Talent is seriously lacking. How can AI education break the situation?

This year's college entrance examination, a composition question that is made in "Dream of Red Mansions", is difficult to stump a large number of candidates.

If it is said that "Dream of Red Mansions" is a problem that "the past" gives candidates. So with the continued scores of the college entrance examination in these days, candidates and parents will encounter confusion from the "future" -to report volunteers.

Who does not want to choose a major that can adapt to the general trend, embrace the rising industry, and even open the atmosphere first?

This choice is not only about personal. Today, with the evolution of international games into scientific and technological disputes, China wants to break through the technical blockade and stuck neck, which is increasingly required for a group of professional talents in key areas to innovate and overcome difficulties.

Speaking of which, what majors do you think of?

01

Yanyuan's unnamed lake, the reflection is Boya Tower, and the dream of countless students.

Opening the professional setting of Peking University, Peking University, which has been known for its literature and sciences in history, now has a graduate department of information and engineering science.

In the past, computer majors were derived from many institutions such as information technology, electronics, computers, integrated circuits, intelligence, microelectronics, software, future technology. From a professional expansion, it is not difficult to see the active adjustment of this century -old institution in the face of the future.

In Peking University, there is an elective course set up by the School of Information Science and Technology, which has become a popular choice for young geeks.

This course is called "Algorithm in the Game AI", and after -class operations are experimented with the environment of King Glory 1V1.

As soon as the king started, the "war fire" rotated to the old opponent next door -Tsinghua.

Chen Huayu is a graduate student in the Tsinghua Computer Department. He joined the "Five Killing Cai Wenji Team" and learned in his professionalism. It took 5 months to invest in the "Glory of the King" and won the championship in the 2nd "Tencent Enlightenment Multi -Intelligence Study Contest".

However, it was not people who played the game, but the AI ​​they trained.

The opponents are also strong, including Peking University, the Academy of Sciences, the University of Science and Technology of China, the University of Science and Technology of China, the National University of Science and Technology, the University of Science and Technology, Harbin Institute of Technology, the University of Electronic University, Electronic University ... Xueba PK, the master's tricks, the sloppy.

Isn't these famous universities we want to dream? Why are the famous school schools "playing games"? Is there any misunderstanding in it?

No misunderstanding! Pull the camera to the other side of the ocean.

When Chinese students were staring at the famous American school business school, American students had already made different choices.

In 2019, the number of people in American mathematics and computer science increased by 9.4%, exceeding business, becoming the second popular major in the United States.

The hottest discipline in computer science is artificial intelligence, which is a well -known AI.

Schools that open AI majors in the United States include Harvard, MIT, Yale, Stanford, Berkeley, Kanegi Mellon, Benna, Corda, CIT, GIT, UCLA ... specialized.

AI professional Heat is the first driving force for Silicon Valley.

GlobalData released a report on the world -renowned data analysis company:

From 2016-2020, Apple acquired 25 artificial intelligence companies. It is rare that Cook, who has always been steady, can't sit still, and has a "re -created an apple" handwriting.

Musk was not idle, and launched the "Superman Program" to establish Neuralink Company, bluntly: To control AI, first coexist with AI.

Google supports the artificial intelligence laboratory DeepMind, which has successfully controlled an overheated substance in a nuclear fusion reactor.

Do you think DeepMind is a bit familiar?

That's right, in 2016, it was the AlphaGo built by DeepMind, defeating the top player of human beings in one fell swoop.

AlphaGo is alarming, and more and more AI research teams are aware of the research and development of intelligent technology. In terms of massive data, complex scenarios, model iterations, etc., too many resources and investment are required.

Therefore, the game has become a trial field for AI.

DeepMind himself, even to study AI, created a game called "X LAND".

X LAND is more advanced in that it can establish a 3D first -person incarnation in the multiplayer environment. This incarnation will perceive the game environment through the first person's perspective, and constantly try and collide the rules of the game. Train yourself to complete the game goals.

Of course, don't just look at Silicon Valley's AI research leading, China is also catching up.

Last year, the Bell Five Science of Kennedy College of Harvard University and the International Affairs Center "The 21st Century Sino -US Science and Technology Competition" forecast:

China has significantly surpassed the United States in the actual application of artificial intelligence; by 2030, China is expected to surpass the United States and become a global leader in the field of artificial intelligence. The Allen Institute of Artificial Intelligence in the United States also said that at the level of scientific research, China's patent publications in deep learning are now 6 times that of the United States.

Since overseas technology companies can train AI through game training, of course, China can also.

Seeing this, you can answer the questions the previous: Why do colleges and universities are playing games -the truth is that they use the glory of the king to improve their artificial intelligence research topics.

Moreover, the glory of the king today has also been in the field of AI.

The 2022 World Artificial Intelligence Conference, a very special king battle show was held. Five e -players played against AI. Just like the world's top Go player in the world of AlphaGo, but this time AI was not from Google, but the strategic collaboration of King Glory and Tencent AI LAB- "King's Uncle".

If the player sparring, game environment balance regulation, hero numerical testing such as "king's unknown", and new gameplay such as "challenging" are the secrets of improving the player's game experience.

Then, "enlightenment" is Tencent's scientific research, education, and industrial cooperation dimension outside the game, creating an open platform for AI scientific researchers and enthusiasts.

Taking the enlightenment AI open research platform as the technical carrier, in addition to hosting the open AI competition in the Chengyu area to help local talent training and academic exchanges, it is necessary to have previously with Peking University, Electronic Science and Technology University, Tsinghua University, China Science and Technology, and Chinese Science and Technology. Based on the cooperation of teachers of 4 universities in universities, the scope of cooperation will be further expanded, and teachers from 10 universities including Fudan and Nanda will be invited to join the science and education consortium to jointly build AI universities training plans.

The glory of the king has done so much. A large part of the reason is that the development of AI in China, especially in the cultivation of technical talents, there is a need for urgent and pragmatic needs.

02

AI is a practical science. After all, technology will leave the laboratory and move towards the industrial scene.

In the process, it is necessary to join the excellent talents. And these AI talents need a comprehensive education system.

Don't always think that there are many Chinese people, you can always discover talents. In fact, my country's AI talent gap is still very large.

According to the "Analysis Report of the Employment Situation of Artificial Intelligence Engineering and Technical Persons' Employment" according to the three departments including the Ministry of Human Resources and Social Security ":

my country's AI talent gap exceeds 5 million, and the supply and demand ratio of 1: 10 is seriously unbalanced. If it is not trained in time, the AI ​​talent gap will exceed 10 million in 2025.

China has realized the importance of AI education and full support for policy. The AI ​​class of Peking University's AI has also explained that students are interested in the AI ​​field.

However, AI research and testing is not easy.

If you are interested, there is no way to let an AI model from clumsy to "smart".

◆ The first difficulty of AI education is the problem of computing power resources.

Simple understanding is the computer's data calculation and processing capabilities. Peking University Ph.D. Lu Yunlong introduced: At Peking University, the school's GPU can only handle general computer vision research and simple games. But if you want to train AlphaGo or AI who can hit the king, efficiency will be much slower.

Even if the calculation is up. How to evaluate whether your AI model is smart enough, can be like AlphaGo, even "smart" than it?

◆ This is the second difficulty facing AI education today -the test environment.

Studying the "Go" used by AlphaGo is a good choice. However, today's artificial intelligence research is not the same as before.

The entire state space of chess games is limited, but the industrial application scenarios in the real world are not so simple, and there are many limitations of information acquisition.

Therefore, it may be good to use Go to test AI's decision -making ability; but as mentioned above, it is definitely not enough to let technology go to the industrial scenario.

◆ Let AI from theory to practice is precisely the third difficulty.

Chinese colleges and universities usually take theoretical knowledge as the system and curriculum learning. This form is unavoidable to disconnect at the moment when the science and technology revolution is changing with a variety of days and industrial upgrades.

The Chinese physical industry chain is very complicated, and the corresponding application scenarios are extremely complicated. It is difficult for college talent training to carry out flexible predictions and adjustments based on such complicated reality whether it is curriculum settings, results research, and scene applications.

Outside the school, the industries and industries that should have been promoted by AI to promote transformation and upgrading may lose the opportunity of transformation because they are not driven by the results of the results of the AI ​​research of colleges and universities.

In a word, the development of AI requires the innovation of the entire chain of industry -university and research.

However, since education is the source of all talents, breaking through this layer of ceiling still needs to be countermeasure from education itself.

Based on the enlightenment platform, the "curriculum+competition+scientific research" symbiotic development model given by King Glory and AI LAB may be a new solution.

03

Looking up at the starry sky, the sea above the head.

From the early blue, dark blue, deeper blue, to the 16 -year -old ALPHAGO, foreign DeepMind and Open AI went forward. The United States once had a good scenery. In China, the "King of the King" developed by King Glory and Tencent AI LAB also proves that China is also catching up in the field of artificial intelligence.

From the perspective of AI research and education needs, the multiplayer competitive strategy game such as King Glory solves the problem of lack of AI scientific research scene -the possibility of 20,000 squares in the game is comparable to the complexity of the real environment. Technology cultivation dish.

Moreover, if the AI ​​model can cope with the incomplete information environment in the gaming environment and be able to feedback and make the optimal decision in time, then it naturally has the opportunity to migrate to the real environment and exert larger application value.

At this step, it is not easy. But then the appearance of the AI ​​open research platform "enlightened", everyone understood that there was a larger blueprint. Relying on the supporting AI provided by the enlightenment platform for the large -scale computing power of strengthening training and a unified enhanced learning framework, it can help teachers and students of colleges and students to accelerate research and development and improve AI training efficiency.

Through the cutting -edge simulation environment, open and diverse technical precipitation assets, teaching and verification support provided by enlightenment, it can make up for the lack of application scenarios and computing power resources from multiple dimensions.

Zheng Dongrun, a graduate student of Harbin Institute of Technology (Shenzhen), requires enough GPUs to train AI before, not only consume resources, but also build a system of training AI, which is also very troublesome. Now the enlightenment directly gives Zheng Dongrun visualized interface, and it is OK to throw what you do.

The enlightenment project team member James compares it to the hardcover room where you can check in at any time.

In addition, based on the enlightenment platform, Tencent also organized two "enlightenment multi -intelligent physical enhancement learning contests" to help teachers and students from colleges and students in a lower threshold in the way of "in the stage of the competition". The accumulation of technology on the top to improve its own scientific research topics.

Chen Huayu of Tsinghua University said that during the rematch of the "Tencent Enlightenment Multi -Intelligent Standard Learning Contest" this year, the AI ​​of training the competition was used to use 32 servers, 1,000 CPUs, and 1 large GPU. He frankly could not provide such computing power resources based on individual or school laboratory.

Zhao Jian, a doctor of Chinese University of Science and Technology, bluntly stated that with this platform, the scientific enthusiasm of laboratory students is particularly high.

In addition, as mentioned above, the glory of the king is still based on enlightenment and establishes college curriculum cooperation plans with 14 universities. Let AI majors complete the construction of a more solid theoretical knowledge system in the classroom.

Since the Ministry of Education issued a document in 2019, 35 colleges and universities were replied as the first batch of new professional construction qualifications of "artificial intelligence", and the artificial intelligence curriculum education of colleges and universities has been developing.

◆ On the one hand, in just three years, many schools have no systematic systematic artificial intelligence education teachers. Most teachers have relevant experience in a certain field.

In addition, high -tech talents are sought after by related industries at home and abroad, high -quality talent resources are working in enterprises, and colleges and universities have restricted talent progress.

◆ On the other hand, aiming at policy dividends and air outlets, some colleges that have not obtained AI professional rights to play "wipe the ball", which blurred the professional choices of students through text games, which also affected the training of talents to a certain extent. For example, the so -called so -called so -called so -called so -called so -called so -called so -called so -called The "robotics professional" has nothing to do with the knowledge of AI in strengthening learning, and more focused on mechanical manufacturing and circuit engineering.

Fortunately, Chinese universities have now launched a multi -level, "AI+X" talent training model for different people. This model can not only be taught and cultivated according to their aptitude, but more importantly, it is conducive to tapping professional composite talents. The "AI+Game" based on the extension of the enlightenment platform is a choice, which is one of the reasons why enlightenment can build a cooperative relationship with major universities.

The enlightenment platform carries "AI ecological raw materials such as technology and resources" and gradually develops into an ecological gathering place for "competition-curriculum-science and education alliance-science and technology consortium".

On the way to the industry to the industry, Tencent released the cooperation of "Tencent X Southwest Jiaotong University Smart Transportation Project", which will cooperate with the professional field of Southwest Jiaotong University to rely on the "enlightenment" platform and use Chengdu as the core demonstration area. The problem of traffic congestion that plagues everyone is committed to increasing the cross -section efficiency by 10%.

This kind of school -enterprise linkage can use corporate resources and channels to promote research on the one hand, to precipitate high -quality cases for subsequent AI education, and accelerate the reserves of talents and technology. Pusher.

Moreover, smart transportation is an important part of a smart city.

When the "daily problem" of traffic congestion is optimized and solved, the next step may be to improve the construction of the city's "traffic brain". After all, Chengdu itself is planned at the level of smart transportation -not only launched the "smart parking" service module last year, but also established a joint innovation center of smart transportation this year. Therefore, the smart transportation projects created by the cooperation and universities relying on the enlightenment platform are becoming one of the puzzles that complement Chengdu's "smart brain".

At this point, although the barriers between AI industry -university -research are not fully abundant, at least find a new starting point that connects with each other, providing a new reference path for more talents, technology, and industries in the future.

In early June, Tencent also signed a cooperation agreement with Chengdu High -tech Zone to land in the Tencent Future Center project in Chengdu High -tech Zone, and explore future science and technology and scene forms such as Quanzhen Internet and other key areas such as culture and innovation, science and technology innovation. Digital transformation and development of urban governance.

At this stage, the "GM AI" approaching the human brain may only belong to science fiction plots.

But there is always a beginning. At the end of last century, some universities in the United States opened the "game majors", which was not only game design, programming, etc., and even "playing" became a separate major.

Everything must be the beginning of the future. In this sense, the king's realization and enlightenment have opened a good start for China's AI education docking in the future.

The word "enlightenment" is very important.If China has perceived the ashes of industrial civilization in the past 100 years, you can also touch the stones of others all the way, but in the new technology wave, there is no ready -made stone to touch.We can only rely on ourselves to find and climb the stairs of human overall evolution.The test field of King Glory is a good start.

- END -