Where is the direction of artificial intelligence? See the advice of the four deans of Microsoft Research Asia

On November 14, Li Kaifu, Zhang Yaqin, Shen Xiangyang, and Hong Xiaowen also appeared at the Microsoft Asia Research Institute in Danling Street, Zhongguancun. These four were the first, second, third and current dean of Microsoft Research Asia. Now Kai-fu Lee is the chairman and CEO of Innovation Works, Zhang Yaqin is the president of Baidu, and Shen Xiangyang is the global executive vice president of Microsoft. On the same day, he appeared at the inauguration ceremony of the Microsoft Asia Research Association.

Li Kaifu, Shen Xiangyang, and Hong Xiaowen are all students of Raj Reddy, a global expert in artificial intelligence technology. The main investment direction of the innovation workshop is AI (Artificial Intelligence). Baidu’s current main direction is AI, Shen Xiangyang is Microsoft Labor. The head of the Intelligent Business Department and Hong Xiaowen’s own research business are also AI. These four people get together, how can you not talk about AI? The theme of this year's Alumni Association is "Let the world be full of AI."

Shen Xiangyang: SmartX is AI

As the president of the First Alumni Association, Shen Xiangyang said: "Because I know that my brother Li Kaifu is coming, and I want to talk about AI, I went to the mentor's office to "steal the teacher." This is of course a humility. Shen Xiangyang is the global executive vice president of Microsoft, the head of Microsoft Artificial Intelligence and Microsoft Research. At present, Microsoft is injecting AI into various products and services. His judgment on AI reveals the Microsoft AI strategy.

Shen Xiangyang said that from the personal computer era to the Internet era, now is the mobile era, and more importantly, the intelligent era, intelligence will bring huge contributions and tremendous impact on this society. Some people say that PC is the extension of human limb tools, and mobile phones are the extension of the human brain. “And in my opinion, all these devices are concrete entities, and only AI is really close to the human brain. In the process of human development, AI will be part of our evolution, and it can have different technologies. Looking back, these The development of humanity in the past, from the previous limbs on the ground, and later upright, to the later Stone Age, to the present information age, what happened? In this process, including the acquisition of knowledge into the evolution of the human brain, to DNA Inheritance, what happened inside? AI is likely to be a breakthrough place." Shen Xiangyang said.

What exactly is AI? If the original thing is called X, from X to SmartX is called AI. You can double it up for all the work today, and it's a bit of AI. A lot of work you can do 10 times or even more. There are a lot of such jobs now, such as marketing, customer support, and most of these jobs are still dominated. If 90% of the work is done for the machine, it is a 10x improvement. Shen Xiangyang said.

At present, among the traditional IT giants, Microsoft is a representative enterprise with relatively fast transition and relatively smooth performance. The performance has been rising and the stock has been rising. Microsoft has just established the AI ​​business unit and is advancing AI at full speed. As a representative of the traditional giants, Microsoft's overall evolution from "cloud" to "AI", its path and method have implications for many manufacturers.

Shen Xiangyang said that the truly useful innovations of artificial intelligence will not come from the ivory tower of the research laboratory, but will come from the integration of cross-border comprehensive innovation, and is a deep technology for large-scale production tasks and research results in all aspects of algorithms, systems and experiences. Combine.

Shen Xiangyang revealed that the core of Microsoft's AI strategy is to promote the popularization of AI in an all-round way, so that everyone can enjoy the tremendous changes brought by AI. Microsoft's AI products and solutions will advance from four dimensions: First, personal assistants. Including Microsoft Xiao Bing and Microsoft Xiaona, the main core is to help people do more things. The second is the application. In the future, all of Microsoft's application products AI will run through it. The third is service. The core is to enable all users and enterprises to use AI, and to more easily enjoy AI technology and promote ecological development. The fourth is the infrastructure. Includes Azure Machine Learning, Azure N Series, and FPGA.

As the head of Microsoft's global R&D system, Shen Xiangyang finally said that everyone knows that this round of AI boom was caused by deep learning, but few people know that the first paper on deep learning was made by Microsoft Research. At present, Microsoft has achieved 152 layers on the deep neural network.

Kai-fu Lee: Is it easier to start an AI venture in China than in the US?

Kai-fu Lee is the first president of Microsoft Research Asia. Both Shen Xiangyang and Zhang Yaqin are recruited to Beijing, so he has a high status in the Microsoft Research Institute.

Kai-Fu Lee talked about the contribution rate of global Chinese in the AI ​​field from 2006 to 2015, and said that many people may see similar data in the White House report, and the data he gave is the research his colleagues are doing. How many of the 50 top journals in the world (about AI) are Chinese? About 56% in 2015, this is a huge number of people who are very proud and optimistic.

Li Kaifu started doing AI research 30 years ago. Someone asked him what is the difference between doing AI 30 years ago and today? Kai-fu Lee said: "The biggest difference is that AI did not admit that he was doing AI 30 years ago. But today everything is called AI."

What value can AI create? Kai-fu Lee believes that many professions will be replaced by AI after 10 years. The intermediary or assistant will basically disappear. There are still many professions such as translators, drivers, workers, etc., which will be replaced by artificial intelligence. This replacement is far faster than we think, because there is a huge amount of data that will surpass and replace humans under deep learning and other technological learning. Simply put, when face recognition exceeds people, security and security can be replaced to a large extent. Speech recognition exceeds people, and a large degree of customer service can be replaced.

Kai-fu Lee believes that doing things must be at the right time. For example, Jeff Bezos (Amazon founder), Bill Gates, Jobs, because they met the computer when they were in high school, learned programming, and created a great company after the university dropped out of school. Kai-fu Lee listed several techniques, including his own AI. "Because the participation time is too early, or the early stage of technology development, individuals have become martyrs."

It’s time to do AI now. Li Kaifu is optimistic about several AI entrepreneurial directions: First, big data, who has the largest data, who has a very good opportunity to generate value. Of course, the biggest data in China is BAT, but banks, insurance companies, hospitals, etc., they also have a lot of data opportunities.

The second is driverless. This is the biggest slogan of artificial intelligence. The global capital power is being driven by unmanned driving. One after another, the top unmanned driving companies in the United States have started their own business. The team of six or seven engineers soon got 50 million US dollars. To 100 million US dollars.

Kai-fu Lee believes that when the data is huge, and the data has very clear boundaries, it can be collected automatically. For example, Didi tells you whether to take a ride. Baidu tells you whether you clicked on the search results, Taobao tells you whether to buy, plus the machine and The best scientists, these are the most critical conditions for the formation of AI applications.

Under such premise, what is the best AI practice company? Li Kaifu believes that the first is to form a closed loop, there must be big data; the second is to have a lot of machines; the third is to find the best scientists; the fourth young is a particularly important force.

Kai-fu Lee said that there are still many problems with AI today. How to solve it? First of all, you don't have to let AI replace people, maybe you can be a good assistant. Now AI has a lot of achievements in medicine, and its judgment has exceeded two-thirds of doctors. Second, some problems can hide errors by using a good UI (user interface). For example, a search engine presents 10 results. Maybe the first result is only 60% correct, but the UI can hide this error. Why does Google have a "just tell you a result" because the result on the search engine is not correct enough? of.

Why do you have some special opportunities to do AI in China? For example, there are still very few AI scientists in China. Although there are so many articles written, it is still relatively rare to apply AI to produce commercial value. There are many opportunities for this.

Air Fryer

Air Fryer No Oil,Multifunction Air Fryer,Electric Air Deep Fryers,Deep Fat Fryer With Timer

Ningbo ATAP Electric Appliance Co.,Ltd , https://www.atap-airfryer.com