Below is the transcript of a CNBC interview with Leo, Gang LIU, VP, SenseTime & GM, SenseTime's Shenzhen Office and CNBC's Eunice Yoon. The interview took place at CNBC's inaugural tech conference, East Tech West, in Nansha, Guangzhou.
EY: So, Leo Liu, VP of SenseTime, he has spent some time in Beijing, along with me, for several years, before moving over to Shenzhen-,
EY: So, I just want to pick up on what Mandy had said. You're a young company, obviously very important in empowering AI, and we know that the government's national directive has been to focus on AI, turn AI in to a $150 billion industry by 2030. So, what does that mean, for some of the experimentation that you're seeing, when it comes to applications? So retailers, for example, or for, as Mandy had said, anything on your mobile phone?
LL: Yeah, thank you. If course, AI, actually, the value of AI is to make a lot of things, impossible in the past, possible nowadays. For example, like, retail, like transportation, we always imagine that, for a lot of repeatable, or repeated, routine work, we always that-, okay, if there's some machine there, to do the job for humans, that's terrific, especially for some low-value work. But nowadays, a lot of applications are already there, for example, (inaudible) you arrived here from Guangzhou airport-,
LL: So, some of our technology has already been running there, but you didn't notice that.
LL: For the security pass, actually, the challenge there is to make sure that you are using your own ID to take the plane, so it's much safer than just to rely on people. So, like, retail, we have quite some innovation, with retailers, like Walmart, like (inaudible), a lot of, like, AI shops have already been opening up in China. So, AI in the past, like, more than 60 years, has been categorized as very interesting, it has been very interesting. But nowadays, it has been labelled as very useful, which means that technology, if we put technology to the industry, it can generate a lot of value.
LL: So, we are quite fortunate, and we are very excited that, the government, this country, the market, even the user, our customers, are kind of embracing the new technology, so we have been very fortunate, I think.
EY: Yeah, well, you mentioned going through the airport. Now, after visiting your headquarters in Beijing-, I think a lot, every single time I go through Beijing airport, about how the security situation has been changed, and the equation has changed, just because the facial recognition technology is largely what you guys are working on, right? So, in terms of application. So, one of the other questions that I want to ask you, in terms of-, about the-, the government policy, is that, you know, obviously, government sets a policy, creates a lot of experimentation opportunities-,
EY: For startups, for big retailers-,
EY: Like you had mentioned, including Walmart, I know that they're doing some experimentation. However, industrial policy doesn't guarantee success. So, we've seen, in various countries, including in China, where the government has set a policy, so, for example, in developing a domestic car industry, it had mixed results. So, what other advantages do you see in AI that make you so optimistic about the prospects?
LL: Yeah. SenseTime is a young company, like you mentioned. The average age of the staff is only 28. So I'm kind of the older generation of the company [laughter].
LL: But it's a young company, but with a long history. In Chinese, we have a saying that [proverb in Chinese], I'm not sure how to put that in English, it's kind of, just-, it's like a quick kick-off, but with a long accumulation, kind of like that. So, first of all, like, SenseTime, we have a long history. Before we started the company, we have a long history of research. Because the value generated by AI is not, kind of, invented or created. You must push the technology to the limit. Which means, if you want to create a value, you have to always try to be better, better and better. Take accuracy, for example. SenseTime is a company focused on visual intelligence, and deep learning, so the-, the criteria for the technology to be valuable is, of course, that your accuracy must surpass human performance. So, I think one advantage we enjoy is that we have a long history we focused on research. And another one is, we are quite value-oriented, which means-, technology, by itself, cannot generate value. You have to apply the technology to the specific industries, or the specific scenarios.
LL: And, as we understand that China is kind of a-, a-, a nation full of-, full of spirit of innovation, so a lot of industries kind of embrace the new technologies, which give us a quite good chance to put our technology to the industries-,
LL: So that value will be generated there, which means that we make-, of course, maybe you have some idea in your industry, but you don't have the way to realize it, in the past, so I've provided that-,
LL: Way of realizing your idea, in your industry, so-, and value generated for you. And the last one I want to mention is that China has been enjoying the quick development, in terms of the economics, so a lot of talent, top talent, are now, kind of, gathering around China-,
LL: For example, like, our big bay, right, that's in the big Bay Area, a lot of talent is coming back, and, take our company, for example, we have staff around 2,000, but we have more-, we have more than 150 full-time PhDs in our company. So, they are continuing focused on advanced technology, and we (inaudible) this technology, the algorithms, the technology, to the specific industries, and the value will be generated there-,
LL: And with the macro support, from the government, from the industry, from the customer, so we're kind of enjoying this era, and doing good.
EY: [Laughter]. Not so bad. Well, I've heard that-, I've heard that quite a bit, actually, especially when it comes to-, when people are comparing, say, silicon valley-,
EY: And-, and China-,
EY: And various parts of China, that are working in AI, that in the United States, there are-, there's a lot of really solid technology, but here in China, it's where all the applications and experimentation is-,
EY: And that that's a big advantage, and part of that is because of the data that is here. I was reading an interesting stat that Chinese companies collect data on 1 billion active mobile phones. That's a lot of phones, and a lot of data-,
EY: So, what kind of position, then, does that allow you to be in, as well as other startups?
LL: Okay. In terms of data, I think we need a clarification here. So, data, of course, is kind of a key success element, to have advanced AI technology. So-, but the data are actually collected and used in different scenarios. For example, when you are doing the learning or training of AI, a lot of data actually are obtained from the internet, with the permission from the user. But when you use the technology, for example, you mentioned that-, that, kind of, huge data, 1 million, 10 million, for example, that-, like I told you, that more than 400 million cell phones are actually using our technology, like face unlock-,
LL: Like the enhancement of pictures. Maybe you-, you-, you didn't notice that, but a lot of phones, the pictures you are taking by the phone, the quality is higher and higher, right? So, actually, it's-, it's-, it's not because of the camera itself, but also because of AI technology. So, (inaudible) have already been running on hundreds of millions of phones, but we don't collect the data-,
LL: It's because we kind of generated the algorithm, so this algorithm, after, is put in the smartphone, the phone, by itself, uses this technology. SO, because China has been enjoying the advantage of, kind of, going to the mobile internet era, quite quickly-,
LL: We skipped the PC era, so this kind of flexibility, and huge amount of internet users, and the spirit of innovation, kind of have given us the scenarios to apply our technology to this huge market.
EY: Now, you made a point to-, to say that you do not collect the data, and there is a lot of concern about privacy issues-,
EY: When it comes to AI-,
EY: Especially with facial recognition, or any other type of body recognition technology, like, either with your hand, or with your eye. So, do you worry about a pushback-,
EY: From consumers, or from anyone else, about-, for-, because of privacy issues?
LL: Yeah. No, actually, not. Actually, on the contrary, not, because-, I'll give you a very specific example. As staff of SenseTime, we do not have a (inaudible) badge. Because the badge-, the purpose of the badge is to open your door, right? When you go to your office every day, you need a badge to walk in the door. We don't have a badge, because our face is our badge. So, we now-, we-, yeah, I mean, if you want to go to the office, I mean, to work overtime, you don't need to bring a badge. I mean, even-, so-, take (inaudible), for example. The employees in our company, they know that they use their face to open the door, and they all, of course, permitted that, because to go in to your office, you use your face to open the door, it's more convenient, it's fast, and it's more secure. And-,
EY: But these are a very, like, selected group of people, who all want to work at your company, so they would-,
EY: Feel more comfortable, with the idea of-,
EY: Having their faces scanned-,
LL: So, my point is that we don't collect or use the data without the permission of the user. So, as long as we get the permission from the user, and the user, of course, enjoys the value generated by AI, and when they get used to use the AI, in their everyday life, they will actually like the AI technology, more and more. So, that's kind of worked for the past more than three years, four years, we are actually just celebrating our four year birthday, this week, so for the past four weeks, we have signed almost more than 700 business customers, and hundreds of millions of users. So the feedback from the customers? Fantastic. I mean, they have-, they have been enjoying the-, the convenience, the safety, I mean, the smarter life generated by AI technology.
EY: Mm. And I think that's what makes a lot of sense, that you see the convenience and the-,
EY: As the benefits of-, of AI, in particular of facial recognition technology, of course there is a trade-off, as I said, with, with privacy. I wanted to push back a little bit-,
EY: And also ask you about the accuracy-,
LL: Accuracy, yeah.
EY: When it comes to facial recognition technology. Because, obviously, the technology's getting better and better, but it's not perfect, and if you're talking about applying it to a lot of different areas, either with-, you know, to try to help with the bureaucracy, or here in China, the national ID system, or trying to use it as a way to withdraw money, say, for example-,
LL: Yeah, yeah.
EY: Then you have to make sure it's pretty accurate-,
LL: Sure. Actually, accuracy is kind of a milestone event in the history of visual intelligence. So, why SenseTime is, kind of, founded in 2014? Because a kind of milestone event happened in that year. In the-, the specific dataset, it's kind of an international dataset, it has more than 1 million faces, so, the theoretical accuracy of human performance is kind of around 97%, so which means that, in that dataset, a lot of face pictures there-,
LL: If it's a normal human being, you will make two mistakes for 100, I mean, kind of, verifications. But in that year, in 2014, the algorithm from SenseTime, for the first-, very first time, in the history of human-, in human history, the accuracy beat the human performance. And nowadays, for example, when you use your ID, to go through the security, to take airplane, it's normal people, right? The security guard is comparing you to your ID photo, in your passport or on your ID. It is accurate, but it's not so accurate-,
LL: So-, because we are human beings, right-,
LL: This theoretic accuracy is almost 97% accurate.
EY: So it's better than the human eye [laughter].
LL: Yeah. But the machine-,
LL: Nowadays, for that particular scenario, it's 99.9-, then seven 9s, behind the point, percent. So it's almost-,
EY: You're very precise. [Laughter].
LL: It's almost no mistakes.
EY: Well, just, I-, one-, one quick last question, because I have only 30 seconds left, but how worried are you about the trade tensions that we're seeing with the United States? I know that, right now, you're focused on the China market, but you guys have been interested in pushing at the technology, globally. IS this something that you're worried about, that you could see a pushback?
LL: Actually, no, because as a tech company, I have been spending my whole career in the IT industry, so I can see that the technology actually is joined by China, US, I mean, Singapore, South Asia, and Africa, the Middle East, so I don't think it's going to be a challenge for us-,
LL: So I'm quite optimistic.
EY: Thank you very much.
CNBC is the leading global broadcaster of live business and financial news and information, reporting directly from the major financial markets around the globe with regional headquarters Singapore, Abu Dhabi, London and New York. The TV channel is available in more than 410 million homes worldwide.
CNBC.com is the preeminent financial news source on the web, featuring an unprecedented amount of video, real-time market analysis, web-exclusive live video and analytical financial tools.
CNBC is a division of NBCUniversal. For more information, visit www.cnbc.com.