Since General Secretary Xi Jinping proposed: "In the current external environment of rising protectionism, world economic downturn and shrinking global market, we must give full play to the advantages of the domestic mega-market, add momentum to our economic development and drive the recovery of the world economy by booming the domestic economy and opening up the domestic circulation", the "domestic circulation" has become a hot topic recently.

So, in the field of artificial intelligence industry, from what aspects of AI technology power can help to open up the domestic cycle? Under the current economic environment and social background, what kind of opportunities and challenges does AI development face? After the deep combination of industry, academia and research, in which fields can AI have better development?

At the recent Shenzhen (International) Artificial Intelligence Exhibition and Digital Economy Summit, four guests from academia and business circles, namely Dong Zhenjiang, president of Shenzhen Artificial Intelligence Industry Association, Cui Baoqiu, vice president of Xiaomi Group and chairman of Group Technology Committee, Huang Feiyue, deputy general manager of Tencent Youtu Lab, and Shen Linlin, Pengcheng scholar and director of Shenzhen University's Institute of Computational Vision, discussed the topic of "Unleashing the power of AI technology to help smooth the domestic circulation" in depth.

This forum discussion mainly revolves around the following four topics.

01 What role will AI play in the domestic circulation

02 How to view the "danger" and "opportunity" of AI development in the current economic environment

03What will be the better development of AI in China after the deeper integration of industry, academia and research?

04The problems that may be encountered in the future development of AI

Note: This forum is hosted and questioned by Mr. Dong Zhenjiang, President of Shenzhen Artificial Intelligence Industry Association.

The following are the specific contents of this roundtable forum.

1. What role will AI play in the domestic cycle

（深圳市人工智能行业协会会长董振江）

 Dong Zhenjiang: Today we are very fortunate to invite you to our site, the last roundtable of the summit today, to talk about one of the hottest topics now is the big cycle, especially the current situation, the play of internal forces may have a greater role in the domestic development.

I would like to ask the experts on several topics. First, in the domestic cycle, especially the "new infrastructure" proposed a few of the most important elements, especially 5G to do support, so artificial intelligence in the domestic cycle will play what role? I would like to ask experts to share with us, first of all, please tell us about Cui Baoqiu.

Cui Baoqiu: The concept of domestic circulation is very good, I just shared in my speech that AI is a big opportunity for China, everyone is talking about it since last year, and in the past few years everyone is talking about the world's two AI giants, one in the United States and one in China, which is not false, China has all kinds of talents, big data, government support and application scenario dividends, so AI can play a very important role in both the internal circulation and the "new infrastructure".

AI is useful in Xiaomi's representative products, especially smartphones. Today, China's smartphone market is far ahead of the world, and several major Chinese cell phone manufacturers are using AI to empower cell phones, making them more and more intelligent and ahead of their time, so the innovation in many points is already leading the world, and China's domestic demand for cell phones alone is very large, so the world is also being driven by Chinese cell phone manufacturers, which is very obvious.

IoT is an opportunity for China. China's income and consumer level have come up, and people's demand for intelligent life is getting stronger and stronger, and there are too many things AI can do in the middle.

The reason why people's demand is getting higher and higher is because of the technological progress, because of the big data dividend, because of the rapid development of AI in China, so whether it is the "new infrastructure" or the internal circulation, AI is playing an increasingly important role, so this is also very exciting for industry practitioners, we are to seize this opportunity to really let AI ignite our internal demand, let AI increase the internal circulation, which I am very optimistic.

Dong Zhenjiang: Cui just mentioned this thing, I think it is 5G came after the connection more, just said IoT actually I think it laid the foundation, with data actually on our late artificial intelligence performance will also bring great benefits, in turn to promote the application level, which is indeed a very big opportunity. Tencent should be the best in this side, especially with the most people contact, and Utu plays a very important role in Tencent, please Huang, what is the role of the current view in the large domestic cycle.

Feiyue Huang: The domestic cycle is the latest macro strategy and layout of the country, in order to adapt to a series of new international situations. In the overall strategy, science and technology itself needs to establish a set of internal circulation. In this internal cycle of science and technology, I think AI has two main roles: one is to play its own role as a core component of the new infrastructure; the other is to help traditional industries upgrade and accelerate the realization of the fourth industrial revolution, which is the industrial Internet that Tencent has been mentioning.

Unlike the first three generations of industrial revolution, industrial automation, electric power electrification, and information digitization, the core of the fourth generation of industrial revolution lies in the embodiment of intelligence. Just look at the wording, you can see that intelligence is of course very close to artificial intelligence.

Especially in recent years, with the increase in labor costs, in the past we are proud of the traditional industries, traditional manufacturing, the advantages are not so strong, in order to enhance our competitiveness, we need to enhance the traditional industries, manufacturing competitiveness, the key lies in how to arm them with new technologies, artificial intelligence is to help these industries to improve the technical strength of a very critical link.

Dong Zhenjiang: It is also the most important dividend of this change. Mr. Shen you as a scholar, I wonder how you view this issue.

（鹏城学者，深圳大学计算视觉研究所所长沈琳琳）

Shen Linlin: I started doing vision when I was a doctoral student in 2002, and at that time, this technology in the academic world would have new things come out every few years, and at that time, when these technologies may come out, the academics are very excited, and once applied to the actual may be quite different from the actual requirements. Even without the epidemic, artificial intelligence will certainly be gradually applied to various industries in these years.

This epidemic, the country's "new infrastructure" is certainly accelerating the course of +AI, especially in China. Of course the AI track is also a major competitive track between China and the United States, and is also a major track where China is likely to surpass the United States. I hope that China does artificial intelligence whether in academia or industry, can work closely together, this epidemic also brings us some warning, such as the Huawei thing. Many chips, in fact, in terms of artificial intelligence industry, many things are actually used in the United States, for example, training algorithms with NVIDIA chips GPU, of course, with the framework is open source, there are many codes on GetUP.

I think we should seize the opportunity to do artificial intelligence industry, close cooperation, the industry technology forward, development beyond the United States, this piece of stuff completely into independent, like AI this piece of the future can well create a world leading position to.

2 the current economic environment, how to view the "danger" and "opportunity" facing the development of AI

Dong Zhenjiang: As Mr. Shen said, I think the academia and industry, especially in AI is not so close, I think many problems are now proposed by industry, the academia to help solve, which is also a very good cycle, but also see the big domestic cycle. Just now you mentioned a problem, I would like to ask for your advice, after this Sino-US conflict, many technologies may be subject to certain restrictions, which indeed brings a certain crisis, but in turn, there are some opportunities inside the crisis, I would like to ask you to talk about each of you, in such a form in the end what difficulties, will bring what crisis? In turn, in response to these crises, whether it is Xiaomi, Tencent, or various industries, what kind of opportunities will be brought by such crises, and how to seize them, I would like to ask you to talk about, if you can give a few examples, or a good scene can also be depicted.

(小米集团副总裁，集团技术委员会主席崔宝秋)

Cui Baoqiu: This is a very big problem, but also many points, I will talk about one or two points that I myself more important, the current world pattern is slowly becoming a lot of people do not like, in the hope of seeing some of the direction, more and more deterioration, first of all, the Chinese government is also considering this, to solve the problem of neck, each enterprise may have to look at, if in which points encounter the problem of neck how we solve, so last year This is my personal view, China should start to lay out and pave the way for all basic technology research and upper layer application technology, so this is a big (direction).

Another point should be optimistic, should be confident, because China is very talented, Chinese people are very diligent, we are in big data, cloud, AI, of course, there are open source form, open source technology, the Internet, so we have a lot of reasons to be confident, so in today's new civilization era of the Internet, open source is the best platform and mode of human technological progress, so I believe that many neck technology Therefore, I believe that many necked technologies will not be really stuck for a long time, as long as Chinese people bury their heads in the research, they may be able to catch up soon, so I think we still need to have cultural confidence and technological confidence, especially I am very optimistic, whether it is the mobile Internet era, or the future Internet of everything, the next generation of the Internet era, China has a lot of intelligent scene innovation, application innovation, application innovation will bring a lot of basic technological innovation, so there is It is possible that soon, China may lead in many points and at different levels, which is one of my more optimistic places.

Dong Zhenjiang: What you just said, I think I have the same feeling, now should be more confident, the biggest one is indeed now is not empty to do research, more is it has really applied there. In terms of open source technology, Tencent should have a greater say here, please also talk about it.

Huang Feiyue: the current economic form, for artificial intelligence enterprises, in fact, opportunities and crises coexist, on the one hand, due to the domestic economic cycle, including the entire international pattern changes, and the impact of the epidemic, we can also notice that in the new context, may be the original offline communication will have a more obvious impact, through the help of digital intelligence technology, through a series of products and solutions, can be very good to help life, work, communication, and even found more efficient than the original.

In this case, we are more willing to accelerate the embrace of new technologies, thereby creating more new opportunities, artificial intelligence as a new technology applications will have more opportunities, opportunities correspond to challenges, we tend to imagine artificial intelligence omnipotent, in the actual application, we will find that many new technologies such as artificial intelligence is actually lower than the application of business expectations, this time there will be a relatively large gap. In the face of such a situation, what we have to do is to seize such opportunities, bury our heads in the technology to do a good job, to put the technology into practice, in the process of such repeated refinement, until the technology application really mature and scale, the enterprise really for such practical application of value recognition, AI will become more universal, which I think is a very good adaptation process.

Shen Linlin: The opportunity for me is the direct feeling is that the original like Huawei, before may find a lot of U.S. universities to cooperate, now because of this matter is not good to find the U.S. universities, began to turn to many domestic universities to cooperate, so Huawei and Shen University also have cooperation, including their president visited us some time ago, this is the biggest opportunity, now the domestic big enterprises turn to domestic universities to cooperate, they may also have more funds to invest in domestic research this, for the academic community this is a very good thing.

It may be the same for other industries, some times it is to go abroad to buy these things, may be the domestic made things than foreign some poor, but also can have the opportunity to become these large companies suppliers, I think this is a very good example.

There are also "new infrastructure" piece, for example, like this period of time, the power industry and we talk about cooperation, like intelligent inspection, which may also be due to the importance of the state invested a lot of funds, they now have the opportunity or funds to do intelligent piece, and like infrastructure here, there are doing building intelligence, information technology companies, also want to find us to cooperate I think this is also the state's attention to do "new infrastructure" investment, many things on the site were originally manual, for example, to see whether the site is covered, whether the helmet, which may be used more, and dumper supervision, these can be used to artificial intelligence technology, all these for the domestic Whether it is the academic community or artificial intelligence companies are a great opportunity.

Dong Zhenjiang: Mr. Shen said this industry-academia research thing, I remember that every year Tencent will send a plan called "rhinoceros bird", just like we send the subject every year, I wonder if there are new changes in the situation this year, especially on cooperation. As Mr. Shen said, a lot of cooperation with foreign countries, may slowly subject to greater restrictions, it provides a very good opportunity for universities, this year, the two enterprises in this area have no new plans or ideas. Mr. Huang, do you have any new cooperation plans from Utu?

(腾讯优图实验室副总经理黄飞跃)

Huang Feiyue: We have a lot of cooperation with many famous domestic universities and institutions. Since 2012, we have been cooperating with Shanghai Jiaotong University, Tsinghua Institute of Automation, Chinese Academy of Sciences Software Institute, Xiamen University, Zhejiang University and other famous universities, including the "Hornbill" project, every year we will initiate or participate in some key direction of the subject, gathering more academic experts to overcome the problem together. We have also achieved very good results, and our partner universities have exported many excellent students to Utu, and many of them have worked in Utu and become our excellent backbone.

On the other hand, we have a lot of cooperation with the industry. On the one hand, Tencent Youtu provides an AI technology open platform to the public, which will deliver a series of visual capabilities such as face detection and recognition, alignment, and OCR to small and medium-sized enterprises for their use.

On the other hand, we also work closely with Tencent AI Accelerator and Industry Accelerator, providing a series of assistance such as capital, mentors, courses, etc., and communicating with the gas pedal finalists on a regular basis. Overall, there is more cooperation from academia to industry, and today we have launched the Tencentlight Club, which also hopes to gather more young scientists to jointly promote the construction and development of science and technology.

Shen Linlin: As you know, Ma Huateng is our alumni, so Tencent and our cooperation with SZU, before the epidemic has been more, like the "hornbill" project, Tencent has a special independent project in SZU, and CCF is independent, we have our own "hornbill "project, which is thanks to Tencent, there is also cooperation with the Tencent Artificial Intelligence Academy, which cultivates artificial intelligence talents, that is, since the first half of last year, Tencent and we jointly develop artificial intelligence courses for undergraduates, which is a close cooperation in the epidemic. Now there is also cooperation with Huawei.

Dong Zhenjiang: Mr. Cui side there is no specific work related to this aspect of cooperation with universities.

Cui Baoqiu: there are many, first of all, the general environment, I joined Xiaomi more than 8 years ago, after joining Xiaomi has been constantly promoting Xiaomi and university cooperation, when Xiaomi is still relatively young, early business run too fast, cooperation is not deep enough, last year Xiaomi Group established the Group Technology Committee, technical cooperation is one of the five responsibilities, very important, technical cooperation includes three aspects, cooperation with universities, joint ventures with academia, cooperation with enterprises, the third includes cooperation with the Chinese government, so the three major directions of cooperation, this is the focus of my work this year.

Mr. Huang also mentioned that in the AI era, academia and the business community should be more deeply integrated, and today many innovations in the field of AI often occur in enterprises because they have application scenarios, products, data, computing power, and many engineers, so many university teachers especially like to cooperate with Xiaomi.

In the past few years, we and Tsinghua University, Peking University, Harbin University, Wuhan University and Sun Yat-sen University and many other teachers have maintained some close cooperation and communication, this year is promoting cooperation with many universities, Xiaomi also took out a lot of money some time ago, set up a number of Xiaomi scholarships, but also to strengthen cooperation with universities from the other side, to strengthen talent training. Whether it is artificial intelligence, big data or cloud computing, IOT, hardware, radio, or materials, etc., these communication technologies are particularly optimistic about the direction of Xiaomi, we will deepen cooperation with universities from time to time, which is a big task for me this year.

In addition, in the past few years, on behalf of Xiaomi, I have been involved in some of the work of the Computer Society, also to promote exchanges between enterprises and academia, this year I was recommended as the director of the CCF Enterprise Working Committee, is planning a lot of CCF front-line technology seminars, held from time to time, also to deepen the cooperation between universities and enterprises, teachers, students and enterprises to communicate more, but also in promoting a CTO club, so that China all enterprise CTOs can cohesion, to the world, to the outside of the Chinese enterprises on technology to science and technology of some words, or judgment, or direction pointers, analysis of views, etc., this is also on behalf of Xiaomi in promoting the integration of industry-university-research efforts, at this stage is very important, in the AI era is also very important.

3. After the deeper integration of industry, academia and research, in which areas of domestic AI will have better development

Dong Zhenjiang: Because this is also a very good integration, especially now that enterprises are facing this situation will encounter many specific problems later, there are also more opportunities to communicate with schools, I believe there will be more and more. With this kind of cooperation, there is an opportunity for internal circulation, I wonder which direction you teachers are optimistic about? Is there any scenario that you think is very meaningful and worth investing in? I wonder if there is a typical scenario from the enterprise? Mr. Huang also mentioned that under the new crown, a lot of work and life patterns have changed, in this case, in which areas do you think there are very good opportunities, or in which directions the company is engaged in, can you briefly talk about them? Mr. Shen, you look from the periphery to the inside, where do you think the opportunities will bring to our business, and from which aspects we may develop favorably.

Shen Linlin: First of all, I do education, I think education is a very good direction, during the epidemic, a lot of education is online, are placed online, online education process, AI education itself is a few directions, AI is getting bigger and bigger, the demand for talent is certainly also getting bigger, the AI education course itself, or education itself is a direction. Of course, there are multiple levels of education, such as high-level, medium-level, originally AI talent is a number of levels, from algorithm talent, to engineers or training engineers, the Ministry of Human Resources and Social Security this year launched a new occupation called artificial intelligence trainer, may be for professional educators, I think different levels of AI education itself is a worthy direction of development.

There is also the process of online education, may have generated a lot of data, different education courses, learning or, have generated a lot of data, how to use AI algorithms from the data to dig something, to promote personalized education or more good content manufacturing, in my opinion this is a very good direction in itself.

Dong Zhenjiang: Mr. Huang, you talk about how Tencent sees this thing.

Huang Feiyue: From the perspective of Youtu, I tend to look at two dimensions, one is in the AI era, which industries can have more widely used talents and greater value, this industry level indeed I look more like AI education, and like finance, medicine, industrial manufacturing, how can the traditional manufacturing industry play a greater value with AI, these areas can generate huge value.

Another dimension to look at is more from the technical dimension, this time AI to see how the future valuable direction, my own more concerned about the future trend of 5G development, with the development trend of 5G, the future of digital content is certainly more and more, so content generation, how to use AI to help produce more valuable content is a very good research direction.

In addition, with the future development of digital content, on the one hand, more and more content, how to ensure the authenticity of the content, on the other hand, how to ensure the safety of the content, for example, for the protection of children and youth content, is also something we need to think about and as.

Dong Zhenjiang: Another one, I recently saw the country mention "new infrastructure", I think a more promising development is the direction of the Internet of Cars, I see Tencent in this direction has also made efforts, I do not know how to promote now.

Huang Feiyue: Telematics is a trend, regardless of driverless, Telematics is certainly very valuable in the future, from a technical point of view is not particularly mature, the future requires greater investment and research and development.

Dong Zhenjiang: At present, we are in the "new infrastructure", and industrial Internet, I think Utu itself is doing very well in vision, will there be to industrial vision to go, there is no such consideration.

Huang Feiyue: In the industrial field, we already have very many applications and programs, for example, our industrial quality inspection program with huaxing photoelectric, based on self-researched vision AI algorithm, Utu joint Tencent cloud to create the AI automatic defect classification system, identification of a single picture only takes 500-600 milliseconds, compared with the manual diagnosis of the film, AI recognition speed increased by 5-10 times, can save huaxing photoelectric more than ten million yuan of costs each year.

Dong Zhenjiang: Okay, thanks. Mr. Cui, from Xiaomi's point of view, which scenes do you think will be easier to catch in the future? Or do you think this thing is worth further development?

Cui Baoqiu: I personally think there are a lot of opportunities, very excited, but back to the main business of Xiaomi, from the perspective of Xiaomi, as I said in the PPT before, cell phones multiplied by AIoT, which are two very thick main line, two core. For cell phones, I think there are two directions where AI can be used, the first is AI camera, a few years ago Mr. Lei said this sentence, he said the battle of cell phones is the battle of the camera, the battle of the camera is the battle of AI, we all use beauty, a variety of AI to improve image quality, today we can all imagine that the phone becomes a camera, AI in the camera can play too many roles, the latest version of the Xiaomi 10 The latest version of the Xiaomi 10, its photo function is the highest grade in the world, the partners score 130 points, which is why we are concerned about this, the AI team, especially the vision team, has made a lot of contributions to the AI camera. AI in the image processing, or photos and video, can do a lot of things.

The other is Xiao Ai classmate, Xiao Ai classmate as an intelligent assistant, natural language processing, semantic recognition, can really understand the user, predict user behavior, can provide users with attentive, personalized, emotional, more close to the human service, which is also a point of infinite investment in technology, which is also the direction of Xiaomi in the strength.

So, around data, a large number of AI areas are worth investing in, AI capabilities, AI resources are worth investing in. In addition AIoT more intelligent devices, intelligent scene innovation, a variety of combinations from cameras to security, to education, to emotional care and some of the actual home environment improvement, very much. So from Xiaomi's point of view around these two core, cell phones and AIoT, this is not only an opportunity for Xiaomi, but also for Chinese companies.

4. About the possible difficulties encountered in the future development of AI

Dong Zhenjiang: Please think about a problem and give us a suggestion. Now from the academic world, from the company's point of view, where do you think the biggest difficulty is when AI is developing forward? How should we solve it.

Shen Linlin: When AI slowly began to enter the traditional industries, including electricity, manufacturing, and a variety of traditional industries, now become + AI, from my point of view, I found that the biggest difficulty in the process of communication with these companies is the communication problem, because they actually do not know AI, of course, may know some skin, they do not know AI system from data collection to algorithm training, to the final deployment of the whole process, what are the difficulties. This time their enterprise may come to say I want to do this thing, can you do the intelligent identification system within a month, many times encountered this problem, to be very patient and he explained the whole process is like, encountered the first problem is whether there is data, usually he and you say is I have a lot of data, ask where is this data, data is not organized This is the first problem, he said there is a lot of data, and you understand that a lot of data is two different things, this is a cross-industry cooperation and communication problems.

This is a cross-industry cooperation and communication problem. Including my cooperation with hospitals, the same and doctors say the generation gap is even bigger, doctors are saying with you that there is a lot of data, the data is scattered in various places, and there is absolutely no organization, this is a very big problem, these years feel slowly become better, especially Shenzhen enterprises, Shenzhen enterprises are often willing to invest some time and costs, they rely on technological innovation to develop up, so they are willing to do Such things. But when communicating with some enterprises, they are not willing to invest in this thing, including data collection this step they are not willing to do, he wants you to do all the systems at once, for me the specific feeling is the problem of cross-industry communication, and whether enterprises are willing to invest, spend time with you to do up AI, application up, I think this thing needs to be invested by both sides, time cost or financial resources I hope that after a few years of understanding this industry and this technology, enterprises will slowly reach a consensus with us, so that AI will be applied more and more successfully in various industries.

Dong Zhenjiang: Yes, it is a process, now the better is that each enterprise has realized the importance of AI, to integrate into it, we need to understand the needs of the enterprise, the enterprise reverse also need to know where the development of technology to go, the combination will be better.

Shen Linlin: enterprises know that AI to use up, but they lack systematic understanding of the AI process, and many companies are not willing to invest in this thing, I think there is no way to do this, including data collection, the first thing to do this thing, first the data collection done, first the platform built, before you can train, before you can do this, the first step are not willing to invest in doing, the latter can not be talked about, because now deep into various industries, each industry data are different, it is impossible to buy a universal placed inside, may do a similar can, but the speed, performance continues to improve, there is also the process of iteration.

Cui Baoqiu: there are several challenges, may not be the same dimensions, I will give two or three examples.

Challenge one: information security and privacy protection.

Today's AI is big data AI, it relies on a lot of big data, so how much data there is how much intelligence, in order to improve AI capabilities today, people need a lot of data, a lot of data is related to people, how to protect user privacy in this case, when a company has tens of millions, hundreds of millions of user data, is a big responsibility, information protection, information security, these are a challenge, so Xiaomi has done a lot of work in this area, to be responsible for the user, to respect user privacy, under these premises in order to tap the value of big data to provide users with attentive service, this is the past eight years Xiaomi to promote two things, one is data science, data intelligence. There is another line is the protection of privacy, these two are actually some of the conflicting, how to grasp the balance between these two lines, no matter what, user privacy is what we need to adhere to, on top of this principle to provide intelligent capabilities, play the value of big data, which is the challenge facing the industry.

Challenge two: about AI ethics.

Based on today's big data AI, made some judgments, AI reasoning, the conclusion is right or wrong, in many cases is not explainable, no explanation, for example, it gives a person to design a career plan, then it is easy it will say that women are not suitable to do science and technology, men are suitable for science and technology, or it will say that women are not suitable for management, not suitable for CEO, or from rural areas, or low-income families better not to study in college, similar to this is I think there may be some kind of discrimination, the big data deception, which is very much a cause for concern.

For example, after the emergence of AI technology, "seeing is believing" this phrase people no longer dare to speak, because AI capabilities can be used by the bad guys, you can change the face of some online bad video to fake, so seeing is believing, hearing is not, in the future, hearing, seeing are not necessarily to believe, so this is another AI ethics Of course, there are two sides to any technology, the use of the bad guys is one side, AI ethics I think it is worth the whole industry to pay attention to. So the challenge is still very big.

Challenge 3: AI for good.

Today too many companies are doing AI to make money, to get users to do AI, AI for good is whether AI can be used to solve the problem of poverty, avoid war, solve the problem of water, basic health care problems, etc., what role can AI play during the epidemic.

Personal data, information security privacy protection, AI ethics, AI for good. I think these are some of the challenges that AI faces.

Dong Zhenjiang: Mr. Cui spoke very well, because I think a technology to long-term, stable forward development, these are the fundamental issues, including the good, including ethics, perhaps from the legal level, including ethics, may be a technical level, a legal level, and from the moral self-restraint level, may have to consider these issues, otherwise technology simply go on is no direction.

Once again, thank you Mr. Shen, Mr. Cui and Mr. Huang for the wonderful sharing, thank you.