论文部分内容阅读
Wh e n A n -drew N g ， one of the world’s leading thinkers on artificial intelligence， announced he would be stepping down from his position as chief scientist at Chinese search giant Baidu， the company’s stock dropped nearly 3 percent in just a few hours. It was a reflection not only of Ng’s prominence and fame， but also of the importance investors have placed on the search giant’s focus on AI. The technology has become a key element of the company’s strategy， and Ng’s departure comes at a time when Baidu is determined to double down on its AI efforts.
Baidu has quickly reaffirmed its commitment to AI， naming Wang Haifeng， an expert in natural-language processing， as Ng’s replacement. CEO Robin Li has recently written numerous opinion pieces in Chinese newspapers about the importance of developing AI expertise. “The disruption in traditional business models， industrial chain， and value chain brought about by artificial intelligence will cause fundamental changes in the global economy，”Li wrote in People’s D a i l y ， a g o v e r n m e n t flagship newspaper， in early March. And also in March， China’s National Development and Reform Commission approved Baidu as the leader of the new National Engineering Lab of Deep Learning Technology and Application， reflecting China’s top-level government dedication to AI development.
In the lab， Baidu will collaborate with top Chinese universities including Tsinghua， exploring a variety of different areas of AI research including visual perception， speech recognition， and human-machine interaction.
Shengjin Wang， a professor at Tsinghua University who studies computer vision and image recognition， and who also contributes to the research done at the national lab， says Baidu has been a leading player in AI， and there is reason to believe that it will be able to maintain its lead.“Personally I think it’s a pity，” Wang says about Ng’s departure. But he stresses that Ng has built a very solid AI team with over 1，300 researchers at Baidu already， giving the company an advantage in human resources. Baidu also has a competitive edge because of the large amounts of data it has collected from its search engine business， he says. Additionally， he thinks the newly formed Intelligent Driving Group， which focuses on automated driving research， shows the company is heading in a good direction.
He predicts Baidu will be able to adjust its AI resources and strategy after Ng’s departure.Some of that adjustment is already under way. A couple of days ago， Baidu launched its first overseas campus recruitment campaign， seeking AI talent for its Beijing headquarters in top American universities such as Carnegie Mellon and Columbia. A Baidu spokesperson says the company particularly hopes to attract Chinese students in American universities. Wang says that China can provide an interesting challenge for young researchers with its cities’ difficult-to-navigate road conditions and many different dialects， forcing Chinese AI researchers to find innovative ways to produce useful applications. Ng himself praised Baidu’s deep bench in AI in his departing blog post. He included Lu Qi， an AI expert from Microsoft whom Baidu brought in as chief operating officer earlier this year. But outside of Baidu， some see Ng’s departure as proof of a systemic problem at the company， one more example of its inability to hold on to important talent. Tong Zhang， who formerly led Baidu’s Big Data Lab， recently joined Tencent， a Baidu competitor， to oversee its AI research. Another senior manager， Jin Wang， who used to head the company’s automated driving unit， has announced that he would resign and start his own company. Last year， a key researcher at Baidu’s humancomputer interaction team， Jiawei Gu， left the company as well. While the company’s AI Group has had successes with face-recognition technology and AI-powered lending， Baidu’s other attempts at broadening its business beyond search have largely fallen short. Its share of search ad revenue has dropped in recent years， and as it faces new competitors like Toutiao， which uses AI to aggregate news stories， the company really needs its bet on artificial intelligence to pay off.
“The Internet is the appetizer， and the main course is artificial intelligence”， one day after Baidu chairman Li Yanhong’s metaphor， his main course chef Ng resigned. This chief scientist in Baidu announced his departure from Baidu by posting a blog on Medium （an English self-media platform） and Weibo （Chinese version Twitter）， together with an open letter on Twitter and other personal social media. It means a new page is open for his position in AI area. He will focus on the cultivation and exploration of new areas， which is what he really likes and is strong at. What is worth mention is that Ng sighed with emotion more than one when taking and interview earlier. He said that the biggest problem of AI is the shortage of talented people losses the rich opportunities.
For Baidu Ng’s leave not only means the company lost one top talent in AT are， but also a second human resource turbulence went through by Baidu’s senior management. In this January， when former Microsoft global executive deputy president Lu Qi joined Baidu as president and COO， the AI tech team led by chief scientist， selfdriving business department led by senior deputy president， president of searching engine company， president of tech system and emerging business community， and financial business department lead by senior deputy president all report their work to Lu Qi. Within fewer than three months， among the above mentioned Baidu mid and senior management， Wang Jin from the driverless vehicle business department took a leave out and Ng said he would consider a resignation soon. Baidu will adjust its AI business after his leave， including deep integration of NLP，KG，IDL，SPEECH，BIGDATA and other central tech. These all make up the Baidu AI tech platform system （AIG）. Wang Haifeng， the deputy president， was appointed as the head of AIG. Nowadays， AI talents are frequently poached by tech giants， how can Baidu retain talents through management and incentive mechanism is a question concerned by the outside. Next how is Baidu AI transform from science research and basic tech founding from jointly power generation of R&D and product applications？ Yu Kai， the then Baidu deep study center’s standing deputy dean， once had a long conversation with Ng beside the swimming pool of Palo Alto Sheraton hotel in America. He highly recommanded this authoritative scholar in AI and robotics study to join Baidu. Before this conversation Ng had cooperated with Google in deep research and led Google Brain project created by the Google X lab. The project took the company huge amount of investment. He pushed the internal deep study algorithm in Google and he’s also called father of Google Brain by the outside world. Ng officially joined Baidu in May， 2014， and was the chief scientist and leader of Baidu research institute， especially being charge of Baidu Brain project.
One resource familiar with Baidu said that AI before Baidu was mainly about scientific research and tech structure building， since the tech and market are immature. Since the participation of Ng， the company built up an AI deep research platform within three years， with employees growing from zero to more than 1000. Ng was also seen as the remarkable figure of Baidu in Silicon Valley and AI. Baidu once held a BIG Speech event in Silicon Valley in the early 2015， when Ng had joined Baidu for over half a year. Several artisans presented that day said， most American artisans might not know what is Baidu but they had all heard of NG， since many of them had taken lessons on his public classes of Coursera project. His Stanford university background is a persuading calling card for Baidu to attract talents from Silicon Valley. Technical staff within Baidu describes Ng as a person with unique ideas and outstanding leadership in this field.
In an open letter， Ng introduced himself as the chief structure engineer of Baidu AI strategy and was proud to lead the company to the rapid development and rise in AI. According to his description， Baidu AI team has over 1300 members including 300 Baidu research institute members. Ai tech daily serve over hundreds of millions of users， comprehensively support searching， advertising， mapping， take-out delivery， security and consuming financial and other Baidu current services. Furthermore， it has cultivated various of AI driven new business and tech such as driverless vehicle projects， DuerOS voice interaction and face recognition. As one of the preachers of AI， Ng showed up in Baidu world conference and brought developers three presents： all voice supported input keyboard， an open Baidu deep research study platform and an open Baidu Brain platform. It easy to see that Baidu wishes to play the role of tech output in AI area， and be the provider of tech supports to enterprises and developers. When taking an interview， Ng said， “AI is just like a new kind of electricity， which made great changes to agriculture， refrigerators and communications a century ago. AI in modern life will also change a lot industries， such as self-driving vehicles， financing and medicals.” However， Baidu also see a great loss of talents when it is taking in a big group of AI top talents and making progress in AI business. The former Baidu IDL deputy dean who once highly recommended Ng to join Baidu had resigned and created a horizon line robotics start-up.
In recent years， Baidu’s young talents， top scientists， and T10 engineers have been poached by other companies or built up their own AI start-ups. In last year’s Yabuli Summer Forum Summit， Yu Kai told reporters that AI had past its initial cultivation period and entered the best period for entrepreneurship. Large companies like BAT are hard to hold AI talents due to the mass and complex structure， which will be rather boring in the future. Yu Kai added that he believed Ng’s leave was for another more exciting AI project. Chief scientist Dr. Lin Hui who has contacted Ng said， Ng made great contribution during his time in Baidu， especially put out profound tech in a simple way， which helps make it more popular. His departure has actually been predicted by us， said Dr. Lin. The transfer of Baidu’s leading figure from academic Ng to doer Lu Qi shows Baidu’s requirement of Ai has changed. Zhao Yong， CEO of Deeplight company， said Ng left Google with the expectation of having a larger influence in a Chinese company. Now he left Baidu for the platform didn’t fulfill his expectations and ambitions in AI area. The specific reasons are only known by internal figures. Ai structure in today’s Baidui is large and deep， which will not be affected by one person’s departure. About Ng’s next plan， Baidu hints that he will focus on what he like and is good at in America. Besides， according to rumors within the industry， he may join Drive.ai， an AI start-up created partly by his wife. This company mainly transfer normal cars into driverless vehicles by tools.
AI was the key word in Baidu’s development in the past few years. After the chef’s departure， where is the AI team made up over 1300 staff of Baidu heading for in the future？ In recent years， Baidu has established deep learning lab， Baidu big data lab， and AI testing lab in Silicon Valley. It build up AI lab in the past year， integrating voice supporting tech departments and AI product applications. Baidu has started to adjust AI business since departure of Ng， including NLP， KG， IDL， Speech and Big Data. Its internal central tech will be further integrated， making up Baidu AI tech platform system. The company has appointed Wang Haifeng， deputy chairmen of Baidu， as general person in charge for AI tech platform system， reporting to Lu Qi. The dean of Baidu research center will be taken by Lin Yuanqing. About half a month ago， Baidu announced through internal emails that the key application of driverless business in AI field will be worked out by AI driving special team made up by driverless business department， AI vehicle business department， Car Life service.
Ng is VP & Chief Scientist of Baidu； Co-Chairman and Co-Founder of Coursera； and an Adjunct Professor at Stanford University. In 2011 he led the development of Stanford University’s main MOOC （Massive Open Online Courses） platform and also taught an online Machine Learning class to over 100，000 students， leading to the founding of Coursera. Ng’s goal is to give everyone in the world access to a great education， for free. Today， Coursera partners with some of the top universities in the world to offer high quality online courses， and is the largest MOOC platform in the world. Ng also works on machine learning with an emphasis on deep learning. He founded and led the“Google Brain” project which developed massives c a l e d e e p l e a r n i n g algorithms. This resulted in the famous “Google cat”result， in which a massive neural network with 1 billion parameters learned from unlabeled YouTube videos to detect cats. More recently， he continues to work on deep learning and its applications to computer vision and speech， including s u c h a p p l i c a t i o n s a s autonomous driving.
Baidu has quickly reaffirmed its commitment to AI， naming Wang Haifeng， an expert in natural-language processing， as Ng’s replacement. CEO Robin Li has recently written numerous opinion pieces in Chinese newspapers about the importance of developing AI expertise. “The disruption in traditional business models， industrial chain， and value chain brought about by artificial intelligence will cause fundamental changes in the global economy，”Li wrote in People’s D a i l y ， a g o v e r n m e n t flagship newspaper， in early March. And also in March， China’s National Development and Reform Commission approved Baidu as the leader of the new National Engineering Lab of Deep Learning Technology and Application， reflecting China’s top-level government dedication to AI development.
In the lab， Baidu will collaborate with top Chinese universities including Tsinghua， exploring a variety of different areas of AI research including visual perception， speech recognition， and human-machine interaction.
Shengjin Wang， a professor at Tsinghua University who studies computer vision and image recognition， and who also contributes to the research done at the national lab， says Baidu has been a leading player in AI， and there is reason to believe that it will be able to maintain its lead.“Personally I think it’s a pity，” Wang says about Ng’s departure. But he stresses that Ng has built a very solid AI team with over 1，300 researchers at Baidu already， giving the company an advantage in human resources. Baidu also has a competitive edge because of the large amounts of data it has collected from its search engine business， he says. Additionally， he thinks the newly formed Intelligent Driving Group， which focuses on automated driving research， shows the company is heading in a good direction.
He predicts Baidu will be able to adjust its AI resources and strategy after Ng’s departure.Some of that adjustment is already under way. A couple of days ago， Baidu launched its first overseas campus recruitment campaign， seeking AI talent for its Beijing headquarters in top American universities such as Carnegie Mellon and Columbia. A Baidu spokesperson says the company particularly hopes to attract Chinese students in American universities. Wang says that China can provide an interesting challenge for young researchers with its cities’ difficult-to-navigate road conditions and many different dialects， forcing Chinese AI researchers to find innovative ways to produce useful applications. Ng himself praised Baidu’s deep bench in AI in his departing blog post. He included Lu Qi， an AI expert from Microsoft whom Baidu brought in as chief operating officer earlier this year. But outside of Baidu， some see Ng’s departure as proof of a systemic problem at the company， one more example of its inability to hold on to important talent. Tong Zhang， who formerly led Baidu’s Big Data Lab， recently joined Tencent， a Baidu competitor， to oversee its AI research. Another senior manager， Jin Wang， who used to head the company’s automated driving unit， has announced that he would resign and start his own company. Last year， a key researcher at Baidu’s humancomputer interaction team， Jiawei Gu， left the company as well. While the company’s AI Group has had successes with face-recognition technology and AI-powered lending， Baidu’s other attempts at broadening its business beyond search have largely fallen short. Its share of search ad revenue has dropped in recent years， and as it faces new competitors like Toutiao， which uses AI to aggregate news stories， the company really needs its bet on artificial intelligence to pay off.
“The Internet is the appetizer， and the main course is artificial intelligence”， one day after Baidu chairman Li Yanhong’s metaphor， his main course chef Ng resigned. This chief scientist in Baidu announced his departure from Baidu by posting a blog on Medium （an English self-media platform） and Weibo （Chinese version Twitter）， together with an open letter on Twitter and other personal social media. It means a new page is open for his position in AI area. He will focus on the cultivation and exploration of new areas， which is what he really likes and is strong at. What is worth mention is that Ng sighed with emotion more than one when taking and interview earlier. He said that the biggest problem of AI is the shortage of talented people losses the rich opportunities.
For Baidu Ng’s leave not only means the company lost one top talent in AT are， but also a second human resource turbulence went through by Baidu’s senior management. In this January， when former Microsoft global executive deputy president Lu Qi joined Baidu as president and COO， the AI tech team led by chief scientist， selfdriving business department led by senior deputy president， president of searching engine company， president of tech system and emerging business community， and financial business department lead by senior deputy president all report their work to Lu Qi. Within fewer than three months， among the above mentioned Baidu mid and senior management， Wang Jin from the driverless vehicle business department took a leave out and Ng said he would consider a resignation soon. Baidu will adjust its AI business after his leave， including deep integration of NLP，KG，IDL，SPEECH，BIGDATA and other central tech. These all make up the Baidu AI tech platform system （AIG）. Wang Haifeng， the deputy president， was appointed as the head of AIG. Nowadays， AI talents are frequently poached by tech giants， how can Baidu retain talents through management and incentive mechanism is a question concerned by the outside. Next how is Baidu AI transform from science research and basic tech founding from jointly power generation of R&D and product applications？ Yu Kai， the then Baidu deep study center’s standing deputy dean， once had a long conversation with Ng beside the swimming pool of Palo Alto Sheraton hotel in America. He highly recommanded this authoritative scholar in AI and robotics study to join Baidu. Before this conversation Ng had cooperated with Google in deep research and led Google Brain project created by the Google X lab. The project took the company huge amount of investment. He pushed the internal deep study algorithm in Google and he’s also called father of Google Brain by the outside world. Ng officially joined Baidu in May， 2014， and was the chief scientist and leader of Baidu research institute， especially being charge of Baidu Brain project.
One resource familiar with Baidu said that AI before Baidu was mainly about scientific research and tech structure building， since the tech and market are immature. Since the participation of Ng， the company built up an AI deep research platform within three years， with employees growing from zero to more than 1000. Ng was also seen as the remarkable figure of Baidu in Silicon Valley and AI. Baidu once held a BIG Speech event in Silicon Valley in the early 2015， when Ng had joined Baidu for over half a year. Several artisans presented that day said， most American artisans might not know what is Baidu but they had all heard of NG， since many of them had taken lessons on his public classes of Coursera project. His Stanford university background is a persuading calling card for Baidu to attract talents from Silicon Valley. Technical staff within Baidu describes Ng as a person with unique ideas and outstanding leadership in this field.
In an open letter， Ng introduced himself as the chief structure engineer of Baidu AI strategy and was proud to lead the company to the rapid development and rise in AI. According to his description， Baidu AI team has over 1300 members including 300 Baidu research institute members. Ai tech daily serve over hundreds of millions of users， comprehensively support searching， advertising， mapping， take-out delivery， security and consuming financial and other Baidu current services. Furthermore， it has cultivated various of AI driven new business and tech such as driverless vehicle projects， DuerOS voice interaction and face recognition. As one of the preachers of AI， Ng showed up in Baidu world conference and brought developers three presents： all voice supported input keyboard， an open Baidu deep research study platform and an open Baidu Brain platform. It easy to see that Baidu wishes to play the role of tech output in AI area， and be the provider of tech supports to enterprises and developers. When taking an interview， Ng said， “AI is just like a new kind of electricity， which made great changes to agriculture， refrigerators and communications a century ago. AI in modern life will also change a lot industries， such as self-driving vehicles， financing and medicals.” However， Baidu also see a great loss of talents when it is taking in a big group of AI top talents and making progress in AI business. The former Baidu IDL deputy dean who once highly recommended Ng to join Baidu had resigned and created a horizon line robotics start-up.
In recent years， Baidu’s young talents， top scientists， and T10 engineers have been poached by other companies or built up their own AI start-ups. In last year’s Yabuli Summer Forum Summit， Yu Kai told reporters that AI had past its initial cultivation period and entered the best period for entrepreneurship. Large companies like BAT are hard to hold AI talents due to the mass and complex structure， which will be rather boring in the future. Yu Kai added that he believed Ng’s leave was for another more exciting AI project. Chief scientist Dr. Lin Hui who has contacted Ng said， Ng made great contribution during his time in Baidu， especially put out profound tech in a simple way， which helps make it more popular. His departure has actually been predicted by us， said Dr. Lin. The transfer of Baidu’s leading figure from academic Ng to doer Lu Qi shows Baidu’s requirement of Ai has changed. Zhao Yong， CEO of Deeplight company， said Ng left Google with the expectation of having a larger influence in a Chinese company. Now he left Baidu for the platform didn’t fulfill his expectations and ambitions in AI area. The specific reasons are only known by internal figures. Ai structure in today’s Baidui is large and deep， which will not be affected by one person’s departure. About Ng’s next plan， Baidu hints that he will focus on what he like and is good at in America. Besides， according to rumors within the industry， he may join Drive.ai， an AI start-up created partly by his wife. This company mainly transfer normal cars into driverless vehicles by tools.
AI was the key word in Baidu’s development in the past few years. After the chef’s departure， where is the AI team made up over 1300 staff of Baidu heading for in the future？ In recent years， Baidu has established deep learning lab， Baidu big data lab， and AI testing lab in Silicon Valley. It build up AI lab in the past year， integrating voice supporting tech departments and AI product applications. Baidu has started to adjust AI business since departure of Ng， including NLP， KG， IDL， Speech and Big Data. Its internal central tech will be further integrated， making up Baidu AI tech platform system. The company has appointed Wang Haifeng， deputy chairmen of Baidu， as general person in charge for AI tech platform system， reporting to Lu Qi. The dean of Baidu research center will be taken by Lin Yuanqing. About half a month ago， Baidu announced through internal emails that the key application of driverless business in AI field will be worked out by AI driving special team made up by driverless business department， AI vehicle business department， Car Life service.
Ng is VP & Chief Scientist of Baidu； Co-Chairman and Co-Founder of Coursera； and an Adjunct Professor at Stanford University. In 2011 he led the development of Stanford University’s main MOOC （Massive Open Online Courses） platform and also taught an online Machine Learning class to over 100，000 students， leading to the founding of Coursera. Ng’s goal is to give everyone in the world access to a great education， for free. Today， Coursera partners with some of the top universities in the world to offer high quality online courses， and is the largest MOOC platform in the world. Ng also works on machine learning with an emphasis on deep learning. He founded and led the“Google Brain” project which developed massives c a l e d e e p l e a r n i n g algorithms. This resulted in the famous “Google cat”result， in which a massive neural network with 1 billion parameters learned from unlabeled YouTube videos to detect cats. More recently， he continues to work on deep learning and its applications to computer vision and speech， including s u c h a p p l i c a t i o n s a s autonomous driving.