Most popular experts call on AI to practice basic

  • Detail

Experts' appeal: artificial intelligence needs to practice basic skills original title: experts' appeal: artificial intelligence needs to practice basic skills

if you want to select the most popular Internet technology in recent years, artificial intelligence is definitely one of the candidates. At present, artificial intelligence is like a big popular sign, and everyone wants to rely on these four words. However, in Wuzhen at the moment, on the scene of the 5th World Congress on the prohibition of bumping and impacting the interconnection of instrument industry, some people poured cold water: "the development of artificial intelligence is still in its infancy", "we are still in the era of weak artificial intelligence", "there are many foam in artificial intelligence"...

experts call for artificial intelligence to improve efficiency and effect, to land solidly, and to make up some lessons as soon as possible

high quality data is not easy to obtain many high-end scientific research achievements and patents from batch production to automatic production equipment.

to some extent, artificial intelligence is a data processing technology that can automatically make decisions through a large amount of data analysis. Therefore, to improve the efficiency of artificial intelligence, we must first ensure that there is sufficient quantity and quality of data. However, in some fields, it is not as easy to obtain high-quality and authentic data as many people think

for example, artificial intelligence has been widely used in disease screening and other fields, and the speed and accuracy of disease diagnosis can be improved through machine reading films (medical images). Weiwenbin, vice president of Beijing Tongren Hospital, is engaged in similar work, but he has found a problem. "Medicine is a complex subject. It is difficult to obtain high-quality data and images, and it is also difficult to read data." He said that there are many kinds of diseases in ophthalmology. For each disease, more than 100000 high-quality images need to be collected before artificial intelligence can make effective judgments. They cooperated with relevant institutions to establish an ophthalmic big data laboratory, trying to collect standardized and high-quality image data across the country

liruicheng, senior vice president of SAP worldwide and President of SAP China Research Institute, also encountered the same problem. Liruicheng tried to use artificial intelligence in their factory to judge whether there were quality problems in each production link of parts, but failed to achieve the expected results. "The most critical difficulty is that the quality of the photos we collected has the problem that 3 cyanamide cyanurate is a nitrogen-containing halogen-free environment-friendly flame retardant." "We are still in the age of weak artificial intelligence," he said What artificial intelligence can do is limited. Don't take it as a panacea to solve all problems

strengthen basic research on computing power and chips

even with data, without the support of algorithms and computing power, artificial intelligence may become an "artificial intellectual disability"

"computing power is productivity." Wangendong, academician of the Chinese Academy of engineering and chief scientist of Inspur Group, believes that in the digital economy era, the concept of "fast" is not only changing with each passing day, but even calculated in hours and minutes. "How many days, hours or minutes can we train a model? If we can train the model in real time, the development of artificial intelligence may be completely different."

the computational power of artificial intelligence determines its development potential. Wangendong believes that it is necessary to strengthen the construction of computing infrastructure and build a computing platform. "Some people say that we have caught up with or surpassed the United States in artificial intelligence, which is a bit optimistic. It should be said that there is still a gap, especially at the basic level. We should also strengthen basic research and investment in chips."

chukibuakolot, senior vice president of Xinsi technology, has the same view as wangendong. In his opinion, the biggest challenge facing artificial intelligence is to design intelligent chips with excellent computing power and simulating neural networks, and continuously improve the mass production capacity

chenzhaoxiong, Vice Minister of the Ministry of industry and information technology, suggested to strengthen basic theoretical research and make breakthroughs in the development direction, theory, methods, tools and systems of artificial intelligence; Strengthen the research and development of key common technologies, and focus on breakthroughs in smart chips, systems, sensors, core algorithms and other fields

find a "circle of friends" for artificial intelligence

the social value of artificial intelligence should ultimately be reflected in specific application links to achieve deep integration with the primary, secondary and tertiary industries. "The integration of artificial intelligence cannot be completed by one enterprise. It requires various enterprises such as hardware and software to create a symbiotic and integrated ecosystem." Liruicheng vividly compares that to truly implement artificial intelligence, we need to "find a circle of friends"

Baidu's experience in developing unmanned vehicles (autonomous driving technology) can annotate liruicheng's views. "To realize the intelligence of cars, there is also a matching thing, that is, the intelligence of infrastructure such as roads." Robin Lee, chairman and CEO of Baidu, pointed out that only cars are intelligent, but roads are not intelligent, and the effect of unmanned cars will be affected. Baidu took over the traffic lights in Haidian District, Beijing, and intelligently adjusted the time of traffic lights by monitoring the number of vehicles on the road in real time. "It is estimated that the waiting time of people in traffic congestion can be reduced by 30% ~ 40%."

"many enterprises gain the experience of artificial intelligence at a certain point in a certain link, but it is a great challenge to truly integrate artificial intelligence with the real economy." Zhoubowen, vice president of JD group and head of AI platform and research department, said that this requires division of labor and cooperation between enterprises

in fact, the "light of the Internet" Expo held during Engel's Graeme Herlihy Internet Conference is precisely helping enterprises find a "circle of friends" for various AI products and technologies. For example, the flexible AMOLED technology demonstrated by vicino can provide more possibilities for the application of various artificial intelligence in different scenarios, such as driverless on-board display and the facilitation of intelligent wearable devices

"insist on industrial interaction and expand integration space." Chenzhaoxiong called for conforming to the general trend of digital, networking and intelligent integration development, promoting the cross integration of artificial intelligence and various technologies, and deepening the integration and development of the same, secondary and tertiary industries. (chenhaibo)

Copyright © 2011 JIN SHI