Interview With Hu Guoping, Senior Vice President And Research Institute Of HKUST, AI: The Impact Of The Bubble On The Head Business Is Small. The Biggest Opportunity For The Future Is In The Education Industry.

During the two sessions, five written proposals submitted by NPC deputy and Liu Qingfeng, chairman of the University of technology, respectively, involved hot topics such as wisdom education, public health, smart offices, private universities, and tax policies.

Under the background of the epidemic, AI application is also the focus of the market.

Around the topic of AI application, technology development and the development of its own enterprise, Hu Guoping, co founder, senior vice president and Dean of Research Institute of the University of science and technology in May 28th, received an exclusive interview with the economic report in twenty-first Century.

Growth point conjecture

Twenty-first Century: this year's two sessions of Liu Qingfeng suggested that we should focus more on the specific application of artificial intelligence. What is the background?

Hu Guoping: first of all, in recent years, the trend of artificial intelligence is becoming more and more obvious. More and more companies combine artificial intelligence with specific industries and scenes. Second, people used software to implement services, and now there are some corresponding hardware, such as special intelligent recorder and flying interpreter. The third is chip technology. From the industry perspective, China's R & D in AI related chips and other basic fields has been strengthened. Fourth, the impact of the epidemic on the application and value of artificial intelligence has been increasing rapidly, including the use of intelligent medical assistant telephone robot, the use of artificial intelligence outcry to achieve follow-up visits, investigation, rehabilitation investigation.

Twenty-first Century: one of your status is the director of the National Key Laboratory of cognitive intelligence. What are the frontiers of research in the current laboratory?

Hu Guoping: to put it simply, a very important aspect of our research is how to make interpretations of the coordination between deep learning and artificial experts.

The basic framework of AI is now called deep learning. But for some industries, the application of artificial intelligences is still a big obstacle. For example, in the medical field, the machine tells the doctor the conclusion of the illness, but the reason for not telling the doctor is that the doctor will have great psychological pressure when he receives the results of the machine. In this regard, we have done a lot of work, that is, to give the reasoning process.

Twenty-first Century: which one do you think will be the biggest growth point?

Hu Guoping: it is inseparable from the basic necessities of life, more of these application aspects of the scene, in the direction of development. For example, AI technology can reach the level of primary doctors. Will it reach the level of two a hospital in the future, can it interact better with doctors? In the future, there will be considerable progress in the application of these fields in terms of output value. From the point of view of impact, I estimate that only about 5% is achieved.

If I want to say "the most", I think it is education. Because the number of people involved in the industry is the biggest. Everyone hopes that the efficiency of learning can be significantly improved, hoping that there will be personalized and virtualized tutors.

There is also a very important field, called intelligent hardware in the era of Internet of things. All appliances, home appliances, and artificial intelligence technology are landing, making the machine smarter, able to understand what people say and understand people's gestures and facial expressions.

AI foam effect

Twenty-first Century: in the past few years, the market is talking about the AI bubble. In 2020, do you think the bubble is still there?

Hu Guoping: in the second half of 2019, it has become difficult for AI enterprises to refinance.

When AI is relatively hot, there are many second or third tier companies crowded in, but in fact, it is unlikely that so many companies will do it simultaneously, such as face recognition, medical imaging and so on. For the head company, there is no degree of bubble.

Twenty-first Century: has it been influenced by bubbles in the market?

Hu Guoping: objectively, it will still have an impact. Including their own mentality, some of the price of AI personnel. Our basic judgement is that after the second half of last year, it is more rational, for example, the rate of increase in talent prices is slowing down.

Twenty-first Century: it has been 12 years since the launch of HKUST. What's the difference for you after going public?

Hu Guoping: in fact, the impact on the company will be larger. Inside ourselves, there is a saying that money is not the same as money. From my work status, busy degree, as well as the pursuit of technology and products, I feel that there is no effect.

Flight direction

Twenty-first Century: in 2019, the company's To C business accounted for about 36%, an increase of about 4 points over the previous year. Will it continue to increase the proportion of To C business in the future?

Hu Guoping: we have more confidence in the development of To C, and we have already seen a relatively clear development potential. But the proportion will not be determined as the goal of efforts.

One is intelligent hardware related products, we have done more scenes of the product segmentation, the full range of products rolled out, enabling users to fly related intelligent hardware single products, better recognition. We will continue to do some more user rated products, including earphones and so on.

The second is the API, which provides the invocation service of AI technology for developers. A larger advantage of the cloud calling service mode is that the cost of manufacturers using these AI technologies is very low, but because of the scale effect, the API economy itself has become a continuously growing source of revenue.

Twenty-first Century: how do we see the relationship between input and performance? Will the company's R & D input be affected by the fluctuation of its performance?

Hu Guoping: deep learning technology can be applied to all industries such as speech synthesis, speech recognition, image recognition, translation and so on. When speech recognition is done, it can be used in To C intelligent recording pen, trial conference and voice electronic medical record. Through this industry structure, input and output will be higher than that of a single technology or product company.

In the field of artificial intelligence, there are obvious "28 principles". So long as we do the head, the ratio of input to output can be guaranteed.

As for the fluctuation of performance, it may have an impact, but I think it has a unique advantage. The company has been very persistent in pursuing and investing in core technology. The company's decision makers and shareholders are very supportive. And just now said the "28 principles". So far, we are ahead of the 30%-40% and get a better return.

(author: Southern finance and Economics National two sessions report group Zhang Sainan)

?