"Engineering thinking is very important and common in real-world work. A recent example is autonomous driving. Baidu founder, chairman and CEO Robin Li recently said in a speech at the Peking University New Engineering International Forum that engineering thinking and scientific thinking are very different, in the field of automatic driving, if you use scientific thinking to do it, it is a step to the sky, to achieve L5 and then expand the scale, commercialization, Waymo is this way; and if you use engineering thinking to do, it is step by step, first in some scenarios to achieve automatic driving, Tesla does this.
At the forum, Robin Li delivered a speech on "simplifying complexity, from scientific thinking to engineering thinking", talking about the construction of new engineering and talent training, he said that new engineering focuses on the engineering thinking of cultivating talents, simplifying complex problems, and combining reality in "compromise" step by step. At the same time, talents in the new engineering era must have the three major capabilities of innovation, cross-border and openness. He also suggested that it is necessary to strengthen the linkage between schools and enterprises, learn from each other and grow together.
In terms of industrial practice and technology landing, Robin Li said that on the basis of engineering thinking, Baidu has chosen the path of gradually changing the driving range of autonomous driving, starting from the fully unmanned driving of some roads, and gradually expanding the geographical scope, so as to move towards full automatic driving at the L5 level.
Baidu has launched an autonomous driving travel service platform "Radish Run", which operates in Beijing, Shanghai, Guangzhou, Cangzhou, Changsha and other places, and has received more than 400,000 passengers.
"I once said that when there is 1 yuan, we will invest in technology; there is 100 million, we will invest in technology; there are 10 billion, we will still invest in technology. The most cutting-edge technology wave can not wait, we must invest and layout 10 or 20 years in advance. Robin Li said.
Baidu has maintained a R&D investment intensity of more than 15% for many years, and its core R&D investment accounted for 21.4% of revenue last year.
In terms of technology trends to promote a new round of scientific and technological revolution and industrial change, Robin Li believes that there will be eight key technologies in the field of artificial intelligence in the next decade, which will achieve quantitative to qualitative changes, namely autonomous driving, digital city operation, machine translation, biocomputing, deep learning framework, knowledge management, AI chips and personal intelligent assistants.
Robin Li said: "The breakthrough of these eight technologies will extensively change people's production and lifestyle, which is something I firmly believe, and these eight technologies will also be the direction of Baidu's firm investment." ”
Li Yanhong said that china has the best soil to cultivate leading talents and strategic scientists, and stressed that in the next 5 years, Baidu will train 5 million AI talents for the society and continue to contribute to the construction of national strategic scientific and technological strength.
<h2>The following is the transcript of Robin Li's speech:</h2>
Good morning, everyone! It is a pleasure to return to my alma mater and participate in today's New Engineering International Forum.
The last time I came back was three years ago, to celebrate the 120th anniversary of Peking University. This year is the 30th anniversary of my graduation from Peking University, originally on August 7th, we had a party at the school for the 87th class, but later cancelled due to the epidemic, and when I was feeling sorry, I received an invitation from our new engineering forum, so I am particularly happy to have the opportunity to share my experience of doing big projects over the years for your reference.
When I was studying, Peking University did not have engineering, studying abroad was computer science, the same school also has a department of computer and electronic engineering, these two departments belong to the College of Science and the College of Engineering, so I always feel that science and engineering are still quite different, and then left the school to enter the industrial world, about the concept and ideas of engineering, are slowly learned in the work.
It can be said that the lack of engineering thinking is the shortcoming of our Peking University graduates. In 2000, I returned to China to start a business, and a large proportion of the early entrepreneurial team came from Peking University. When discussing work, you can often see the difference between engineering thinking and scientific thinking.
For example, the sorting of search results, I asked for 1 second to produce results (at that time the industry standard was almost 3 seconds), the team told me that I could not do it, users search for a word, such as "e-commerce", the web page with this word on the Internet may be millions, according to the relevance from the first to the 1 millionth sort, but also to support a large number of users concurrent access requests, is unrealistic, and we are a small company, has limited computing resources.
I said no, who told you to sort from the first to the 1 millionth, which user will turn to the 100,000th page of results to see? Your algorithm only needs to be able to find the most relevant 1000 results from countless web pages and sort by relevance, we just need to let the user turn 10 pages, really can't find what he wants, just another search term (query).
You see, this is the difference in thinking between doing engineering and doing science. Our classmates are trained by science, from the first ranking to the 1 millionth, I have worked for a few years, I know to compromise, only the top 1000 is good.
You compromise, hundreds of millions of people can enjoy the convenience of information at your fingertips every day, and if you don't compromise, you will always be a lab product.
Engineering thinking is very important and common in practical work. A recent example is autonomous driving.
To do it with scientific thinking is to ascend to the sky one step at a time, to achieve L5 and then expand the scale and commercialize, Waymo is this way. But many Turing Award winners say that L5 is too difficult to achieve in a few decades.
To do it with engineering thinking is to take it step by step, and first realize automatic driving in some scenarios. Tesla has taken such an approach, it first does L2, follow the car on the highway, change lanes, achieve automatic parking in the parking lot, and so on. This way is successful, the market is very accepting, the car is selling, every day countless people drive Tesla cars for them to collect all kinds of data for free, Tesla thus has an unparalleled super-large-scale autopilot-related data, so investors think that Tesla is more likely to make unmanned driving. At the same time, Tesla also made a lot of money by selling cars, forming a virtuous circle, and the automobile industry has also seen major changes that have not been encountered in a hundred years.
Tesla's path is an engineering logic of thinking, a compromise on a scientific problem, a gradual change in the degree of automation, from L2 hope to L3, L4, L5. But is this the only viable way to ramp?
We think there is also a way to gradient, that is, the gradient of the driving range of autonomous driving.
Now in any scenario of full unmanned driving can not be done, the country's 5 million kilometers of road, one step to the sky to achieve unmanned can not do, then can not start from the simplest 50,000 kilometers, 100,000 kilometers? Technically, in places where there are fewer people and cars mixed, in places where everyone obeys traffic rules more, where traffic lights are set up more reasonably, it is possible to achieve full unmanned driving, we can run up in these places first, continue to learn in the process of actual operation, continuously improve, and gradually expand the geographical scope of unmanned vehicle services, so that we can also go to L5.
Baidu has taken such a gradual approach, we have set up a special transportation company called Radish Run, to provide everyone with autonomous driving travel services. Today in Beijing Yizhuang, Guangzhou Huangpu, in Changsha, Cangzhou and many other places, ordinary users are able to hit the radish fast running unmanned car, and many times also free.
Today, our driverless cars have received more than 400,000 visitors.
Knowing how to turn difficult problems into simple problems, knowing compromises, knowing how to take steps, I think is the main difference between engineering thinking and scientific thinking. We have more scientific thinking in education and less engineering thinking, and I hope that Peking University's new engineering department can pay more attention to this aspect and cultivate more talents with engineering thinking.
At present, global scientific and technological innovation has entered an unprecedented period of intensive activity, a new round of scientific and technological revolution and industrial transformation is reconstructing the global innovation map, the application of technology is faster than people imagine, and the construction of new engineering sciences has also come into being in this context.
Last year, I made a judgment:
In the next decade, there will be eight key technologies in the field of artificial intelligence, which will achieve quantitative change to qualitative change, namely autonomous driving, digital city operation, machine translation, biocomputing, deep learning framework, knowledge management, AI chips and personal intelligent assistants.
The breakthrough of these eight technologies will widely change people's production and lifestyle, which is something I firmly believe, and these eight technologies will also be the direction of Baidu's firm investment.
The birth of technological achievements stems from Baidu's strong investment in technology for many years.
As a technology-based enterprise, Baidu has maintained more than 15% of its R&D investment intensity for many years, and its core R&D investment accounted for 21.4% of revenue last year.
I once said that when there is 1 yuan, we will invest in technology; there is 100 million, we will invest in technology; there are 10 billion, we will still invest in technology.
The most cutting-edge technology wave can not wait, we must invest and layout 10 or 20 years in advance.
For example, eight years ago, when Baidu decided to invest in self-driving technology, we thought it was the top project of artificial intelligence that would revolutionize human mobility and life.
Thanks to the long-term layout and innovation path choice, in the ranking of professional research institutions, Baidu's autonomous driving technology is now in the global leader camp and the only Chinese enterprise in this camp.
Innovative undertakings call for innovative talents. Focusing on the future industrial application, I also hope to strengthen the linkage between school and enterprise, and we can learn from each other and grow together. In this regard, Baidu has also done a lot of exploration and experimentation. Since 2015, we have supported the Ministry of Education's industry-university cooperation collaborative education project for seven consecutive years, aiming to train teachers for colleges and universities, and have invested more than 10 million teaching and research funds. We have trained more than 3,000 college teachers in more than 700 colleges and universities, and participated in the preparation of a series of artificial intelligence textbooks.
At that time, I went to the United States to study, and I liked the emerging discipline of artificial intelligence. But my professor commented that this is useless, the practicality is poor, the industry does not recognize it, and you can't find a job on this.
From the beginning of "learning artificial intelligence can not find a job", to today's "do not learn artificial intelligence can hardly find a job", technological development has put forward higher requirements for talent training. Looking a little forward, the layout is a little earlier, I think this is also the essence of the construction of new engineering.
The construction of new engineering talents focuses on cultivating teachers on the one hand, and on the other hand, on improving the core abilities of students. I believe that talents in the new engineering era must have three capabilities, that is, innovation, cross-border and openness.
The first key word, innovation.
More than 20 years ago, when I returned from the United States to found Baidu, I hoped that like Silicon Valley startups, I could recruit a group of engineers with about 5 years of work experience, but in the end, I almost all fresh graduates were recruited, because at that time, the Internet was a new thing, and no school could cultivate the right technical talents, let alone experienced engineers.
All the problems we faced at that time needed young people to find ways to solve themselves, and the process of solving problems was innovation. In the end, many of these employees grew into the core backbone of the company.
Today, more than 60% of Baidu employees are R&D personnel, and it is these innovative young people who have the dream of changing the world with technology, starting from a line of code to create today's Baidu.
The second keyword is cross-border.
In the fields of autonomous driving and AI chips, our business is eight or ten years in advance, and talent training is earlier than business layout. There is interdisciplinarity and cross-border behind the technology in these fields, so there is a greater need for talents with cross-border thinking. We have established and improved a compound talent training mechanism to encourage more talents to flow across fields and roles.
We firmly believe that only when the comprehensive ability of talents is improved, the enterprise will be more dynamic, and Baidu's technical strength can always be at the forefront of the industry. This is also what I want to say, young people should dare to cross borders, take the initiative to cross borders, and invest in the future.
The third key word, open.
You may have some experience that artificial intelligence technology is becoming more and more complex, but the application is becoming more and more simple, which is largely due to the open source and open concept advocated by the industry.
For example, Baidu's first self-developed, feature-rich industry-grade deep learning platform Flying Propeller is open source and open. Developers do not need to learn artificial intelligence theory, nor do they need to write algorithm code from scratch, they can efficiently design and develop applications, which also greatly accelerates the diversification and scale of artificial intelligence technology.
For example, a 12-year-old elementary school student, although he does not know how to program, successfully developed an AI program to detect whether the mask is worn as a standard by participating in the flying paddle live course. At present, the flying propeller platform brings together more than 3.6 million developers from all walks of life.
Today's China is undoubtedly the best era for science and technology practitioners. We have enough technical application scenarios, the world's most complete industrial chain supply chain, the continuous improvement of new infrastructure, and the best soil for cultivating leading talents and strategic scientists, and large-scale innovation results will inevitably emerge in China.
Not long ago, General Secretary Xi Jinping stressed at the Central Talent Work Conference that it is necessary to build a large number of first-class scientific and technological leading talents and innovation teams, give play to the role of national laboratories, national scientific research institutions, high-level research universities, and scientific and technological leading enterprises, and organize industry-university-research collaborative research around key national areas and key industries.
As Mr. Cai Yuanpei, the old president of Peking University, said, "Educators are not for the past, not for the present, but for the future." ”
Here, I would also like to solemnly promise you: in the next 5 years, Baidu will cultivate 5 million AI talents for the society, continue to contribute to the construction of national strategic scientific and technological strength, and live up to its mission and live up to the times.
Thank you!
Source: Thoughtful
![Book Recommendation 36: Robin Li's favorite five books: have the knowledge of a big coffee[fig]](data:image/gif;base64,R0lGODlhAQABAIAAAP///wAAACwAAAAAAQABAAACAkQBADs=)