Remove capital and hot spots. What should the AI industry do next
1. Baidu ranked 7th "Ai" in the 2016 annual hot search list, and ranked 3rd in the top ten technology events, "alphago wins Li Shishi"
2. 1458 AI start-ups received financing in 2016, with a financing scale of $8.9 billion
3. Microsoft AI research team is composed of more than 1000 researchers and more than 5000 engineers
there are hot spots, capital and technology, which is the trend of the AI industry in 2016
whether it's the Google deepmind alphago robot war against Li Shishi at the beginning of the year, or the homicide caused by Tesla's driverless driving, and then the robot broke the glass and injured the audience at the exhibition, artificial intelligence has become a topic we must talk about every day
when talking about the confrontation between alphago and Li Shishi, Pichai, CEO of Google, said, "the significance of this victory will create a new situation for us, but mankind is always the ultimate winner." He believes that human beings will create artificial intelligence that can help us in various industries, from life, work, learning to travel, and even treating cancer and predicting the weather
according to the tracking survey data of ventures scanner, an investment research institution, as of November 2016, a total of 1485 AI entrepreneurial functions have received financing, with a total financing amount of more than $8.9 billion, the highest in history. Among these enterprises that have received investment, 44% are artificial intelligence in the field of machine learning. In the whole AI industry, 35% of enterprises choose the machine learning route as the industry leading direction. It can be seen that investment institutions are betting on the artificial intelligence industry
according to the financial times, Microsoft spent $12billion on research and development in 2015, a third higher than Google, and is the largest enterprise in the world. One third of Microsoft researchers are engaged in research on AI related projects. In October last year, Microsoft transferred Shen Xiangyang, a top AI expert, to take charge of the world's largest ai r & D team, which is composed of more than 1000 researchers and more than 5000 engineers. Public opinion believes that Microsoft's move is to use artificial intelligence to reverse the passive situation in the post PC era
it is such an industry that makes people look promising, but it is full of impetuous atmosphere everywhere. Rui Yong, vice president of Microsoft Research Asia, believes that the outside world does not view AI rationally, and even because alphago's victory in go is over interpreted as "AI has defeated mankind". The public lacks a correct understanding of artificial intelligence, and cannot well understand what artificial intelligence is and what role it can play in human society
this is really an era of popular interest.
on Sunday, Li Feifei, associate professor of computer science at Stanford University and director of Artificial Intelligence Laboratory, was invited to participate in GIF 17 activities. As the first speaker of the event, she obviously received "preferential treatment" from the media
what do you think of alphago
how long do you think it will take for the scenes in Western world to come true
what do you think of the TV program "the strongest brain"
do you think artificial intelligence will destroy mankind
"to be honest, I don't know alphago very well"
"I haven't seen western world"
"I haven't seen the strongest brain"
"I haven't read the book out of control"
friends, stop first! Do you know what kind of person li Feifei is? That's why he asked such a "retarded" question
Li Feifei is not Kevin Kelly. She is a serious academic researcher. At that time, Feifei li of Stanford University and Kai Li of Princeton University jointly studied visual learning and jointly created the Imagenet project. When we mention Imagenet, we should mention WordNet, which is a dictionary grouped according to the semantics of words. It is a tool that all AI scholars must learn and use. Imagenet can be said to be the annotation version of WordNet. It provides image annotations for 110000 synonym sets of WordNet. Under each image annotation, there are an average of 1000 pictures with similar content to explain what this synonym set means. All pictures have been manually reviewed and annotated. So far, more than 21800 synonym sets have been annotated with pictures, and there are 14.2 million pictures in the whole Imagenet
AI can recognize objects more efficiently and accurately and understand the meaning of images by using such databases and retrieval data. This is a project that only scientific researchers have the patience to do, but the results can indeed benefit future generations. WordNet, approved in 1985, is still used by researchers and still has a place in the field of artificial intelligence. Watson robot of IBM also uses this database. It is conceivable that when the image annotation of Imagenet is completed, it will make a fundamental breakthrough in the artificial intelligence image recognition industry
however, it was such a 40 year old woman who was asked by various media during the event. This kind of question needs to be answered by media critics, which is not worth answering. However, Li Feifei's coming to GIF 17 to deliver a speech can be said to be shameless for the media. Perhaps the media just want to make their article title appear "Professor Stanford", "Li Feifei" and "alphago" to attract the attention of readers. If you don't understand the industry or can only express some simple views, then don't occupy other people's time. Moreover, the online live broadcast platform asked questions about AI, constellations and boyfriends in the interview of female anchors. This is really an era when the whole people rub hot spots
according to Li Feifei, although Beijing is very cold, AI is really hot. But she stressed that AI is now popular, and the more she hopes to have someone calm down and do research. Similarly, in the face of such hot speculation, Rui Yong also has a similar view. He believes that it will take some time for AI technology to change daily life, and that it will take about 500 years to realize real AI. "I don't object if you ask me to add another 0 later."
is the AI you see in the media really the real trend of AI? This is not the case. It makes people misunderstand and and even start to feel panic about the development frontier of this industry, which is the inescapable responsibility of the media and academia
academic summits that have never made headlines
at geekpark GIF, techcrunch disrupt, CES, WWDC, 36kr wise and other large-scale events, I saw the presence of AI industry giants such as Li Feifei, ruiyong, Wu Enda, Ben Goertzel, etc. in the low-cost sales of Jinan gold testing machine, I have to say that through media packaging and headlines, academic bulls have become "red men" in the eyes of the public, and their views represent the industry attitude Their ideas represent the direction of the industry
but in fact, academic conferences such as CCAI, AI world, IJCAI, AAAI, IEEE and so on can really lead the industry frontier and determine the development direction of artificial intelligence. At these conferences, what Daniel said and today's discussion are completely different. Here is their real point of view
AAAI 2017 conference will be held in San Francisco after the lunar new year. The full English name of AAAI is the Association for the advance of artistic intelligence, or "American Association of artificial intelligence", which is one of the major academic organizations in the field of artificial intelligence. AAAI was founded in 1979. This organization aims to promote the research of artificial intelligence and promote the orderly use of artificial intelligence. In addition, this organization also helps the public to increase their understanding of AI and improve the teaching level of AI practitioners
best paper in recent years:
2015 "from nonnegative to general operator cost division"
2014 "correcting selection bias in causal inference and statistical inference"
2013 HC search: learning heuristic algorithms and cost functions for structured inference smile: disruptive multi case learning
2012 "learning SVM classifier through indefinite kernel", "basic it is a document summary of data reconstruction with a diamond cone with a vertex angle of 120 ° or a steel ball with a diameter of 1.59 and 3.18mm"
these are the contents you can't understand, which are constantly pushing the AI industry forward. On GIF 17, if someone can ask Li Feifei about any of the above questions, I believe she will be much happierIn May this year, Professor zhouzhihua of Nanjing University, who was selected as an AAAI academician, said on his microblog that because the original time of the AAAI 2017 conference conflicts with the Chinese Lunar New Year, through the efforts of many Chinese scholars, AAAI 2017 decided to reschedule to the end of the Spring Festival. "This is the first time that AAAI has considered the Chinese Spring Festival and made a major move to change the time and place. It is believed that as the influence of Chinese scholars increases, more top-level meetings will be considered."
in the list of academicians and papers of AAAI, there are more and more Chinese and Chinese names. Since the "document abstract based on data reconstruction" published by Zhejiang University of China was rated as the best paper of the year in 2012, the names of Chinese AI researchers have increasingly appeared in the list of accepted papers, and the influence of Chinese AI scientists is also increasing
aaai received a total of 2590 submissions in 2017, and 638 papers were hired, with an employment rate of 24.6%. Compared with 2132 papers submitted last year, 549 papers were hired, with an employment rate of 26%. The proportion decreased, but the number of papers increased. Among them, the number of papers with Chinese authors' names is increasing, close to 1/3
as for the papers of Chinese authors, in addition to seeing the contributions of papers from domestic first-class research institutes such as Tsinghua University, Peking University, Nanjing University and National University of defense technology, we should pay more attention to the contributions of enterprise researchers today. Baidu, Tencent, Ctrip, Ipin and other domestic AI research teams have published many papers on AAAI 2017
the "Baidu talent think tank tic" of Baidu AI team uses AI to intelligently manage human resources. Baidu's complete intelligent talent management system is the world's first. Before that, both Google and Facebook had conducted small-scale tests inside the company, while Baidu was the only company with 10000 people. In addition, the team of Wang Jin, vice president of Baidu and general manager of the autonomous driving business division, was also selected
papers written by Ctrip scientific research team based on the self-developed universal recommendation system have also been included by AAAI. The system uses the automatic coding technology of deep neural network to improve the recommendation effect of tourism products, improve the scene conversion rate by 13 times, and improve users' purchase and actual experience
based on topic model, Ipin AI team uses deep learning technology to model document semantics, which increases the semantic correlation between sentences and makes up for the defect that topic model only has word frequency relationship. They applied this technology to the socio-economic map and developed a system similar to Watson robot, which can intelligently identify user questions, answer user questions and help users make efficient decisions
in addition, there are Yan Shuicheng, President of 360 Artificial Intelligence Research Institute, Li Hang, director of Huawei Noah Ark laboratory, and Tencent Youtu laboratory
Copyright © 2011 JIN SHI