Summary: Currently, AI competition has been upgraded to national competition.
At present, the competition of AI has been upgraded to national competition. For the first time in 2017, the "AI China Threat Theory" appeared. Former Google Chairman Schmidt publicly complained that the US government's lack of support for enterprises in the field of artificial intelligence will enable the United States to give up China's dominance in this field "within 10 years."
Behind the complaints, China’s artificial intelligence has accumulated and achieved certain achievements, especially in the application layer, which has reached a level similar to that of the United States. For example, in mobile payment, China's mobile payment penetration rate is currently 77%, ranking first in the world. Behind a large number of applications, artificial intelligence plays a key role from brush face payment to algorithm optimization. Americans can't help but drink a pot of old vinegar.
Is this really true?
When we did the "Comprehensive Interpretation of the Development of Artificial Intelligence Industry in China and the United States", we found that the number of artificial intelligence enterprises and the number of talents in China are only half of that of the United States; the United States has a comprehensive layout, while China is a enterprise or a talent, at the industrial base. , technology layer, application layer, uneven distribution, only the application layer is slightly accumulated.
The complaints of Schmidt cannot, after all, cover up the huge industrial gap between China and the United States.
Instead of focusing on who threatens, it is better to focus on technological innovation. This is the problem that every AI company should always think about, and it is also the duty of a technology company.
However, the industry is not calm enough now. There have even been some signs of concern. Looking back at 2017, there have been three breakthroughs in the field of artificial intelligence. Algorithms, policies, and funds have all created milestones. The industry is cheering. This situation is like the situation of the Internet bubble flooding at the end of 1999.
Looking forward to 2018, a large artificial intelligence, insufficient excellent projects, insufficient top talents, and lack of scenes, how will the three major problems be solved?
Three breakthroughs in 2017
1, the breakthrough of the algorithm
To say that artificial intelligence was introduced into the climax of public opinion in 2017, it would be necessary to mention the game of chess. AlphaGo from Google beat the world's number one Ke Jie 3-0, and then AlphaGo Zero gained more than AlphaGo, won the 100 games and overtook all the old versions in 40 days.
The previous generations of AlphaGo mainly used deep learning algorithms, which were initially trained with thousands of human figures.
AlphaGo Zero skipped this step and learned to play chess on the game. It started from chaos and used reinforcement learning. The system begins with a neural network that knows nothing about Go, combining the neural network with a powerful search algorithm to self-game. In the process of game, the neural network constantly adjusts and upgrades, predicting each step and the final winner.
Reinforcement learning is actually a branch of machine learning. Reinforcement learning is a kind of supervised learning with mark delay. It pays attention to achieving a goal through multiple steps and appropriate decisions under a series of scenarios. It is a problem of sequential multi-step decision making.
The results of AlphaGo Zero suggest that AI is not only a deep learning, but also that intensive learning is worth studying.
For the past three decades, deep learning has been considered a heterogeneity in academia, and the efforts of Geoff Hinton and his colleagues have made deep learning mainstream for speech recognition, image tagging, and other user experiences of countless online tools.
Interestingly, near the end of the year, Hinton, the father of deep learning, released a new paper Capsule, categorically claiming to abandon the theory of backpropagation and deep learning, and want to practice a new AI "Kung Fu" after 30 years of skill. Circles outside the circle.
Self-subversion or another leaps in AI. Li Feifei greatly appreciated this and issued a special statement: no tools are eternal, even backward communication and deep learning. What is important is that basic research continues to advance.
2. Policy breakthroughs
The top-level design of 2017 has clearly indicated the direction of industrial development. It can be expected that after 2018, a new round of development climax will be set up.
In order to seize the major strategic opportunities for the development of artificial intelligence, build the first-mover advantage of China's artificial intelligence development, and accelerate the construction of innovative countries and world science and technology powers, in July 2017, the State Council issued the "New Generation Artificial Intelligence Development Plan", proposed three Step by step, by 2030, the theory, technology and application of artificial intelligence will reach the world's leading level.
The Plan aims to develop five artificial intelligence 2.0 technologies (including deep learning, cross-border integration, human-machine collaboration, open-mindedness and autonomous control) to solve problems in the four major areas of technology, industry, society and national defense. It is worth mentioning that the planning also mentions the introduction of artificial intelligence and programming courses in primary and secondary schools. Artificial intelligence education starts from the dolls, and there is no one in the limelight.
Following the release of the “Planningâ€, on November 15th, the Ministry of Science and Technology held a new-generation artificial intelligence development plan and major science and technology project kick-off meeting in Beijing, and announced the establishment of four major platforms for artificial intelligence based on Baidu, Ali, Tencent and Keda Xunfei. It marks the beginning of a comprehensive start-up implementation phase of a new generation of artificial intelligence development planning and major science and technology projects.
As entrepreneurs and entrepreneurs, what kind of artificial intelligence technologies and products are developed in 2018, and how to develop artificial intelligence technologies and products? Open the "Planning", especially the content of "cultivating high-end and efficient intelligent economy", we must find some ideas: "Develop vigorously develop artificial intelligence emerging industries, transform technology into applications, realize intelligent hardware and software, intelligent robots, intelligence Innovations in vehicles (cars, boats, airplanes, rockets, etc.), VR/AR, smart terminals and IoT infrastructure devices; accelerate the upgrading of industrial intelligence, promote the transformation of traditional enterprises, and make manufacturing, agriculture, logistics, finance, and business And large-scale application of artificial intelligence in various fields such as home and home; vigorously develop smart enterprises, promote enterprise intelligence upgrade, promote the application of intelligent workshops; create artificial high-tech innovation highlands, and encourage the construction of industrial clusters centered on talents, enterprises and production factors, Industrial Park."
3. AI investment and financing breakthrough
A change in the low-key of the previous two years, the capital of 2017, high-profile gathering in a handful of smaller-scale AI entrepreneurship projects.
July 11, $410 million! Shangtang Technology refreshes the single round financing record in the AI ​​field!
On October 31, $460 million! Defiance Technology won a $ 460 million C round of financing, once again refreshed the financing record!
In 2017, a series of dazzling financing incidents broke out.
In 2017, China's AI investment and financing reached a record high, with a total investment of 58.2 billion yuan in one year.
In the hot investment field, VC's attention to computer vision and image, natural language processing and intelligent robots continues throughout the year, and its trend is basically in line with the "Comprehensive Interpretation of the Development of Artificial Intelligence Industry in China and the United States" and "The Interpretation of China and the United States" released by Tencent Research Institute in August. The forecast of the China-US Artificial Intelligence Venture Capital Trend Report.
It is worth mentioning that the domestic AI chip unicorn appeared. For a long time, China's information industry has been subject to people. The backwardness of industrial core chips is not only the lack of technology and capital, but also the lack of industrial ecological awareness. The long investment cycle of AI chips, large amount of money, and small output make many investors and enterprises discourage it. The $100 million financing will be used to develop the productization and marketization of domestic AI chips, which will help promote the industry's road to independent development.
More porridge is less, and the bubble is brewing. Since the number of startups has fallen from the previous two years, the funds in 2017 are obviously biased towards the middle and late stages, and most of them are some mature projects. The amount is quite huge.
In 2018, will investors be more enthusiastic about AI startup projects?
Many AI start-up projects belong to the “three missing one†project. Without original technology, lack of application scenarios, lack of maturity, and only one concept, it is difficult to push open the door of the market in the laboratory.
Three major problems in 2018
1, a lot of money, the project is not enough
The current AI industry is facing the risk of foaming, mainly reflected in the large amount of investment supply and the small amount of project supply. The market places high expectations on entrepreneurial projects, and the actual product experience is not good.
The bubble is about to appear. In the report “The Comprehensive Interpretation of the Development of Artificial Intelligence Industry in China and the United States†published by Tencent Research Institute, two signals that triggered the industry bubble were analyzed:
First, there are many funds and the project is lacking.
Based on past data and the situation in the first half of 2017, the number of new companies in the United States will fall to the bottom this year. On the occasion of 2017, the number of new companies in the United States ranges from 25-30. At the same time, the cumulative financing of the United States continues to grow rapidly, and will eventually stabilize in the range of 138-150 billion yuan.
After 2018, the growth of the number of AI companies in China and the United States will recover, but it will remain flat. During this period, the venture capital circle will find it increasingly difficult to find a new potential project. Due to the scarcity of new companies, it is often only possible to vote for some projects.
By 2020, the cumulative number of AI companies in the United States will exceed 1,200, and the accumulated financing will reach an astonishing 200 billion yuan. The growth of Chinese AI companies is uncertain. According to the industry development cycle, China's artificial intelligence industry will pick up in 2018, the number of new companies will rise to more than 30, and the cumulative amount of financing is expected to reach 900-100 billion yuan.
Second, the cycle is long and the revenue is difficult.
In layman's terms, artificial intelligence expectations are greatly overestimated. The deep learning that led the current AI boom originated from the neural network research in the 1980s and 1990s. In many cases, cutting-edge research consists of minor modifications and improvements to existing methods that have been designed decades ago.
In 2006, after the breakthrough of the deep learning algorithm, the market was hot, but the maturity of related AI technologies and products was still limited, and even ridiculed as "artificial mental retardation." Many projects and technologies take a long time to get popular with consumers.
From the perspective of investment and financing trends, funds pouring into artificial intelligence will continue to increase.
One basis is that according to incomplete statistics, there were about 353 investment and financing events in the field of artificial intelligence in China in 2017, a slight decline from 2016. However, the amount of investment has soared and the total financing amount is nearly 60 billion yuan. In the government's encouragement and industry mergers and acquisitions, China's AI investment will continue to increase substantially in 2018.
Another basis is that industry mergers and acquisitions have begun to intensify. According to data provided by CB Insights, nearly 140 artificial intelligence startups have been acquired since 2011, and in the first quarter of 2017, 34 artificial intelligence startups were acquired overseas, two for the same period last year. Times. In 2018, this trend will continue. At the same time as the funds grow, the number of Chinese AI companies cannot increase at the same time. According to the industry development cycle, China's artificial intelligence industry will rebound in 2018. It is expected that the accumulated financing will reach RMB 900-100 billion, and the number of new companies will only rise to around 30.
The funds are numerous and the projects are short, the cycle is long and the revenue is difficult, and the project is more expensive every day. This situation is similar to the first Internet bubble in 1999.
2, a lot of things, people are not enough
YoshuaBengio, the algorithm god, once said: "Deep learning is now hot. The current dilemma is the lack of experts. A PhD student will take about five years to develop, but five years ago, no doctoral students began to engage in deep learning, which means that the field is now There are very few experts, which can be said to be precious and extremely scarce." This is the dilemma faced by AI three years ago, and it has not been improved or even become more severe.
Artificial intelligence competition is based on top talent. It is said that there are no more than 50 top talents in the field of deep learning in the world. Andrew Ng said that the main reason for the lack of talents in deep learning is first of all data. It is not easy to solve data in certain areas; secondly, computing infrastructure tools. Including computer hardware and software; finally, engineers in this field have a very long training time. Therefore, technology giants and others have recruited talents through the acquisition of startups.
As the country's future development direction, AI technology plays a vital role in economic development, industrial transformation and scientific and technological progress. The research and development of AI technology, landing and promotion are inseparable from the collaboration of top talents in various fields. In the process of promoting the rapid development of the AI ​​industry, the leading role of AI top talents is particularly important. They are the key factors driving the development of artificial intelligence.
However, the development of talents in the field of artificial intelligence in China is extremely lacking.
According to the "2017 Global Artificial Intelligence Talent White Paper" published by Tencent Research Institute, there are about 20 research laboratories in China focusing on artificial intelligence, college teachers and reading about 7,000 students. The number of existing employees in the industry is about 39,000 people. Far from meeting the demand for millions of talents in the Chinese market.
From the perspective of industrial development, the distribution of talents in the field of artificial intelligence in China is seriously out of balance.
The artificial intelligence industry consists of the base layer (chip/processor, sensor, etc.), technology layer (natural language processing, computer vision and image, machine learning/deep learning, intelligent robot, etc.) and application layer (speech recognition, face recognition), etc. Composition, at present, China faces two problems at the industrial level:
Problem 1, the industry is unevenly distributed. The main employees of China's AI industry are concentrated in the application layer, and the talent pool at the base and technology levels is weak, especially on the processor/chip and AI technology platforms, which seriously weakens China's international competitiveness.
Question 2: Supply and demand are seriously unbalanced, and the talent gap is difficult to be effectively filled in the short term. In the past three years, China’s job seekers who are expected to work in the AI ​​field are growing at an annual rate, especially at the basic level of AI positions, such as algorithmic engineers, with supply growth of more than 150%. Despite the growth rate, it is still difficult to meet market demand. However, because the time and cost required for qualified AI talent training is much higher than that of general IT talents, the talent gap is difficult to be effectively filled in the short term.
Insufficient talent is a key factor that constrains the development of China's AI industry.
In recent years, the main purpose of Google's continuous acquisition of AI companies is to “buy†a group of the world's best experts. In a rapidly growing field of artificial intelligence, these experts are all outstanding. Other technology giants are also moving by camera.
It can be assumed that the flow of talent will also intensify. The introduction of talents needs to continue. In 2018, it is impossible to alleviate the hunger and thirst.
3, the scene is a lot, the road is not good.
If we sort out the AI ​​industry events in 2017, the combination of artificial intelligence technology and industry, the nine hot areas are everywhere.
Among them, the suspense among the three hotspots of medical care, finance and unmanned driving is greatly appetizing the public's appetite.
Suspense: Where is the signal of the transformation of AI medical care?
As a field of people's livelihood, medical reforms have changed year after year, but it is helpless. Venture capital also has ongoing support for AI+ Healthcare. In 2017, VCs flow into the AI+ medical field every month, and the accumulated financing amount of all medical artificial intelligence companies in China has exceeded RMB 18 billion.
The layout and application of smart medical technology in science and technology enterprises has taken shape. IBM Waston has been used in clinical diagnosis and treatment. It has been introduced to many hospitals in China in 2016; Ali Health focuses on creating medical imaging intelligent diagnosis platform; Tencent in August 17 Launched Tencent's film to help doctors screen for esophageal cancer. Tuma deep-dimensional investment of 200 million yuan in November, is introducing deep learning into computer-aided diagnosis system, XtalPi recently raised $15 million for a new generation of smart drug development technology to solve drugs Efficiency and success rate issues in preclinical studies.
Regrettably, despite the government's green light, companies have invested in human and financial resources, but artificial intelligence has not exploded in the medical field. What is the reason? It is that artificial intelligence requires a large amount of shared data, while hospital and patient data is like an island. How to break the barriers of all parties to ensure health while ensuring data security? This will be an important signal to promote the rapid development of smart medical care.
Suspense 2, how does AI deeply instigate finance?
The same data problems faced by smart medical care are also in the financial sector. A large number of highly credible data are held in the hands of major banks. How can AI extract these data to promote the innovation of financial technology, which is the brains of entrepreneurs The subject of thinking.
At present, face recognition and fingerprint recognition technology, as a solution for verifying customer identity, remote account opening, and brushing face payment, to solve financial security risks, have been gradually developed.
The technology of how to use the knowledge map to mine potential customers and further tap the potential needs of customers is also mature, and the problem of data sources needs to be solved urgently.
The US technology company FutureAdvisor first developed the "robot financing consultant." Subsequently, such robotic financial advisors quickly became popular around the world.
In 2017, smart investment is even more hot, and it is regarded as the next vent. However, the robot stocks, the result is lost.
Suspense three, when will the smart car be listed?
Driverless cars are called "four-wheeled robots", but when will their development be like a smart phone, and one person will completely subvert the traditional mobile phone and promote the transformation of the entire industry? This answer is still a suspense.
In 2017, the intelligent vehicle-making forces in the automotive industry continued to move, and some of them have successively handed over the answer sheets, allowing the products to be tested by the market, while some of them are still being cultivated in the greenhouse, waiting for the results. The reason why it is called "greenhouse" is because all walks of life have high hopes for it, especially in investment and financing. Although the number of investment events is small, the amount is 23.4 billion yuan.
Baidu announced the opening of the Apollo platform. Alibaba cooperates with traditional car companies such as SAIC. At the beginning of the year, Tencent successfully became the fifth largest shareholder of Tesla and led the first pure electric product of Weilai Automobile.
Time is racing against us. In 2017, the unmanned vehicle was taken to the Beijing Fifth Ring by the traffic police investigation. On December 20th, a Baidu Apollo unmanned vehicle team was tested and started in Xiong'an New District. At the beginning of 2018, the unmanned pilot operation base in Shunyi District of Beijing was officially launched. It became the first area in the city to launch the driverless trial operation since Beijing introduced the first new automatic driving regulations in China. Who will be on the road in 2018? Industry and consumers are waiting to see.
Conclusion
Looking back at the disillusionment of the Internet bubble in 2000, many people still find it incredible. At that time, the development of the industry was changing with each passing day. Software applications, Internet service ISPs, and ICP outbreaks of online content often did not seem like a sigh of three autumns.
In April 2000, the Nasdaq index rushed to the top of history, more than 5,400 points. But unfortunately the bubble burst and the capital market collapsed. The Nasdaq index quickly fell. In the middle, I experienced the 9/11 terrorist attacks and the Enron incident. The winter lasted for three years and slowly walked out of the trough.
Today's AI industry is booming, similar to the early stages of the Internet.
The industry has a dazzling aura, shoulders the heavy responsibility of the national strategy, and is highly dependent on the capital market channels. Public opinion is highly popular, and practitioners are all three high social elites.
However, the fly in the ointment is that the products listed are not experienced well, the application scenarios are slightly insufficient, and the projects of fish and pearls have appeared.
Not to mention how Nasdaq, the A-share market alone, is often tempted. The sensible capital market has become hormonal and the taste of the bubble is getting stronger. Whether it is the government or the enterprise, everyone should take precautions against future risks.
China's leading manufacturer and supplier of solar panels, all black solar panels, PERC solar panel technology in China, we specialize in solar panels, solar power systems and more.
Jinko Solar Panels, Longi Solar Panels, Canadian Solar Panels, Trina Solar Panels, JA Solar Panels, Solar Panel, Monocrystalline Solar Panel, MONO Solar Panel, Polycrystalline Solar Panel, All Black Solar Panel, 400W Solar Panel
Power X (Qingdao) Energy Technology Co., Ltd. , https://www.solarpowerxx.com