This is Bloomberg Business Week with Carol Messer and Bloomberg Quick Takes Tim Stenovic on Bloomberg Radio. Kai Fu Lee has spent nearly forty years at the forefront of artificial intelligence. He's the chairman, CEO, and co founder of sign Ovation Ventures. It's a venture capital firm with about two point five billion dollars in assets. It focuses on AI, healthcare, and other high tech and has made more than three hundred investments.
It's one of the first Chinese VC firms to have a presence in the US, and he's also the co chair of the Artificial Intelligence Council at the World Economic Forum. He's the former president of Google China and before that was an executive at Microsoft and at Apple. His newest book is AI Ten Visions for Our Future, and unlike his last book, it's a work of fiction. He and his co author, the sci fi novelist chen Cho Fan right ten stories. They're all set twenty years from now,
in ten forty one now. These stories illustrate a world that's layered with artificial intel agents. They cover themes like AI displacing workers, deep fakes, and how a I will reimagine education of course, also imagines what could go wrong. Kai fu Ley joins us now from Beijing. Kai ful, great to chat with you. Thanks so much for taking the time. Why are these stories? Why is sci fi? Why is it the best way to communicate what you envision AI doing to society in twenty years. Hi great
to have to be a here. AI is such an important technology, perhaps the most important in the history of mankind, and yet it's sometimes viewed as rocket science, very hard to understand. So I wanted to see if there's a way to make it accessible to people and that it's actually not only understandable, but also fun to read. That's why I uh found a science fiction writer to work
with me. He writes the stories, I write the analysis, and then we make sure that the stories are realistic within the twenty year time frame, so that the stories you read will be roughly what life will be like twenty years from now. Each story has an introduction, it also has an analysis after the fact. What were you trying to do here? I wanted to explain the how the technology works and what it's good for, why it's why it's good at what it's good at, and also
why it is almost always a double ex short. What makes it incredibly powerful also creates some downside. For example, AI is great to help Facebook and Amazon make a lot of money by showing us things that we will watch and buying, But then are those things good for us because that's not being considered in the AI run by Facebook and Amazon. So there is a flip side to this, and the story is really not just tell the technologies but also bringing up issues that they anology
causes and also suggests possible solutions. Do you think the companies that deploy AI today that are working on solving these issues, do you think they're taking into account the pitfalls the perils that you associate with AI. I think they are trying. But because it's a new technology, so uh, they're the tools to combat the problems aren't there yet, and the engineers building the products weren't aware of the problems.
None of the problems were intentionally created, you know, issues with respect to personal data and explainability and what happens when AI makes a mistake and accountability uh, and also bias. So these were not intentional by the companies or the engineers who build it. But now that we know these are real problems, we need to put the training regulations UH and investigating new technologies to come back techno oology problems.
Do you think the incentives, though, are aligned for companies that do work on this type of technology to create solutions or are they mainly concerned with us getting to to buy more, to engage more, to to read more, to spend more time on the platform without being concerned about the consequences or while you really hit the problem on the head that it is exactly because the incentives are not aligned. So I think the only way to
fix it is to realign the incentives. And in the short term, the internet companies could have an incentive realignment if there were either regulations or certain parties that may do an AI audit, or maybe there would be a watchdog functions that are estimating percentage of fake news and things like that that forces them before their brand and reputation to consider that because if they lose their brand and have a some kind of a media disaster, they
will lose customers. So that would is one way to externally force an approximation to an alignment, but there are also other ways. Why can't we not encourage building of applications and products that help us in the long term. So if there were a product that could make me more intelligent, more knowledgeable, more wealthy, m more happier, then that would be a product I would want to buy more than I would want the Facebook news speed, um or YouTube. So I think there are ways to align
user interests. I think people just um the company's just felt like they got such a good AI tool. It can make money directly, so use it for that singular purpose, um and forget really that companies should be focused on user needs first, and users want more than um just
watching some videos and and reading some blogs. There are long term goals that we have, and when we have such a powerful AI tool that can optimize our long term goals, it would be such a unfortunate, um events if nobody tried to build something that that's for our long term interests. So let's stick with Facebook because it's one of the companies that you mentioned. So who should
regulate Facebook? Should it be the U S Government or should there be some sort of international body that takes these difficult questions into account to come up with some regulatory framework. After all, for all intents and purposes, Facebook is a global company, right right. I'm not an expert in regulation, and I think regulation, in my opinion, is just usually the last resort because, especially given the tools that the US has generally to use, such as breaking
a company up, that's just too brittle. So I think there needs to be ways to push the company into an alignment with the goals that we the users have, and I mentioned a few of them. Uh. You know, when I work for Google, there were third parties that measured the relevance quality of Google Search Engine, and Google
paid a lot of attention to that. So if there could be similar third party watchdog measures of level of fake news, number of deep faith, and uh, how much spam uh, and then it's measured externally, that would put a pressure on the engineering team to figure out at least to spend part of their time to make the content higher quality, uh, instead of just getting us to
click more and more. So, okay, I feel let's take a step back and just think about how we interact with AI ine just just take us through, because I think that there are a lot of people who don't necessarily think about it. Each and every day that we are interacting with this technology, we are aariensing this technology so so what does it look like for in our daily lives globally. Sure, I'll give a couple of examples.
So one example is that um Ai language capabilities to converse with us will grow significantly in many cases indistrestinguishable from human human communication. And that means Ai will become a companion. UH. And that companion can learn our interests, can try to help us improve ourselves. That would be an educational companion or to figure out how how we might be healthier. That it might be a more of a doctor or psychiatrist companion, or it could just be
a friend, a power for for a young kid. And they take and take shapes in UH, both in the in the phone and the computer of course, but but also v R and a R will be really realistic by then that if we put on something like glasses, we will see figures and avatars superimposed on the glasses and and then there there there is as though they were in the actual environment. So it would be um
really virtual reality that's indistinguishable from the real reality. UH. In transportation, autonomous vehicles should just about work by then get all the kings worked out. Of course, throughout the way we will see incrementally robot buses and shuttles and trucks and so on. But truly, the steering feel wheel free car that we can either share right or buy
will be just about fifteen to twenty years from now. UM. That would dramatically change our habits because we will have um something like ten hours a week freed up that when we're in the car we can educate and entertain, communicate, no longer have to drive or look at the road um. And also cars can be on demand available um. And they will be smart, they will talk to each other, they will avoid accidents, they will miss each other by one's enemy there in a very calculated way. They will
drive more safely, probably eventually reducing fatalities. So it will be more efficient, lower costs, save us time, more comfortable um and and also safer uh. And finally, in healthcare, a I will be able to with a very high likelihood predict the next pandemic because it could be watching
a lot of data all over the world. And AI will be used to alongside with scientists invent new drugs, and that the cost of new drugs probably could be as little as ten of new drugs inventions today and what that means is today's pharmaceuticals can only go after
blockbuster drugs for common diseases. But in the future, when the cost is slashed because AI is proposing treatments that are highly accurate working with scientists reducing the cost, that means rare diseases for which is not economical for pharmaceuticals
to book go after will become possible. And and also AI will be keeping us healthy by taking all of our m R I and gene sequencing and blood to regularly monitor our health and give us hints on how to be healthier and live longer, and both of which will happen in twenty years. You you paint a picture that I think is really optimistic when it comes to this type of technology. But I'm wondering what your biggest
fears are regarding AI. Yeah, AI, like any technology, is neutral, and it's good or bad because of the people who program it have good or bad intentions. And the examples that gave earlier are all constructive, but there are definitely cases where AI can can do terrible things. AI can be used by UM companies to change the way we think. We already saw that to influence the way we think, and that's extremely dangerous. AI can be used in uh.
The most disastrous situation is autonomous weapons. Imagine a thousand dollar drone combined with recognition of individuals with a bullet connected to it, and that becomes an assassination tool, and it's deployed in the tens of thousands, it becomes a tool for genocide. So that essentially makes terrorist groups that have a few million dollars they can almost um uh do what countries do to create havoc can damage without risking any terrorist lives. So that's something that's um I think,
extremely dangerous. And and also unlike nuclear weapons, which uh offers a unique property that is mutually destructive, so people don't launch it before fear of retaliation, but autonomous weapons are not totally destructive and not easily detectable, So that's another serious problem. And and lastly, we already saw the facts, but that's really just the beginning. AI will be able to fake many things. It will be able to make videos that look really real when it's off fabricated UM.
It will be able to talk like someone quote quote room evidence. UH may maybe maybe at the point where people can't tell if it's fake or real anymore, not to mention rumors on the and false rumors on the Internet. And also AI will be able to fake emotion, so you know the scenario we saw in the movie Her. You know, while AI doesn't have any feelings, it can fake feelings to a point that could be used to do bad things as well. So I think that interest
cases are just as significant. But my optimism is based on the fact that all technologies that have been incredibly powerful, you know, electricity, automobile, internet, PC, UM, they have caused problems from time to time, but ultimately they end up doing a lot more good than evil because hopefully we are learning from experience and have some wisdom to um to self managed so that we figure out ways to counter the bad guys and encourage the good guys. Well,
how do we do that? It ends up being all constructed because I can't help, but think that those do we counter Yeah, yeah, because they can't help, But think that those problems are of a different magnitude than problems created by other types of modernization. It really makes me think of the sci fi series Black Mirror and the way that that show really lays out so so many of the dystopian futures that you could think about when it comes to this type of technology, how do we
prevent that type of future from overtaking us. Yeah, I'm a fan of Black Mirror. It's partly inspired the book, and I think in the Black Mirror, though, it's usually uh a technology is used in the wrong way by some some part bad person and then it leads to
usually a disastrous ending. In in my book, I try to create scenarios where many of the bad endings were saved by the by the good guys, and that might be saved with some clever technology that combats the bad technology, or with some smart tools prevents the back technology from propagating, or maybe there's some new regulations like the AI audit.
So what I wanted to propose in the book is that yes, AI and like any other great technology, will create a lot of problems, but many of the problems can be greatly reduced if we let technologies work on solutions that address the problems created by technologies themselves. I gotta ask about your interpretation of of what's been happening in China and your interpretation specifically of President She's cracked down on technology I'm wondering how you interpret Beijing's measures
against d D, Shu Shing or Jack Moll for example. Oh, these are not really technology firms per se. And we invest in technology, invest in robots and AI and healthcare and you know, new cures and new drugs. Those are actually heavily encouraged. UM. What those companies are d D and Ali Baba and Tensent and others are really I think the issue is that they have a lot of power and a lot of days done, and that gives them even more power, and that power needed to be checked.
So not at all, you know, consistent with what US and EU are doing to rein in Google, Facebook, UM and Amazon. Given what's happened in China over the last year, when it comes to reigning in these tech firms, has it changed the way that you're thinking about investing in Chinese companies? Does it make you think differently about how to deploy your money. Well, actually, we've always been investors in deep tech, so now it's just a better time than ever to keep doing what we have been doing.
UM deep tech hasn't been as favored. They haven't seen as much growth five or ten years ago compared to the Internet giants. But I think going forward, deep tech um is going to generate the best returns, probably globally, certainly in China. You told Bloomberg a few years ago back in of your money was invested in in Chinese companies. Is that consistent roughly yes, still the case, even more
given COVID. How so what do you mean, Well, we're unable to travel, so we can't really see a lot of companies outside China, and you think you need to see companies in order to know whether or not to invest in them. That would help to build trust, to see the people, to understand them. So I think generally the answer is yes, we have made one investment in the French company without seeing them, but that detainitely the exception,
not the rule. So that brings me to the next question is how do you think the pandemic will will affect China, the tech industry and in China's place in the world. The pandemic has really in China pushed further the adoption of automation and robotics. Initially it may have been for reducing social contact UH by having robots in UH factories and restaurants and even homes, I get my um takeoff delivered by by robot nowadays, So I think the robotic revolution is probably the big big step forward
that will be helpful to push China's technologies ahead. Um. On the flip side, UM, the other big thing that's happened globally is the work from home and and that has really streamlined work for a lot of companies and also enabled the use of uh ai and cloud technologies. And that's why where I think US has really done by far the best job, partly because the US has always had stronger enterprise technologies, but also partly because Americans
were um working from home longer than the Chinese. So I think each country has got some benefit um um out of COVID, even though it's terrible disaster for for the whole world, I feel things have changed just quite a bit since you picked your computer science as your college major years ago. It was of course before you went to get your PhD from Carnegie Mellon. So if you could go back in time, what advice would you give your eighteen year old self, what advice would you
give to college students today? Um, I would say back then I going or today, following your heart, do what you really love to do. Is It's the same advice I give to myself back then and also to a young person today, perhaps even more important today because um, a young person is not just competing with other young people, but also AI is taking away routine jobs. So it pushes us to do things that a I cannot do, whether it's creativity or compassion, and it forces us to
be the best we can be. And the best we can be is doing things that we're good at and the things we love. CAIFOOLEI thank you so much for taking the time and joining us in Bloomberg business. We really appreciate it.