(Original title: AI's Scenario Times: How Far Is Strong Artificial Intelligence?)
Reporter Yang Qingqing reports from Beijing
In 2017, if you want to ask the biggest changes in the science and technology circle, the answer is hard to escape the word “artificial intelligenceâ€.
No matter whether Xafeo heard the emergence of the two conferences during the two sessions this year, Baidu still opened the AI ​​capability and released the “proverbial mass production of L4 unmanned vehicles next yearâ€; whether Tencent released the AI ​​medical imaging product Tencent or the Alibaba ET city brain. The smart applications behind “stealth†cities are all related to artificial intelligence.
As the market has continued to pay intense attention to the field of artificial intelligence, policy dividends from the national level have also come. In March 2017, artificial intelligence was first written into the work report of the Chinese government; in July, the State Council issued the “New Generation of Artificial Intelligence Development Plan†and proposed the idea of ​​developing new generation of artificial intelligence in China in 2030; in December, “promoting a new generation The “Three-Year Action Plan for the Development of Artificial Intelligence Industry†was published, highlighting four key tasks of artificial intelligence in the next three years, involving 17 products such as intelligent networked vehicles, intelligent service robots, intelligent drones, and medical imaging-assisted diagnostic systems. field.
However, for the general public, this year, in addition to several optional smart speakers on the e-commerce platform, the life around them seems to remain the same: the car still needs manual operation, and even faster in the navigation through voice switching When it comes to the route, mistakes are still inevitable. Tide-like patients inadvertently flow into the top-tier cities of the top-three hospitals, hoping to have a conversation with the experts. Urban traffic under the control of smart brain becomes more smooth? "There is not much feeling." The digitally interviewed drivers and passengers told the 21st Century Business Herald reporter.
If artificial intelligence really is the future, how far is this future?
Scenes
According to data from PricewaterhouseCoopers, by 2030, artificial intelligence will contribute 15.7 trillion U.S. dollars to world economic GDP, surpassing the sum of the current economic aggregates of China and India; by 2035, artificial intelligence is expected to promote world labor productivity. The increase of 40% has led to a 14% increase in global GDP.
In the process of rapid changes in the AI ​​economy to the aforementioned time nodes, China has its own ideas. In November, the Ministry of Science and Technology jointly announced a list of the first batch of national new generation artificial intelligence open innovation platforms: Relying on Baidu, Alibaba Cloud, Tencent, and HKUST News to build autonomous driving, urban brain, medical imaging, and intelligent voice. Innovation platform.
"The establishment of a national artificial intelligence open innovation platform is guided by the direction of the country's strategic level." In an exclusive interview with a 21st Century Business Herald reporter, Li Jingmei, head of Alibaba Cloud's ET City Brain, said, "This is by no means a matter with local governments. Single-point cooperation means that future domestic research institutions and companies can participate in the innovation of the city's brain."
As an enterprise relying on the national city’s brain innovation platform, Alibaba Cloud had already set the ET city brains to Hangzhou in October last year. In a short period of one year, the ET city brain took over the traffic lights at 128 intersections in Hangzhou. The total travel time on the roads in the pilot areas decreased by 15.3%. The time spent on the backbone of Hangzhou had 4.6 minutes to save time. Artificial intelligence was used to handle traffic accidents. The accuracy rate reached 92.3%.
“The current ET city brain has entered cities such as Hangzhou, Suzhou, and Macau. The scene mainly focuses on transportation.†Li Jingmei told 21st Century Business Herald reporter, “ET city brain can realize intelligent perception and processing of urban events, targeted Social governance, light traffic control to ease traffic jams, and intelligent dispatch of operational vehicles."
Compared to the city brains that just started smart transportation, the development of smart voice is more mature. Both the smart speaker of this year's hot and the smart home products that have continuously evolved are all carrying or attempting to carry the function of intelligent voice interaction. .
According to data provided by the HKUST to the 21st Century Business Herald reporter, as of November 2017, the cumulative number of terminals of the information flight open platform has reached 1.5 billion, and the number of daily interactions has reached 4 billion. In an interview with a 21st Century Business Herald reporter, Liu Qingfeng, the chairman of HKUST, said that voice will be an important entrance to the era of the Internet of Everything, and it is also the threshold for truly making machines intelligent.
"In terms of the transformation of technology into applications, HKUST is taking root in various industries." A person from HKUST told reporters in the 21st Century Business Herald that currently HKUST's communications include smart education, smart cities, smart medical services, and smart cars. , justice, smart customer service, smart home and other fields, trying to integrate intelligent voice and industry.
If the infiltration of smart voice applications is more like “smooth things,†the unmanned industry’s news has always been “thundering.†Li Yanhong, chairman and CEO of Baidu, said at the 2017 Baidu World Congress that Baidu’s Apollo open platform will mass-produce China's first driverless micro-cycle bus “Apollon†in 2018, mainly in the automotive towns, parks, and scenic spots. Airports, city-level unmanned pilot areas and other restricted areas to achieve L4 automatic driving.
This also means that Baidu has, to a certain extent, fulfilled the slogan of committing “three years of commercial use†of unmanned vehicles at the end of 2015. At the same time, Li Yanhong announced that Baidu will launch auto-driving production models with JAC and BAIC in 2019 and Chery in 2020.
One of the key points of Tencent's military formation is medical care. This year, Tencent released the AI ​​medical imaging product "Tencent", which contains six personal intelligent systems that involve diseases including esophageal cancer, lung cancer, glycogenosis, cervical cancer, and breast cancer.
According to Yang Ying, product manager of Tencent Films, Tencent Yingying processes millions of medical images every month, the accuracy rate of early screening system for esophageal cancer is over 90%, and the accuracy rate of lung nodule early screening system exceeds 95%, and 3 mm can be detected. And above the tiny nodules, sugar net recognition accuracy is as high as 97%.
Burst point
Although the four domestic scenes of artificial intelligence are slowly spreading out, the key to the problem is, which one will quickly become the deceleration of the next round of commercial growth?
According to CCID Consulting's “2018 China's Artificial Intelligence Industry Development and Investment Value†report, the core scale of China’s artificial intelligence industry exceeded RMB 70 billion in 2017, of which hardware products accounted for 55%, support layer accounted for 31%, and software products accounted for Only 14%. In the subdivided product structure, sensors, wearables, and unmanned/assisted driving were among the top three with 31%, 25%, and 19% respectively, followed by voice recognition platforms, machine vision systems, and service robots. , smart drones and other products.
“Intelligent customer service, voice processing and simultaneous translation are the most mature and easiest scenes at present, and they are easy to see and feel.†Around the future development of several scenarios, Microsoft’s China IoT and artificial intelligence senior product market Manager Li Lan commented to the 21st Century Business Herald reporter.
Another scene that Li Hao is optimistic about is artificial intelligence assisted medical treatment. “Through technical means, artificial intelligence has been able to improve the efficiency of repeated medical links such as film viewing, and at the same time it can conduct comprehensive monitoring after comprehensive patient data. These are difficult to match with only human resources.†Li Wei pointed out, “The medical field Some of the scenes can already be started."
Xiang Yang, deputy general manager of CCID Consulting's artificial intelligence industry research center, expressed different views. He pointed out that service robots in the intelligent voice industry are expected to be the starting point for AI's entry into the home, but the devaluation of future business growth will continue to be in the unmanned industry. “All aspects of the unmanned driving industry are already very obvious, including product forms, ecological environment, etc., and a trend has been formed.†Xiangyang told the 21st Century Business Herald reporter.
Xiangyang’s judgment is based on demand. “The driverless driver can free his driver’s hands, and at the same time increase the safety of driving on the road. Under this just-needed role, industry development is inevitable.†Xiangyang pointed out to reporters: “The current intelligent voice is not the most important solution in life. The problem of pain point is to increase the added value of products, so although it is very cool to make, commercialization still requires continuous exploration of the scene.
However, an industry insider who declined to be named told reporters in the 21st Century Business Herald that although driverlessness may be expected, it is still not easy. “In the process of industrial development, Internet companies not only need to cooperate with the traditional depots, but also involve many social and ethical aspects. For example, in the case of the elderly on the right and the wall on the left, it can be achieved through technology to judge the collision or collision. But it cannot exceed the ethical threshold."
In smart cities, a number of industry insiders said that it is still at an extremely early stage. "What we can do now is to monitor vehicles or individuals on a regular basis. The so-called city brains are more likely to stay at the level of smart transportation. There is still more space to be tapped in this area." The aforementioned industry sources commented, "The entire smart city The system has not yet been fully set up. It will take a long time to truly achieve landing."
challenge
Although Baidu has basically fulfilled its "three-year commercial commitment," it is still far from the qualitative leap of the unmanned field.
“Baidu’s mass production is the mass production of L4 commercial vehicles. More vehicles are fixed-route vehicles. The personnel and the surrounding environment are not that complicated.†Xiangyang told the 21st Century Business Herald reporter bluntly, “The technology of L4 commercial vehicles has been relatively mature. In fact, the trend science and technology has also launched a prototype car, and foreign Otto unmanned trucks have already been on the road."
However, as an L4 passenger vehicle with a larger market space, it still needs to wait. The 21st Century Business Herald reporter learned that currently the industry generally infers that the time point for L4 passenger cars is around 2020. In addition to the ethical and regulatory dilemmas of this premise, even L4 unmanned passenger vehicles are technically difficult.
"Nvidia has introduced several high-performance computing chips, but the power consumption is as high as hundreds of watts. In comparison, smart phones have the lowest power consumption or less than 1 watt, and reducing energy consumption is the direction that the industry needs to work together. Xiangyang said, "A trend in the industry is that the trained model is loaded on the terminal chip, but it is still best for the terminal to handle the computational expansion."
In addition, the precision of high-precision maps, the fusion of multi-sensors, and the judgment of machine models all require breakthroughs. "It's very complicated to think about this matter. What's more, when it comes to real development, there are bound to be many pitfalls in the process." Xiangyang pointed out.
Despite the difficulties, the unmanned industry has at least a relatively well-accepted production time point. In contrast, regardless of medical imaging, smart voice, or smart cities, it is difficult to have a clear commercial node.
“Technically speaking, the speech recognition rate under the current multi-user dialogue is high, but there is still a gap from the actual application. Smart speech has a good application in quality inspection, but it has not reached a 100% recognition rate.†The afore-mentioned HKUST News source told a 21st Century Business Herald reporter that “including the recognition of various dialects will, to some extent, involve the issue of insufficient input-output ratio and cost-effectiveness. These are places where technology companies need to continue their efforts.â€
The difficulty of medical imaging lies in data. “The labeling of medical image data requires authority and requires the participation of experts. The threshold for experts is very high, which leads to a huge cost.†Xiangyang said, “In addition, the isolated islands of medical data are serious. This requires medical associations, enterprises, and institutions. With the completion."
In smart cities, Li Jingmei told reporters in the 21st Century Business Herald that some industry pain points are inevitable. "On the one hand, it is the commonality of data, on the other hand, it is the guarantee of data quality. Different organizations have different data format specifications and standards, among which old data is also mixed. In addition, data sharing is a problem, and technical protection issues also need to be addressed. In the future business will continue to face." Li Jingmei said.
Jiang Hao, a global partner of Roland Berger Greater China, told the 21st Century Business Herald reporter that the application layers and solutions for smart cities need multiple companies to cooperate and participate because of their complexity. “Smart transportation, smart medical care, smart education, smart buildings, etc., many industries require specific, placeable solutions, and no one company can 'unified rivers and lakes'.â€
Incremental encoders provide speed, direction and relative position feedback by generating a stream of binary pulses proportional to the rotation of a motor or driven shaft. Lander offers both optical and magnetic incremental encoders in 4 mounting options: shafted with coupling, hollow-shaft, hub-shaft or bearingless. Single channel incremental encoders can measure speed which dual channel or quadrature encoders (AB) can interpret direction based on the phase relationship between the 2 channels. Indexed quadrature encoders (ABZ) are also available for homing location are startup.
Incremental Encoder,6Mm Solid Shaft Encoder,Hollow Rotary Encoder,Elevator Door Encoder
Jilin Lander Intelligent Technology Co., Ltd , https://www.jllandertech.com