Artificial intelligence: application threshold is lowered, and technology dividends will be realized

　　The next 5 to 10 years will be a period of great dividend realization for artificial intelligence. During this stage, all industries will make full use of the technological dividends of the past 65 years and compete to implement artificial intelligence, thereby promoting the development of various industries. -- Yuan Hui, founder and chairman of Shanghai Zhizhen Intelligent Network Technology Co., Ltd.

　　During the COVID-19 pandemic, the application of artificial intelligence has advanced by leaps and bounds, and all sectors of society have high hopes for artificial intelligence technology, and expect that artificial intelligence can become one of the core driving forces of the digital economy after the pandemic. In the "14th Five-Year Plan" and the Outline of the 2035 Vision Goals released in March, my country listed "new generation artificial intelligence" as one of the seven major areas of scientific and technological frontier research, and artificial intelligence has risen to a national strategy.

　　On April 19, the Boao Forum for Asia 2021 Annual Conference paid special attention to "Artificial Intelligence in the Post-Pandemic Era". Many experts were invited to discuss the industrial applications of artificial intelligence and the opportunities and challenges faced in participating in social governance online and offline.

　　Artificial intelligence technology urgently needs breakthroughs

　　Whitfield Diffie, the 2015 Turing Award winner and a famous American cryptography and security technology expert, has made great achievements in the field of Internet security and is known as the "father of public key cryptography technology." He was a member of the Stanford University Artificial Intelligence Laboratory. He said that artificial intelligence must have more computing power to achieve the development vision of artificial intelligence that we want to see. At present, human predictions about the computing power of artificial intelligence are still too conservative.

　　Yuan Hui, founder and chairman of Shanghai Zhizhen Intelligent Network Technology Co., Ltd., said: "From a technical perspective, artificial intelligence faces a 'bottleneck' problem. This is not the bottleneck of any country, but the bottleneck of the artificial intelligence system and the core of artificial intelligence technology development." He believes that the current artificial intelligence technology system and technical framework have not broken away from or exceeded the technological accumulation of the past 60 years. "Artificial intelligence technology should achieve greater breakthroughs in the future. This is a common challenge faced by scientists around the world."

　　Open source improves application efficiency

　　Baidu Group Chief Technology Officer Wang Haifeng said that there will be more and more integrated innovations in the artificial intelligence industry in the future.

　　He said that with the development of artificial intelligence technology today, from the perspective of scientific research, the problems faced are becoming more and more complex, but the application threshold will become lower and lower, and more people will not need to worry about core algorithms and can solve application problems through a good platform or model library.

　　When introducing PaddlePaddle, my country's first open-source, fully functional industrial-grade deep learning platform independently developed by Baidu, Wang Haifeng mentioned that PaddlePaddle currently has more than 2.6 million developers, and these developers do not need to start writing from the first line of AI algorithm code, but directly call the framework. On similar platforms, the threshold for AI applications has been greatly reduced. At the same time, this has also promoted the faster application and promotion of AI and the faster advancement of industrial intelligence.

　　In recent years, deep learning open source platforms are becoming an important choice for China's various industries to rapidly deploy artificial intelligence. The latest data shows that in the comprehensive market share of China's deep learning platform, Google, Baidu, and Facebook are firmly in the top three, accounting for more than 70% of the market share. Among them, Baidu's share increased by 3.38%, ranking first in growth rate, and its comprehensive market share ranked second, almost the same as Google, which ranked first.

　　Promote the implementation of various industries

　　In terms of industry applications, Yuan Hui said: "The next 5 to 10 years will be a period of huge dividend realization for artificial intelligence." During this stage, various industries will make full use of the technological dividends of the past 65 years and compete to implement artificial intelligence, thereby promoting the development of various industries, including medical care, education, etc.

　　"During the COVID-19 outbreak, when people could not communicate face to face, AI applications solved the rigid needs of people in production and life." Du Lan, senior vice president of iFlytek, said that as a leading AI manufacturer, iFlytek, which reaped the benefits of AI during the outbreak, has attracted much attention. During the domestic outbreak, when the second wave of imported cases increased, airport, customs and community workers were often seen using iFlytek translators to communicate with foreign guests. iFlytek's telephone robot also participated in the screening of 9 million people in Wuhan and was also used in South Korea's epidemic prevention and control.

　　"The technology and application of artificial intelligence have greatly increased the value of data." Duran particularly emphasized the data security issues of artificial intelligence. She said that the Personal Information Protection Law and the Data Security Law have gradually transitioned from the discussion stage to the stage of being formulated and issued, which will bring very good prospects for industrial development.

　　"Currently, artificial intelligence mainly comes into contact with biometric-related data. These data should be managed in a layered manner in advance, and operated in a standardized manner by institutions with confidential qualifications, and then shared after desensitization processing," said Duran.