Learn layout, assist in training Go, and walk with artificial intelligence.

  Core reading

  Since the "man-machine war" in 2016, the influence of artificial intelligence on the world of Go is unprecedented. Breaking stereotypes, studying layout, interpreting trends, and assisting training … … Artificial intelligence is becoming more and more involved in the field of Go. Players’ chess skills have been generally improved, training teachers have been reinforced by the situation, and the promotion of running games has gone to the cloud, and the explanation of watching games is more intuitive … … With the extensive application of new technologies, Go and artificial intelligence have joined hands to create a new situation.

  On September 1st, 2020, the first stage of China Men’s Go League was over. In eight rounds, no player or team remained unbeaten. The pattern is no longer strong and weak, and the competition is more suspenseful. Behind this is the overall improvement of the players’ strength. "It is a good phenomenon that there is no victory. The game is very intense and the level is very close." "Chess Saint" Nie Weiping commented after the game.

  This change is not unrelated to the help of Go artificial intelligence (AI) in recent years. Breaking stereotypes, studying layout, interpreting trends, and assisting training … … With the increasing participation of artificial intelligence in the field of Go, the human brain and computer are constantly learning from each other. According to industry insiders, compared with before artificial intelligence was integrated into the field of Go, the strength of professional players generally improved a lot.

  Artificial intelligence is widely used in auxiliary training.

  In the seventh round of encirclement, the two players who ranked first in Go in China and South Korea lost at the same time. China’s "seven-time champion" Ke Jie’s 15-game winning streak came to an abrupt end; South Korea’s "post-00" world champion Shen Zhenluan’s five-game winning streak has also come to an end. Just a few days ago, these two players just set a new record on the Goratings World Go Rating List — — Shen Zhen’s score reached an unprecedented 3,800, and Ke Jie’s score of 3,723 also set a new personal high.

  Since the man-machine war between Li Shishi and Alpha Go in 2016, the impact of artificial intelligence on Go is unprecedented. The computer’s precise calculation, learning ability and layout are jaw-dropping. Today’s Go AI software can even make the two sons compete with the top human masters. "Deep powerlessness" once pervaded the professional chess world.

  It has long been a compulsory course for professional players to use AI as technical guidance and study different "problem-solving ideas". China Weiqi Team and Tencent Intelligent Weiqi "Jueyi" renewed their contract until 2022. By providing auxiliary functions such as chess splitting, winning rate, coincidence degree and situation analysis, the special training AI has assumed multiple roles such as coach, sparring partner and chess friend. Before the competition, chess players often use AI to make targeted preparations.

  The "post-90 s" players were the first to grow up through "online chess", and now it’s time to come to the new node of AI. The equalization of training resources has gradually narrowed the gap between new and old players. Young people who are new to professional chess are not short of opportunities to compete with masters, and the growth rate of "post-00" players is further accelerated. "Now every game of chess has become bad. Sitting in front of the board, everyone has similar things. The outcome depends on improvisation and adjustment." China chess player Chen Yaoye said.

  Under the trend that chess players generally benefit from artificial intelligence, who can learn more nutrients is more likely to stand out. After Ke Jie’s defeat in the "man-machine war" in Wuzhen, he reconsidered Weiqi, and his core abilities such as overall view and judgment went up a step. As one of the fastest players in South Korea to accept AI, Shen Zhenqi has made a qualitative leap after learning thousands of games with AI a year and improving the layout shortcomings.

  AI and professional Go are inseparable. With the help of this springboard, in the past, players’ collective chess moves turned more to collective and AI. "Everyone’s collective research will help to digest the ideas of AI faster, which in turn will help more people accept new changes." Hua Xueming, the leader of China Weiqi team, said.

  Chess players’ understanding of Go is constantly updated and expanded.

  Artificial intelligence is regarded as a great change that has never happened before in the world of Go. From the initial dissatisfaction and shock, to the frustration and confusion after the defeat, and then to the research and consultation after in-depth understanding, most players have experienced complex emotional changes. Ke Jie once lamented: "Watching AI play chess is like a fairy fight. We can only appreciate and learn."

  Fujizawa Hedeyuki, a Japanese Weiqi master, once said that there are a hundred chess skills, but I only know seven. When the wave of artificial intelligence hit, everyone realized more and more that this was not modesty. The previous formula of strokes is often overturned under the thinking of AI, and the players’ understanding of Go is constantly updated and expanded, and more possibilities are found on the chessboard. "For example, many strokes of AI are considered according to the overall situation, and there is no fixed stroke in a local situation. This is a chess improvement." China chess player Shi Yue said.

  From opponents to friends, professional chess players and AI find a longer-term way to get along. Ke Jie often analyzes his judgment and understanding through AI, and the data feedback of AI also helps him to absorb and digest it selectively. "I even feel a little dependent on AI, which is something I didn’t dare to think about a few years ago."

  In the past years, chess players have often been puzzled by where to win or lose a game of chess. The accurate analysis of AI can help to get close to the correct answer and make people take fewer detours. At the same time, as the mystery of AI is slowly unveiled, players are increasingly discovering that there is no absolute positive solution to Go, and it is impossible for computers to analyze all the changes. In the end, the competition is still different understandings of chess.

  When the smoke of AI attacking the "last piece of wisdom highland" of mankind is exhausted, the technological changes and the impact of ideas have not damaged the charm of Go, but have brought opportunities to promote the project. After all, the game between people and the emotional and psychological wits of both sides are still beyond AI’s satisfaction at this stage. In other words, artificial intelligence just allows us to treat Go in a smarter way.

  New technology is integrated into all links of Go.

  Before the start of men’s armor this season, the Chinese Go Association held the first online warm-up match in history. With the help of science and technology, the face-to-face "hand talk" of AI practicing chess and "cloud" training, which has been inherited for thousands of years, presents a different look.

  What AI has changed is not only the professional masters in the stadium. In the past, some ordinary chess fans didn’t understand the complicated calculation on the chessboard. Now, with the artificial intelligence explanation and auxiliary game watching system, even "amateurs" can see the situation and win or lose at a glance through the data and judgment fed back by AI, and watch the game more intuitively.

  For a long time after "Alpha Go" was born, AI Go software was only a scientific research project of big companies. Today, AI is already a tool that ordinary chess fans can control. Among all kinds of game software developed by the people, there are not only publicly sold software, but also many open source software that anyone can download and use, and even some "technical stream" chess fans personally adapt the software.

  The teacher problem that once plagued the Weiqi training industry has also been alleviated by using AI as a teaching tool. The "artificial intelligence characteristic curriculum system" that appeared in the past two years has fully stimulated the interest of young people in learning chess and can improve their chess skills in a relatively short period of time. Competition, study, training, training, research and explanation … … The shadow of artificial intelligence has appeared in every link of Go, and it has become a tool to "stimulate human potential".

  Of course, the infiltration of artificial intelligence also brings some "side effects". This year, affected by the epidemic, a large number of Go games were transferred to online. The readily available game software, the game environment that can’t be supervised in real time, and the AI tricks that are difficult to define clearly make the threshold for online cheating lower and lower. Insiders admit that although there are some technical firewalls, there are still dead ends and loopholes, and some "suspected cheating" can’t catch the "real hammer" in time. China Weiqi Association revealed that in the future, anti-AI cheating measures will be improved and the environment will be purified in official competitions through information blocking, space monitoring, using traces, coincidence analysis and dishonesty punishment.

  Even though artificial intelligence is a "double-edged sword", the fallen chess pieces are ultimately in the hands of human beings. As long as the new technology is "controlled and served by people", there is no reason not to try to learn and apply it. Walking with artificial intelligence, the profound world of Go has not been eclipsed, but it has opened up a new situation and has a better chance to bloom brilliantly.

Stop production, close factories, OEM … How to break the dilemma of automobile overcapacity?

  Recently, the China Machinery Industry Federation released the operation of the machinery industry in the first half of 2019, in which the data of the automobile industry showed that the growth rate of total profit turned from positive to negative, and it showed negative growth for the first time in more than a decade. In the second quarter, the utilization rate of automobile manufacturing capacity was 76.2%, down 2.1 percentage points from the first quarter. The National Bureau of Statistics’ "National Industrial Capacity Utilization Rate" data shows that the capacity utilization rate of China’s automobile manufacturing industry was 77.2% in the first half of 2019, down 3.8% year-on-year.

  Compared with the normal range of capacity utilization rate of 79% ~ 83%, the figure of 77.2% means that China’s automobile capacity utilization rate has fallen below the "safety line". However, as the signs of recovery in China’s automobile market are still not obvious and the market is weak, the sales data in July showed that the new energy automobile market, which has been growing at a high speed, experienced negative growth for the first time, so the situation of idle production capacity may continue to deteriorate. How to find a way out for excess capacity has become a top priority for many auto companies.

  The production capacity of car companies exceeds 60 million.

  In fact, the word overcapacity has been in parallel with the rapid development of China’s automobile industry. As early as 2006, the State Council clearly pointed out that the automobile industry was overcapacity. The National Development and Reform Commission also issued documents to control new vehicle projects and appropriately improve investment access conditions, aiming at curbing overcapacity. At that time, the annual sales volume of China cars was 7.2 million.

  However, with the irrational growth of China’s automobiles, many automobile companies have long forgotten the overcapacity in the face of double-digit annual growth. 10 million vehicles, 20 million vehicles … … The huge increase in automobile production and sales has made many automobile companies see the great potential of the China market, and they are overly optimistic about the market expectations. Many automobile companies have started to invest blindly and continue to expand their production capacity reserves. The news that "the fourth factory of XX automobile enterprise is officially put into production" and "the foundation laying ceremony of the fifth factory of XX automobile enterprise is officially held" is constantly seen in newspapers. According to the report "Capacity of China Automobile Industry at the End of 2015", by the end of 2015, China’s automobile production capacity was 31.22 million. Among them, during the period from 2011 to 2015, China’s new automobile production capacity increased by 10.87 million vehicles; By 2016-2017, the new production capacity will be about 6 million vehicles. At present, China’s automobile production capacity is still in the stage of continuous expansion. While auto companies are still investing and building factories for the rapid growth of the market and think that the auto market will climb further, China’s auto market suffered the first decline in sales in 2018, and began to continue to slump. The problem of overcapacity in the auto manufacturing industry, which has not attracted enough attention, has finally been exposed.

  In 2018, the planned annual production capacity of China’s automobile industry has exceeded 60 million vehicles, among which the announced production capacity of new energy vehicles will exceed 20 million vehicles by 2020, which is 10 times the sales target of the Medium and Long-term Development Plan for Automobile Industry. According to the data released by China Association of Automobile Manufacturers, the annual automobile output in 2018 is only 27.809 million, and the growth rate of production capacity reserve is far greater than the growth of market demand, so it is inevitable that a large amount of production capacity will be idle.

  The capacity utilization rate of many car companies is less than half.

  Although the capacity utilization rate of China’s automobile manufacturing industry in the first half of 2019 given by the National Bureau of Statistics is 77.2%, in fact, the capacity utilization rate of many automobile companies has already been lower than this figure. In the automobile industry research report released some time ago, Guosen Securities pointed out that in 2018, the capacity utilization rate of Geely Automobile, FAW Group, Chery Automobile, BYD Automobile, Jianghuai Automobile, Dongfeng yueda Kia, Changan Ford and Haima Automobile did not reach 70%. According to foreign media reports, in the first half of 2019, the capacity utilization rate of Ford’s factory in China was only 11%; The capacity utilization rate of the joint venture between Peugeot Citroen Group and Changan Automobile is only 1%, and that of the joint venture with Dongfeng Motor is only 22%.

  According to public information, Changan Ford has five vehicle factories in Chongqing, Harbin and Hangzhou, with a total production capacity of about 1.6 million vehicles. After Ford’s sales in China reached its peak level three years ago, it sold a total of 957,000 new cars in 2016, and then continued to decline. In 2018, the output was 387,000, and the capacity utilization rate was 24.2%. In the first half of 2019, Changan Ford sold a total of 75,000 new cars, a year-on-year decrease of 67%.

  Another hardest hit area with idle production capacity is French car companies. Founded in 1992, Dongfeng and PSA each hold 50% of the shares. It owns two brands, Dongfeng Peugeot and Dongfeng Citroen, and four production bases, among which the first, second and third factories of Shenlong are located in Wuhan, and the fourth factory is located in Chengdu. The data shows that the production capacity of the first three factories is 300,000, 150,000 and 300,000 respectively, and the total production capacity of the four factories is 990,000. According to the production and sales data released by the Federation, in the first half of 2019, Shenlong Automobile produced a total of 64,000 vehicles, down 61.6% year-on-year. Changan Peugeot Citroen has produced only 107 vehicles, down 96.9% year-on-year, while its first-phase factory has a production capacity of 200,000 vehicles. Serious idle production capacity has attracted the attention of PSA Group executives. Previously, in the second quarter earnings conference call, Philippede Rovira, the global chief financial officer of PSA Group, said that PSA Group would solve the problem of idle capacity of Shenlong Automobile by renting factory facilities.

  The overcapacity problem of Renault, another legal brand, in China is also very prominent. The total production capacity of Renault Wuhan plant is 150,000 vehicles, and only 48,000 vehicles were produced in 2018, with a capacity utilization rate of 31.9%. In the first quarter of this year, Dongfeng Renault’s capacity utilization rate was only 15.2%, with 5,700 vehicles produced and 5,800 sold.

  The situation of Korean car companies is equally bad. Since entering the China market in 2002, Hyundai Motor has quickly won the favor of the market by relying on the advantages of high cost performance in joint venture brands, and in 2013— In 2016, the annual sales volume exceeded one million. However, the sales of Hyundai Motor in China plummeted because it didn’t follow the changes of consumers’ tastes and launch products that kept pace with the market demand. In 2018, Hyundai Motor’s sales in China dropped to 790,000 vehicles, only half of the total production capacity of 1.65 million vehicles in China. Among them, Hyundai Motor Chongqing No.5 Factory was completed in 2017, and the operating rate has been very low. According to BusinessKorea, Hyundai Motor is preparing to transform its Chongqing factory into an electric vehicle factory to revive its declining business in China.

  The situation of the same brother Dongfeng yueda Kia is similar. According to the 2018 annual report, Dongfeng yueda Kia has three factories in Yancheng headquarters, with a total production capacity of 890,000 passenger cars per year. Based on the sales volume of 370,000 vehicles of Dongfeng yueda Kia in 2018, its capacity utilization rate is less than half.

  Japanese car companies have insufficient production capacity and are still expanding.

  Although the overall capacity of China’s automobile industry is overcapacity, the capacity utilization rate of some automobile enterprises is not only higher than the safety line of 80%, but even the capacity is insufficient. According to the data released by Bernstein, a research consultancy, at present, the capacity utilization rate of Daimler and BMW joint ventures in China has exceeded 90%. The capacity utilization rate of GM’s joint venture in China is 88%. Volkswagen’s joint ventures in China have also exceeded 80%. The capacity utilization rate of the joint venture between Honda and Toyota in China even exceeds 100%. Relevant persons of Dongfeng Honda even revealed in an interview with the media that the capacity utilization rate of this enterprise is close to 140%.

  Due to a serious shortage of production capacity, Toyota is still starting a new factory. At present, Toyota’s two joint ventures in China, FAW Toyota and GAC Toyota, have eight factories in Changchun, Chengdu, Tianjin and Guangzhou. The total production capacity of FAW Toyota’s three factories and five factories is 622,000 vehicles, and the total production capacity of GAC Toyota’s three factories is 480,000 vehicles. FAW Toyota sold more than 720,000 vehicles in 2018, with a target of 745,000 vehicles in 2019, and GAC Toyota sold 580,000 vehicles in 2018. 720,000 vehicles plus 580,000 vehicles, which shows that Toyota’s eight factories in China have been working overtime and overloaded in 2018, and the capacity utilization rate has exceeded 100%.

  In 2019, Toyota’s sales target in China was 1.6 million vehicles, up 8.4% year-on-year, and the target in 2020 was 2 million vehicles. Toyota’s production capacity in China is only 1.1 million vehicles, compared with 2 million vehicles, there is still a gap of nearly 900,000 vehicles. Therefore, at the off-line ceremony of the brand-new TNGA Asian Dragon on March 22nd, FAW Toyota announced the completion of a new TNGA plant, and simultaneously started the new plant capacity increase project. According to the relevant EIA report, FAW Toyota has started to increase the capacity of 240,000 new energy vehicles, including 120,000 in Tianjin Xinyi Plant and 120,000 in Changchun Fengyue Plant. The 12-million-yuan expansion of GAC Toyota’s third plant has also been officially launched. At the same time, the fourth factory with a total investment of 4.988 billion yuan and an annual production capacity of 200,000 new energy vehicles is carrying out preliminary work, which is expected to be the production base of new energy vehicles exported by Toyota to Asia in the future.

  Where is the way out for idle capacity?

  A few car companies urgently need to expand production capacity, but most car companies still need to deal with idle production capacity. For some car companies, how to digest excess capacity has become a major event related to the survival of enterprises.

  For example, Shenlong automobile, which has been in the vortex of various rumors recently. Since 2015, the sales of Shenlong Automobile have been falling. 705,000 vehicles, 600,200 vehicles, 377,500 vehicles and 253,400 vehicles. The number in the first half of this year was 63,000 vehicles, down 60.05% year-on-year. The decline in sales volume is undoubtedly accompanied by the idleness of production capacity and manpower. How to solve this problem? There are rumors that Dongfeng Motor wants to sell its PSA shares. After clarification by both parties, it is reported that Dongfeng Motor and PSA Group will lay off employees and sell the factory … …

  In addition to unconfirmed news, it is certain that many auto companies have taken action early in order to transform idle capacity. Dongfeng yueda Kia temporarily closed an automobile manufacturing plant in Yancheng, Jiangsu Province, and plans to start producing electric vehicles in the plant by the first half of 2021.

  Compared with closure and suspension, changing production is a relatively good choice. Since last year, BAIC Group has adjusted its industrial structure, product structure and industrial chain, and transferred the Beijing branch of BAIC to Beijing Benz, becoming the third factory of Beijing Benz, and turning it into the production base of Mercedes-Benz high-end new energy vehicles in the future.

  Some companies have chosen to work for the new car-making forces. For example, Jianghuai OEM Weilai, Haima OEM Tucki, Changan Suzuki OEM Luchi Automobile, Dongfeng yueda Kia OEM Chinese Express & HELIP; … Although the OEM model has been controversial, after the cooperation between the two parties, traditional car companies can effectively revitalize idle production capacity, and new car-making forces can also save the construction cycle and quickly reach mass production.

  In addition to OEM, it is also a good idea for traditional car companies to warm up with new car-making forces and set up joint ventures. For example, Guangzhou Automobile Group will set up a joint venture with Weilai Automobile, and Tianjin faw xiali will set up a joint venture with Bojun Automobile. For the new car-making forces, they can obtain the qualification of new energy production and sales, at the same time, save the cost and cycle of investment and factory construction, rely on the manufacturing advantages of traditional car companies, and create brand-new products through joint development (the best cooperation state) to achieve the purpose of reducing costs and increasing efficiency. For traditional car companies, idle production capacity can be reused. Bundling the brand promotion of new forces to build cars can also enhance their brand influence. So far, the two sides have achieved win-win cooperation.

  Expert opinion

  Is OEM the best way to solve idle capacity?

  Compared with stopping production, closing factories or selling land, OEM for new car-making forces has become a good choice for many enterprises with serious overcapacity.

  However, many professionals believe that OEM is only one of the transition strategies. "The loss of fixed assets can’t stop, and OEM earns points." Whether it is a traditional car company or a new force to build a car, OEM is not a long-term solution. " An insider of an independent brand said.

  Recently, it is reported that the draft management measures for OEM have been drawn up, which requires R&D investment, production capacity and sales volume as restrictive conditions, and the threshold for OEM and OEM car companies has been greatly improved. For example, in the past three years, the domestic R&D investment has reached at least 4 billion yuan; In the past two years, the sales volume of pure electric passenger cars worldwide has reached at least 15,000; The OEM contract must be signed for 3 years or more, and the annual production capacity of the OEM in the same place should reach at least 50,000 vehicles; Enterprises need to have paid-in capital of billions of RMB; At most, it can only be contracted by two car companies.

  "The OEM model is more like a transitional means. Because with the growth of output and sales, it is essential to build a factory and have independent research and development and production strength. This is also conducive to solving possible production technical problems and improving production efficiency. " Cui Dongshu, secretary general of the National Passenger Car Market Information Association, said.

  This view has also been confirmed by Shen Hui, chairman of Weimar Automobile. Shen Hui believes that the cold winter of the automobile and capital markets has not dissipated, and capital will be more concentrated in the head enterprises that have achieved batch delivery. "Generally speaking, high-quality production capacity will not be idle, and idle production capacity will not be of high quality. Therefore, encouraging the coexistence of various production modes such as OEM is also an important lever for utilizing and upgrading idle and backward production capacity." However, Weimar chose a self-built factory with higher cost and greater difficulty instead of OEM because it attached importance to product quality and quality. "The self-built factory can control the quality and quality on the basis, and can continuously upgrade and improve, and at the same time realize C2M customized production."

  Simply speaking, C2M customized production is to make products meet the individual needs of consumers. Zhu Weihua, secretary-general of the Information Service Committee of China Automobile Industry Association, said frankly: "The overcapacity of automobiles needs C2M rescue."

  Zhu Weihua believes that at present, all kinds of personalized customization are pseudo-customization in nature, and car owners have only a limited number of choices in the end. Car companies are still selling by production in essence, but they only produce more configuration varieties and give users more choices. It is impossible to talk about the user placing an order and then producing it. As a result, car owners give up their personalized choice, and car companies waste their costs on spare parts inventory and dealer inventory, and there is no incentive to optimize production and manufacturing costs.

  Judging from the grim situation of the current automobile market, it is possible for car sellers to satisfy the needs of a small number of users only by working very hard to please users. If users are not given more choices, it is impossible to sell a large number of cars. The larger the car companies, the more they need C2M, and then consider the differences between car networking and electric car service packages, and consider the personalized marketing of these personalized services.

  In addition to starting with products, many people in the industry also suggest that enterprises should integrate and hold a group to keep warm. Dong Yang, the first vice chairman of the International Automobile Manufacturers Association, once said that the lack of cooperation is an important shortcoming among China brand enterprises. If we can get better development without joining forces in the past, then in the current market environment, it is even more important for China car companies to strengthen cooperation and learn from each other’s strong points for mutual benefit and win-win results.

  Giving full play to the synergistic effect and establishing alliance relations are the forms to cope with the increasingly severe competition in the market. Last year, FAW, Dongfeng and Changan took the lead in establishing strategic cooperation in four areas: forward-looking common technological innovation, automobile full value chain operation, expanding overseas markets and exploring new business models. In addition, traditional car companies, internet companies, travel service companies and new car-making enterprises have begun to cooperate hand in hand in various fields. The cooperation with new power enterprises has also gradually expanded from the earliest OEM production to in-depth cooperation in research and development, procurement and sales.

  When the downward trend of the automobile market encounters industrial transformation and upgrading, there will be a large number of car companies whose operating conditions are worrying. Zhu Huarong, president of Changan Automobile, once said that in the next 3-5 years, the shutdown and transfer of car companies will no longer be information. Most automobile enterprises will be eliminated. Faced with the uncertainty and complexity of the market, it is necessary for car companies to integrate and develop before, between new forces and traditional car companies to achieve a win-win situation.

  Text/reporter Li Dongying