experience what you just described. One can perhaps more easily grasp how it would change the way you thought about the meaning of life, but can you share with us how it changed your thinking about the meaning of artificial intelligence?

Kai-Fu Lee: Yes. They're related. I was obsessed with work prior to the diagnosis. Really treating work first and everything else, including family, friends, lower priority. And the moment I got this diagnosis and looking at the possibility I may only have one or 200 days to live, made me realize that really work is not that important. What was important to me was my family, my friends, and especially those who've spent tireless time taking care of me during the chemotherapy. And it made me realize that the thing I thought I loved the most, which was my work, actually I didn't even want to spend one day working. I just wanted to wrap everything up and spend time with people I love.

Kai-Fu Lee: And that was an important revelation. And as it relates to artificial intelligence, one of the big issues we're faced with is the future of work. Many people believe work is the meaning of their lives. And to the extent they believe that, should artificial intelligence displace their job and take away their work, the big issue for the world is not just what happens to the unemployment rate or can we have universal basic income, but rather with so many people facing the loss of meaning of their lives, can humanity survive?

Kai-Fu Lee: So this revelation to me was that maybe we were all wrong. Maybe I was wrong in thinking that work was the center of my life. And given love is much more higher priority, what can we do to share what I've learned with the world, including the artificial intelligence community?

Brad Smith: I'm glad that those days are now in the distant rear view mirror. It's given many of us the opportunity to work with you and talk with you in new and additional ways. You shared in your first book, AI Superpowers, a little bit of your history. And one of the things that I found really interesting is the application that you filled out to enter PhD programs in computer science in 1983. And you wrote about artificial intelligence then. You said it was, "In part the understanding of what makes intelligence possible." And then you said, "It is men's final step to understand themselves. And I hope to take part in this new, but promising science." So your connection with AI and what it means for people to understand themselves clearly has a long history. You've seen so much unfold. When you think back about the course of AI and you compare it to what you might have expected in 1983, how has the future compared to what you might have hoped to see?

Kai-Fu Lee: AI had a lot of false starts, and disappointments, and winters, but I think we're now realizing my original aspiration of AI. AI is an omni-use technology that is being used in all kinds of industries to create a lot of value. We surprised ourselves how well machine learning algorithms work. It is now beating people and showing us that the things we thought required intelligence like medical diagnosis and playing chess and Go, really could be done computationally. And it makes us now focus on the things that AI cannot do. And they will probably either lead to a greater understanding of the human mystique of how we think, or it will lead to more breakthroughs delivering super intelligence. So I think we're ready to move towards that last step. It took 40 years, but I think we're basically there.

Brad Smith: If you were walking down the street and you met a proverbial person who had fallen asleep 40 years ago and they heard that you were working on this thing called artificial intelligence, how would you describe what AI in 2022 actually is?

Kai-Fu Lee: I would describe AI as using a completely different model of thinking. We were able to use mathematics and huge amount of data to learn and make decisions, make predictions, and build systems that exhibits phenomenal capability and intelligence. And that is different from the human brain in the sense that if you have a task that's relatively single domain with a lot of data that contributes to the final decision or prediction, then AI will beat humans every single time and by far. But if you have things that require analytical ability, common sense, and creativity or feelings and self-awareness and love and compassion, then AI is nowhere.

Brad Smith: Well, let's dig into that, because I think it's a fascinating dimension. If we start with where AI is excelling, one of the moments in time you've noted came in 2017 with the AlphaGo match. Can you tell us a little bit about that match and what it meant in terms of illuminating where AI may have the biggest impact on people's jobs?

Kai-Fu Lee: I think the AlphaGo was the single breakthrough that woke everybody up that the era of AI is coming. And that's because many people assume the game of go was not only so complex, but also required humans a lifetime to master. And also it required not only thinking tactically, but also strategically and even with Zen. And it's something that in the AI community people think couldn't be done with the class of algorithms that were being used. And all that was shattered when AlphaGo beat the Korean and then the Chinese champion. And it led to a lot of people thinking. I read in Dr. Kissinger's book, that this was the event that woke him up and see how this will affect the future of diplomacy. And it woke people up in China in thinking that this was the game that the Chinese invented, but now this English AI product is beating the best Chinese mind and it's something the Chinese companies and government need to focus on. So that was the wake up call for a lot of people.

Kai-Fu Lee: And in my first book, AI Superpowers, I related it to the Sputnik moment that woke the Americans up, that this space is a huge opportunity and any country and any company that doesn't want to be left behind need to focus on it. So there's global attention by industries, by academics, and by VCs and entrepreneurs. And just in the last four to five years, huge progress has been made. And because of the nature of AI, it is emulating human intelligence. So by definition, it will have implication on human jobs because once it can do certain task, then the next question is, can we use this software algorithm, which has nearly zero marginal cost to run, to replace what people do and save money?

Brad Smith: It's interesting because to help people think about the future, sometimes it takes an event, but it's not the only way to help people see the future and imagine it. And there are many people in the tech sector that try to talk about the future. You're one of the very, very few, maybe it's a group of about one, that has actually literally used science fiction to help people see the future. AI 2041 is filled with interesting stories. Do you have a favorite?

Kai-Fu Lee: Actually, my favorite is the first one. Even though the later chapters, many fancy technologies come in, such as quantum computing, and AI drug discovery, and other exciting things, my favorite is the first one because it talks about what is already happening today. The stories progress from technologies that could be doable today all the way to futuristic. The first story takes place in India. And as we know, India has been pretty successfully moving away from the caste system. And the story is one in which an insurance company happens to become too powerful, owning many apps in India from social, to commerce, and so on. And it ends up recreating the connections in the caste system and interfering with the heroine's life because it feels that if she were to fall in love with someone with a much lower caste, it would create problems for her in the society. And it starts to disrupt and try to break up her romantic involvement.

Kai-Fu Lee: I think that story is interesting because first an insurance AI story sounds pretty boring, but Chen Qiufan made it quite interesting. Secondly, it shows that the AI externalities cannot only be developed when the AI app has a misaligned interest with the user. In this case, the insurance company has an aligned interest with the user. They both want the user to be healthy and therefore have lower insurance, yet it can come in to interfere. And lastly, I think the story shows the AI problems, and issues, and opportunities are universal. So this story took place in India and the other nine stories take place in different parts of the world, I think to show that we all face similar opportunities and challenges with disruptive technology like AI.

Brad Smith: Well, I just have to say congratulations because the phrases insurance, artificial intelligence, and romance are not usually found in the same sentence, but you have managed to achieve that, I think remarkably well in your book. And when we think about the impact of AI on jobs, you have a chapter called "The Job Savior." And it starts with the story of someone who creates a business that involves job reallocators. What is a job reallocator in AI 2041?

Kai-Fu Lee: My co-author, Chen Qiufan, and I imagined that in the future, particular types of jobs would be displaced one category at a time, let's say entry level accountants followed by paralegals, followed by other types of jobs. And when that happens, we imagined that a lot of companies would have to replace that workforce. And then it would create an issue for the government of massive unemployment one category of job at a time. So the government ended up creating a program that funds job re-allocators who will go in and say, okay, how many are you going to lay off per month? And then the job reallocators would have a program that helps each individual based on that individual person's aptitude and interest, and try to identify a new job. And if that new job required retraining, it would help train them. So, that's the definition of a job reallocator.

Brad Smith: It's not the most uplifting part of the book, but it's very enlightening. It's sobering. It raises the question, gee, I wonder if my job is going to be reallocated in 2041, but what I also found interesting was your words after you told the story, because you said, "There are three capabilities where I see AI falling short, that AI will likely still struggle to master, even in 2041, creativity, empathy, and dexterity." Can you describe those capabilities and why you think AI will fall short in replacing jobs that have heavy doses of them?

Kai-Fu Lee: Right, by creativity, I mean ability to think across domains, think strategically, analytically and come up with new thoughts, ideas, or solutions that are out of the box. These are beyond today's AI because AI still requires a objective function, something to target, something to learn, something that a human programmer would say maximize minutes on my website, or increase my margin on my next product, or reduce the percentage of misdiagnosis in medical diagnostic program. So the human would give the goal. So here we have cases where the human thinks not only creatively about the goal, but also out of the box with concepts that didn't exist before. So I think that's an area that we currently don't know how to do with AI and we'll likely still struggle and certainly not be done with in 20 years.

Kai-Fu Lee: The second is compassion. And that's the human to human unexplainable sense of connection, or love, or empathy. I think it's something that's still hard to explain as a human phenomenon. Where does it take place in our brain, how do we emulate it, and can computer feel it? So we're, again, nowhere because AI is still quantitative and calculating and it doesn't feel anything. If you shut off an AI program, it stops running. It doesn't feel sad or feel bad. If it beats the world champion in Go, it doesn't feel happy. So it has no feeling and we don't know how to program it. We don't even know where to begin.

Kai-Fu Lee: The third one, dexterity, is one we're likely to continue to make incremental progress, that has to do with many years of our evolution and also we're endowed with our hand-eye coordination. So the fine tuning motor skills combined with thinking and problem solving. I think altogether, we call that dexterity. AI is getting increasingly better, but there's still many tasks that fall way short. So I think AI will chip away at that one, but certainly not be done in 20 years.

Brad Smith: One of the other points that you've raised, especially in AI Superpowers, is what all of this could mean for economics and for jobs in developing countries. If you think about the events of the last couple of years, including with COVID, what's your current assessment of what you think AI means for the economies of the developing world?

Kai-Fu Lee: Well, it depends on which developing countries we're talking about. I think, for example, China still considers itself a developing country, but in terms of AI development and the giants that have been created by AI, I think China is already catching up with the US. I would, for the purpose of this discussion, be thinking about maybe some of the economically more behind countries in Africa, South America, Southeast Asia, and so on. I think the economies that depend our hope to become China-like, or India-like, will have difficulties. China-like would be more of a manufacturing future where using lower cost labor, it could take percentages of outsource manufacturing business and build up its economy that way using lower cost labor. The India-like would be the white collar outsourcing to take away jobs that are lower paid and maybe less desirable by the Western world. I think both were excellent methods that brought up China and India and their economies.

Kai-Fu Lee: And those are not likely to be replicable in the future, because manufacturing will increasingly become outsourced to the robots and white collar work that has been outsourced to India will increasingly be done by software as well. So I think developing countries need to plot a new path. I think that developing countries should think about all three skill sets, but probably empathy is one that can be extended the most, that is turning into tourist industry, maybe having elderly care within the country. And we already know some countries in Southeast Asia are exporting healthcare service professionals to other countries. So those things are not likely to be doable by AI. They should also, to the best they can, educate a small percentage of the creative types, because that will be certainly the highest economic benefit, but the countries may have limited resources. So only a small percentage can do that.

Kai-Fu Lee: Dexterity would be another area. So that would be creating phenomenal arts and crafts work, that would be making products like the Swiss watchmaker type of things that AI cannot do, or maybe humans will still treasure handmade products. So those are the possible directions for developing countries.

Brad Smith: Is there some advice that you would consider universal? In other words, if you met a parent whose child, during the next 20 years, will have the opportunity to go to university or pursue something else regardless of where they live in the world. Is there advice that you would offer in terms of what that student should study?

Kai-Fu Lee: First, I would say focus less on rote learning because you can't ever beat AI at that. Secondly, develop your creative skillset. Third, improve your soft skills, communication, teamwork, empathy, compassion, gaining people's trust. I think that's incredibly important. And also I would suggest to follow your heart, follow your passion, because only when you do what you love can you be the best that you can be. And in a world where not just other people, but AI are competing for jobs, you really can't afford not to do what you're most likely to succeed at.

Kai-Fu Lee: And finally, I would say, think about new professions. Just as internet created many professions, AI will create many, many professions as well. These are probably jobs and opportunities that don't exist today and be on the lookout for them. Just like, you know, today data scientists are a great job to have, but 20 years ago they didn't exist. Uber drivers are another interesting class of jobs, but 10 years ago, they didn't exist. Be proactive and adventurous and don't believe too much on conventional beliefs. "Let's take this job. It's a safe job. We've been doing this for generations in the family. So you should do that." This is the time to really question that kind of conservatism.

Brad Smith: It reminds me of when I was a college student, when I would go home for a holiday vacation, I would take an airline flight to the Midwest. And I still remember flying back to the East Coast. By coincidence seated next to a woman and she said, "There's this industry that's going to explode in the future. It will be called software." And I knew next to nothing about it. I never imagined that it would be part of my future. But when I think about what you've written about in AI 2041, and I think even a little bit farther ahead to what I'll call the world in 2051, I have to believe that there's another industry that's just getting started today and it's called sustainability. And it has to do with everything that will relate to the climate. In fact, it's maybe the most universal issue, the one issue that even in a sometimes divided world can pull us all together. What role do you think AI will play as the world grapples with that issue and potentially this emerging industry?

Kai-Fu Lee: Yeah. I can see two areas in which AI can play a significant role. One is just to understand what sustainability, climate and these issues really involve. We observe the damage that's done to our planet, but the causes are not immediate. So it's a great way to apply AI. So first we have to collect the data, and then we use AI to diagnose the problem, and then we use AI to help find solutions. So that's a direction that could take a little bit of time because we might not have all the data we have. So that would be a great additional tool and that would have to be done globally and with collaboration. And that's an area, even given today's geopolitical challenges, I can see possibly good outcome because it's very clear every country needs to cooperate. So that would be a direction I could hope to see.

Kai-Fu Lee: The second is that I see the biggest culprit today to still be the use of fossil fuel that is not good for the environment. And how do we move towards an alternative energy? It seems like we're at the brink of moving towards distributed solar-based plus battery storage. We've already made huge progress, between 80 to 90% cost reduction on solar energy and also on the batteries that can store the energy for some period of time. So in my book, the last chapter, I talk about the future where this distributed energy that can really power communities, and buildings, and businesses, and farms at a much lower cost than today's electrical grid, energy that might be one-10th the cost of today. And that I think will dramatically make the climate and environment issues much better, but also interestingly, they will create low cost energy. When you have low cost energy, you'll be able to make drinkable water, you'll be able to dramatically reduce the cost of production combined with automation.

Kai-Fu Lee: So in making of a product, you would be reducing the cost of energy, cost of labor, and also to some extent, the cost of materials, creating potentially a world of platitude. And that really comes down to the role that AI can play. AI and automation can make robotic and manufacturing work much more effectively and cheaply. And we will be transforming more and more of the world's resources and processes into manufacturing. So energy will become not a dependency on oil and natural gas, but making batteries and solar panels with better and better technologies and costs. And also that will spill over to agriculture. We will be having vertical farms and 3D printed stem cell-based meat that will further improve the environment and also turn more things into manufacturing.

Kai-Fu Lee: So my belief is that among the AI processes, automation will be able to reap multiple benefits as energy, agriculture and food become manufacturing problems. And if AI and automation can reduce the cost of manufacturing, then it will bring about not only a better environment, but lower cost of goods, lower cost of energy, and lower cost of food, and hopefully eradicating hunger and poverty in the process as well.

Brad Smith: What I find so interesting, Kai-Fu, is there are some people, especially in the tech sector, that talk only about the positive side of what technology or AI will do. There's other people, sometimes they're in the tech sector more often, they're outside the tech sector, who talk only about the negative side. One of the things I really love about both of your books, AI Superpowers and AI 2041, is you talk about both, the opportunities and the challenges. At the end of the day, are you an AI optimist, or are you an AI pessimist?

Kai-Fu Lee: I am definitely an AI optimist, but I hope I am a realistic optimist who also looks at the downsides and tries to find solutions. So what I aim to be is a problem solving optimist rather than just starry-eyed naive optimist.

Brad Smith: Well, you speak with a lot of experience and the wisdom, I think that has come from it. Let me just say thank you for joining me today. I am one of many who has benefited from all the amazing things you have done since that day in 2013 when you thought you might not have so many years to live. So thank you, Kai-Fu. Thank you for what you've been doing. Thank you for joining me.

Kai-Fu Lee: Thank you, Brad. Thanks for having me.

Brad Smith: You've been listening to Tools and Weapons with me, Brad Smith. If you enjoyed today's show, please follow us wherever you like to listen.

Brad Smith: Tools and Weapons is produced by Corina Hernandez, Mark "Frosty" McNeill, and Jordan Rothlein, with production assistance from Emma Foley. Our executive producers are Carol Ann Brown and Aaron Thiese. This podcast was edited and mixed by Jenny Cataldo, with production support by Sam Kirkpatrick at Run Studios. Original music by Angular Wave Research. Tools and Weapons is a production of Microsoft, made in partnership with We Are Listen and A_DA.