The picture shows Dr. Kai-Fu Lee reading photos
Text / Bi Xiao Nan
In 1988, 27-year-old Kai-fu Lee was reading a computer doctor at Carnegie Mellon University. He was involved in the development of a black-and-white chess game called "Othello" and defeated the then-current Othello champion in the fall. This is the very first time the machine has defeated the human championship.
Like most later AI systems that made breakthroughs in chess and sports (AlphaGo), Othello also used statistical-based machine learning algorithms. This experience buried a seed in the heart of Kai-fu Lee, which also made him one of the earliest Carnegie scholars to start studying statistical methods.
In developing "Othello", Kai-fu Lee realized the enormous potential value of this technology. He brought this technology into his speech recognition project led by his mentor Professor Rudi and succeeded in raising the machine's speech recognition rate to 96%. This became the basis for the study of voice technology by Silicon Valley giants such as Apple and Microsoft after that, and Professor Ruidi also won the 1994 Turing Prize for this research.
After leaving academia, Kai-fu Lee has served in well-known companies such as Apple, Microsoft and Google. During his tenure as Microsoft's worldwide vice president, he founded Microsoft Asia Research Institute and served as the first dean (also called Microsoft China Academy). Zhang Yaqin, Shen Xiang Yang, Wang Jian, these are now "light and heat radiation to most of the Chinese science and technology circle" name, are Microsoft Asia Research Institute early members.
It is worth mentioning that the current president of Microsoft Research Asia Hong Xiaowen and Kai-fu Lee out of the door, the two also worked in Professor Rui voice recognition programs on cooperation. Wu Jun, author of "The Beauty of Mathematics," believes that Professor Ruidi's achievements can not be separated from the outstanding work of Kai-Fu Lee and Hong Xiaowen.
After leaving Google in 2009, Kai-fu Lee founded Innovative Works and served as chief executive officer, focusing on innovative investments in technology and cutting-edge technology trends. Kuangshen, Yu-potential, the fourth paradigm, the most potential of these artificial intelligence startups in China are behind the innovative factory presence.
From academia to the business community, from the United States to China, from serving large enterprises to focusing on start-ups, Kai-fu Lee has undergone several successful transformations. In a sense, these changes are exactly the challenges that current artificial intelligence entrepreneurs and startups are experiencing. Kai-fu Lee's experience for today's artificial intelligence entrepreneurial wave has a positive guiding significance.
To this end, "China's Road to Artificial Intelligence" has deliberately chosen Li Kaifu, founder and CEO of Innovation Works, as our next dialogue object and discussed with him how artificial intelligence may bring to Chinese society and even to human society in various fields Change, well-being and confusion.
If Robin Li and Baidu are among the firm executors of artificial intelligence in China's Internet companies, Kai-Fu Lee and Innovation Works are among the most pretentious evangelists of artificial intelligence in the field of venture capital. Industry, research and education are the three key words of this dialogue.
Kai-fu Lee believes that data, computing, algorithms, is the development of artificial intelligence, the three major elements. The three chase each other, advance each other. In this wave of artificial intelligence, the rapid development of data and computing power has made artificial intelligence reach a critical point that can be popularized and applied on a large scale.
As a model from the transformation of technical experts to business managers - Kai-fu Lee said that there is an essential difference between doing research and doing business, both in the purpose of there is a huge difference. What scientists are pursuing is to do what their predecessors did not do, and entrepreneurs are pursuing the needs of users and the market. Whether a scientist can successfully transform itself into an entrepreneur mainly depends on whether he can make the transition from pursuing technological innovation to pursuing a technical and practical mode of thinking.
Kai-fu Lee also believes that scientists to entrepreneurs in the transformation of the road is hard and tough. In the future, the proportion of scientists in AI CEOs will fall, while the proportion of people with 2B sales experience will rise. This is because "AI itself is a 2B business"; and, in general, a company doing 2B Sales must have a CEO as the largest Salesman. & rdquo;
Kai-fu Lee admits that there is still a gap between the Chinese research in the field of artificial intelligence and the world's top level. But he believes that China has the largest and youngest AI researchers in the world, as well as a wealth of data and friendly policies that will make China an AI power.
Kai-fu Lee think we should make the property industry, research and scientific research. Let those big academicians continue to do new things and let the Chinese industry maximize the value of what they know from 1% to 50%. The latter "is what Chinese companies are good at" rdquo ;.
Kai-fu Lee said that programmatic, repetitive, relying on memory and contact can grasp the skills will be the most valuable skills can be completed by the machine; the other hand, those who best reflect the overall quality of human skills is artificial The Most Valuable and Intelligent Learning Skills for the Smart Era. Artificial intelligence on the substantial improvement of production efficiency, partial replacement of human labor, fundamental changes in lifestyles, inevitably touch all aspects of human life such as social, economic, political and other aspects.
The following is the dialogue Record:
Bi Xiao Nan:Before that, I had interviewed a lot of entrepreneurs in the field of AI. One of the most important features is that most of them are scientists or scholars who have changed their mindset. The thinking characteristics of their technical talents are still obvious. However, in this wave, who can take the lead in turning a science and technology talent into a leader-level entrepreneur may be a very important barrier to their mutual competition. How does this type of technology-based talent transform into an entrepreneur, and what barriers do they need to break through to themselves or to this group of people?
Kai-fu Lee:In fact, there are many places to turn. The biggest point is that scientists from doctoral study to write essay to do the whole process, the pursuit of one thing, is to do things not done before the predecessors, the greater the better the better, the newer the better, but not the pursuit other.
However, when we do business or do the investment or do the product, the pursuit must be to meet the needs of users and the market. When your ultimate goal is such a huge shift, a scientist can not very successful in the conversion of their own role, to see if he can put aside the pursuit of the previous goal, and to accept the new goal.
Because these two objectives are in fact often repulsive, of course, occasionally can be combined, but most of the time is repulsive. Because you may want to be a great technology, a great product, and a market success because of the directionality possible. But in fact every decision you make every day is whether I should give up this function or not, and I want to copy some of my competitors Function, or I want to grab your market share, earn money. These are very cruel realities.
Bi Xiao Nan:Almost two types of people here, a class of people means that some people find that I really can only do technical personnel, he will find partners, team combat; the other is that he himself can achieve the transformation. What do you think AI's entrepreneurs need more different characteristics than other entrepreneurs do?
Kai-fu Lee:I think from the computer technology, AI is relatively close to science. Computer is basically a project, in the field of engineering which used to do the system, the operating system, the database, safe, they are actually engineers thinking or thinking than scientists, so their transition is relatively easy.
AI scientists really need to understand mathematics well and be able to invent new algorithms that require the brightest brain in the world to collide and compete. This group of people doing AI, as well as those who do computer theory, generally do not become entrepreneurs. So, I'm going to hold a question mark for the transformation of AI scientists into entrepreneurs.
Of course, you would say that today we are not a lot of successful transition, it flew from Sida to Sogou, then to Face ++ and other computer vision companies. This is true. That is because you do not want to be the CEO who makes up 99% of the company's value. So these people must go forward and constantly correct their thinking. The transformation of these people is indeed very successful, but this transformation is also very hard and tough. So, I would predict the next call or two dials AI company, its CEO's proportion of scientists will decline.
Of course, the future of AI talent is still very critical, they will be an important co-founder, there will be a lot of shares. Speaking in the industry, maybe they continue to say. As a scientist, I think these are good. But AI is also entering a new phase, because AI itself is a 2B business. It should be sold to the business, not sold to the user, so an AI leader would especially need a person who can do 2B sales. In general 2B Sales (sales) company, you must CEO as the largest Sales (sales).
Bi Xiao Nan:There are many voices in the industry today that actually we are still less solid in basic theoretical research and our own knowledge creation. However, it seems that the investment circle and entrepreneurship now do not mention the AI as if they were outdated. Now the development of artificial intelligence in China is not there a hurry attitude?
Kai-fu Lee:I believe that AI will have a huge impact on humankind for the next 10 or 20 years. Today there are a large number of young AI researchers in China, supported by a large amount of data and a very friendly policy. I think China will become an AI power.
However, I think some people may misunderstand certain aspects and think that these technologies can solve all the problems in the future. This is actually not enough. Our research still needs to be strengthened. In this research section, the United States, Britain and Canada are far ahead of the world. Therefore, there is still a gap between China and the scientific research here, especially those highly intelligent and experienced scientists who need time to make up.
As for whether there is any bubble in this industry, will it be too hot? I think there is actually no industry. Because the value of AI is just beginning to be realized, if it can do 50% of the human task or even make it than we do Better, that bonus is far from seeing, 1% may not have seen.
My idea is to make the industry belong to the industry, research and scientific research. Let those academic buffalo continue to do new things and let our industry maximize the value of what is known from 1% to 50%. This is a very good thing for Chinese companies.
Bi Xiao Nan:So in fact we have completely deviated from a range of amateur markets, into a professional state, but still far from the top there is a certain distance.
Kai-fu Lee: Yes, but I will not be too worried. Because of this top experts in the above discussion, AI scientists are very open. They came up with new ideas that would be published in a timely manner and posted online without going through the one to two year approval cycle. And these latest ideas, will be passed quickly online. For example, the recent Alpha Zero, the result came out, a week later papers on the line, and a few weeks later, a lot of imitation of its open source program has also been put online.
Of course, there are exceptions, if it is a big company, the company may limit you from publishing. For example, you made a special breakthrough in Google, Facebook, Microsoft or in BAT, whether this will be published or not, and you have to see the company's approval.
Bi Xiao Nan:We have been talking about AI development, artificial intelligence development can bring unlimited imagination, science and technology can reach an unlimited ceiling. But I actually read your essay recently. I published a column in the Financial Times. I do not know if I'm not sure what I'm going to do, but the point is this. You mentioned the passage that artificial intelligence should replace it in at least five to ten years Customer service these basic work.
Kai-fu Lee:First of all, let me say, AI is not a panacea, it is in a single area, which can make some judgments and decisions based on big data. So, for example, telemarketing, customer service, some of these more repetitive assistant jobs, the parts of those people's jobs are rapidly being overtaken by machines. There are also some blue-collar jobs, such as inspectors and pipeline simpler work, will soon be replaced by the machine. Over the next 10 to 20 years, the proportion of this replacement will be growing.
In the next 10 to 15 years, I think we will do all kinds of tasks in our work today. Fifty percent of the quest machines can do as good or better than others, but this does not mean that 50% of the people must be laid off. Maybe 30% of my work has been replaced, but I can make up another, AI can be used as my tool.
Bi Xiao Nan:Innovation Workshop has a lot of attention and enthusiasm for this field of artificial intelligence. What is your layout for the AI field?
Kai-fu Lee:Because I started as an artificial intelligence, I started to touch the field of artificial intelligence 37 years ago. I've had contact with several companies in the past. Apple, Microsoft, Google. What I see today is that the current amount of data and the speed of the machine has finally reached a state where artificial intelligence can be popularized.
Of course, there are algorithms, data, computing power, algorithms, these three are advancing with each other. The era of artificial intelligence has finally arrived, we must be able to work hard in the innovation workshop, so that it can produce value as soon as possible.
As an investment institution, we also hope to get a good return from it. As an early participant in artificial intelligence, I also have a group of investors and engineers who love artificial intelligence all around me. We all want to do some public-benefit work to make artificial intelligence and the popularization of relevant knowledge more extensive .
We have set up artificial intelligence engineering institute to help the traditional industries to do the transformation or promotion of + AI. Because AI is not the same as the Internet, AI really has the potential to help each company, as long as you have the data may generate value, to make better choices.
In addition, we have also made some attempts at education. For example, we are also considering a bigger course for the training of trainees. In addition, we also invested in a number of artificial intelligence training projects. These areas may not necessarily be able to make money, but I think this is of great benefit to the popularization of artificial intelligence.
And we made an artificial intelligence contest, called AI Challenger (Global AI Challenge). Not only do not make money, but also give bonuses, is to allow students to participate in, subsidies for them to enter this area. Let those who are interested in artificial intelligence talented people, you can quickly enter, glowing, let the field do more.
In addition, I personally will do some speech at home and abroad to share. As an investor, I can speak up, able to be very honest AI can do what, what can not do, the opportunities and challenges are presented to let everyone know. Because I think knowledge is power, and when we all know it, it may be better to work together to solve the solution than one of the few who make artificial intelligence.
Bi Xiao Nan:If you use a very short paragraph, let you describe your greatest imagination and expectations of this era of artificial intelligence, what do you think? What should be the best AI in your mind?
Kai-fu Lee:I hope AI can replace our repetitive and boring work and free up more time so that human beings can really do what they are good at and love. This can have great value, let us love each other. Artificial Intelligence can even help humans find the purpose of existence, I think this is what it should do.
Interview with China's Leaders in AI Series By BI Xiao-nan, a well-known young scholar and media man, has an in-depth dialogue with leaders of Chinese artificial intelligence officials, industry, academia and research institutes, and explores how artificial intelligence can bring benefits to Chinese society in all fields, confused. See more High-end record talk show "China's Road to Artificial Intelligence" jointly produced by Caixin Video and Lan Ting Capital.