The reporter | Zhou Xiang
There are always those people who brave hardships to travel thousands of miles to a strange country and dedicate their lives to a group of strangers.
In 1986, John Hopcroft, a professor of computer science at Cornell University, was awarded the Turing Prize for fundamental research in the design and analysis of algorithms and data structures.
Twenty-one years later, John Hopcroft was invited to Nanchang University and gave a lecture on “Future Directions for Computer Science”. The following year, John Hopcroft became an honorary professor at Beijing Institute of Technology.
If you look through Hopcroft’s biography, since 2007, “China” has become the most common word on his resume.
In early 2011, invited by The Administration of Foreign Experts Affairs, PRC, John Hopcroft began to devote himself to the construction of computer science and personnel training in Chinese universities. He has successively become honorary professor of chongqing University, Shanghai Jiaotong University, Huazhong University of Science and Technology, Jilin University, Suzhou University, Peking University, Harbin Institute of Technology and many other well-known universities in China.
Among these Chinese universities, Shanghai Jiao Tong University is most closely associated with Hopcroft. In mid-December 2011, Hopcroft flew to Shanghai to teach a course on Computer Science theory in the Information Age to students at the Zhiyuan School of Computer Science at Shanghai Jiaotong University. Since then, he has taught “Computer Science Theory in the Information Age” to Zhiyuan students in the winter semester and “Automata Theory” in the summer semester. In addition, Hopcroft’s only doctoral student is Yixuan Li, who graduated from Shanghai Jiao Tong University.
In the recent 19th “Computing in the 21st Century” international symposium co-held by Microsoft Research Asia and Hit University of Technology, AI Science and Technology Base had the honor to interview this old professor who is nearly 80 years old. During the interview, the professor spoke earnestly and pointed out the shortcomings of Chinese education, which is worth our warning.
The following is the transcript of the interview, compiled by AI Science and Technology Base camp:
The biggest problem in Chinese education
AI Tech Base: Why did you come to China to offer CS courses?
John Hopcroft: Early in one’s career, one may just want to establish a personal professional reputation. But as you get older, you want to do something that makes more people better, and the opportunity to teach in China is probably a once-in-a-lifetime opportunity.
If you help China change its education system, it will benefit tens of millions of students, and it will benefit millions of individuals, which will improve the economy and people’s living standards. And I want to do something to make the world a better place.
AI Tech Base: How has China’s education changed since you came to China?
John Hopcroft: Not much. But I believe a lot will change in the coming years. The Chinese government knows it needs to change, they don’t need foreign experts to tell them what’s wrong with Chinese education, because they already know, much more than foreign experts do, but they need help to change the situation, because a lot of it is cultural.
AI Tech Base: What do you think is the biggest problem in CS education in China?
John Hopcroft: The first is cultural, the objective criteria by which people judge success. University presidents and some government employees, whose terms tend to last five years and who understandably want to be promoted, may put pressure on junior faculty to push for more research funding and publish more papers.
AI Tech Camp: Do you think online education (MOOCs) are the future? What are the disadvantages?
John Hopcroft: Educators say that the most important quality of a good teacher is that he cares about students’ success, but MOOCs have lost that relationship. I think the main function of MOOCs is to get information out to the public.
Second, many industries require two weeks of training every five years or so to extend the validity of their certificates, so MOOCs are very useful in training, but not in education.
Advice for Chinese students and teachers
AI Tech Base: What do you think Chinese students need to pay attention to most?
John Hopcroft: First of all, the current Education system in China, for example at the basic education level, is aimed at passing exams to get into good universities. With long school hours and homework after school, students spend almost all their time “studying”.
When I was in school, I went to school at 9 am and 3 PM every day, and there was no homework, I could do whatever I wanted in my spare time after school. In fact, a lot of education actually takes place outside the classroom, such as interacting with other students, exploring the world and so on.
Asian students tend not to be that good at problem solving, and they need to tap into more creativity. China needs to reduce homework in basic education and spend more time on real education.
Q: What do you think Chinese teachers need to pay attention to most?
John Hopcroft: We need to realize that the overall standard of living in China is not high enough. People spend most of their time “living”. Usually both parents of students need to work, they need to spend their time to earn money. But I believe that once the standard of living in China improves, then teachers and parents may focus on the success of students’ education. Besides, teachers and parents should not tell students what they should do, but let students find their interest.
I was talking to a Chinese student who started a startup with an advisor who was both their teacher and their boss, which is never allowed in the US because it involves a conflict of interest. For example, when the student consulted the teacher, the latter might put the interests of the company first rather than the interests of the student.
Why did I become an academic? In fact, it is influenced by elementary and middle school teachers. And I want to pass that on to the next generation of talent. I believe that With the improvement of living standards, China will gradually change.
AI Tech Base: What advice do you have for students in other fields who want to switch to CS? (What programming and math skills are required)
John Hopcroft: In my opinion, CS will affect every discipline in the future. However, it is easier for biology majors to learn CS than for CS majors to learn biology.
AI Tech Base: Do you think age affects CSer’s career?
John Hopcroft: As a programmer, it’s exciting when you start out, but to be honest, after 20 years of writing code, you get bored, so I think it has an impact. But we are automating more and more jobs, and there will be less and less work that needs to be done by people in the future, and maybe people will start work at 25 and retire at 45, and society will still be able to produce for everyone.
The secret of success
AI Tech Base: What’s been the biggest change in CS over the decades?
John Hopcroft: There are a couple of things:
First of all, we are entering the information age, in the era of industrial revolution, we let the mechanical task automation, now we are intelligent task automation, such as driving a car, language translation, etc., some of the reasons is the emergence of a large amount of data available, especially in recent years, so we can use these data to train the computer to do what we haven’t been able to do.
Moreover, after 40 years of studying how to make computers more useful, computer science is now more focused on what computers can be used for, so computer science is beginning to play a role in the field of applications. Computer scientists started interacting with researchers in other fields, and that was a big change in CS.
AI Tech Base: What is your current research direction and focus?
John Hopcroft: My current research direction is deep learning. Deep learning has been used in many fields and it works well, but we don’t know why it works so well. My personal interest is to study the internal mechanism of deep learning.
AI Tech Base: How do you stay motivated to learn?
John Hopcroft: I’ve always had a curious mind. When I became interested in something, I followed it, and my career has benefited from it. It made me realize that some education actually takes place outside the classroom.
AI Science and Technology Base: How to find a suitable research direction as a scholar?
John Hopcroft: There are two types of research, applied research and basic research, but it’s not about which research is more important, it’s about why you do it. If it’s applied research, it may be because the company needs it. If it’s basic research, it may be because you’re interested in it and it doesn’t have to be practical. It depends on what kind of research you do, if you choose to do basic research, then you just want to explore, and when you find something interesting, you follow up.
AI Tech Base: What do you think contributed to your success today?
John Hopcroft: I was very lucky to be born into a good family. My parents are very loving and have a stable family, but neither of them went to college, so they want me to have a better life than them in the future, and they know that education is very important. Also, during my primary and secondary school years, I had many teachers who were very concerned about me. I believe I was luckier and had more opportunities than other kids.
In fact, those who study early childhood education find that the earlier a person does something, the greater the lifetime impact, and they believe that the first three years of a child’s life may be the most important.
Why is this important? Because not everyone is created equal. Imagine a child born into a stable, affluent, middle-class family. The other kid was born to a single parent and a drug addict and grew up in a complicated environment, and I don’t think those two kids have the same chance of success.
Limitations and potential of AI
AI Tech Base: What do you think are the limitations of deep learning?
John Hopcroft: First of all, deep learning is essentially pattern recognition and high-dimensional space, and within that range, we can do certain things. , for example, we are now training a depth network to identify the bike, it is to look at the pictures and compare the two pictures, when you see a bike, you will be the feature information extracted, car cushion can take, for example, you can ride it home, and so on, but the deep learning method and the feature information can not be extracted. So we need a big revolution to extract functional information and do more complex things.
AI Tech Camp: What do you think of the AI threat theory of Elon Musk and others? What jobs will AI replace in the future?
John Hopcroft: I’m talking about the next 10 years. They’re talking about hundreds of years from now.
When I was a kid, there was a job called “elevator attendant”, but now this job has disappeared, truck driver, taxi driver, cashier and so on will disappear in the future. It is clear that in the future only a small number of people will be needed to provide services that are needed by the masses, so we need to restructure society.
It should be pointed out that the concept of working for a company is actually a recent phenomenon, only 150 years old. Before 150 years ago, most people in the world worked from home (working for themselves) and then exchanged goods. So we have to restructure society so that people can have a basic income even if they don’t work, something the German government is now considering. Not only that, they’re thinking about what to do and how to engage in meaningful activities when most people aren’t working. Because when the masses are idle, the state becomes very unstable.
The gap between China and the United States in AI
AI Tech Base: What is the biggest gap between China and the US in AI?
John Hopcroft: I think there was much earlier research on AI in the United States. First of all, a lot of faculty [in the U.S.] are doing research in this area that they’re interested in, but in China, it’s often the senior faculty who say to the junior faculty, “Let’s solve this problem.” I think China needs to fix the problems in the education system first, and then China will catch up with the US in AI very quickly.
At present, There is a lack of basic research in China, and a lot of research is done following the us. This is something that China needs to change. When China was a developing country, it could look at what other countries were doing and copy it. But China is becoming a leader in the world, so it can no longer follow, it has to lead, and that will happen in the next 5 to 10 years.
Q: How can China catch up with the US?
John Hopcroft: The main problem is education. China has to improve its university education. There are many lead programs in China’s top universities. There is no doubt that they have cultivated many top talents in the world, but this is only a small part. What about the majority of the rest? That’s the problem. China needs to give world-class education to even the average students.
Q: How can China improve the level of basic research?
John Hopcroft: There’s a lot of work that needs to be done to raise the level of basic research in China, but I think the priority is to raise the level of undergraduate education. Why not graduate education, but undergraduate education? In my opinion, graduate education will not improve unless China improves undergraduate education first, because the top students from the Lead program go to the United States for their PHDS.
If you improve undergraduate education, you will produce a lot of top students, and the quality of graduate students will be greatly improved. So I suggest focusing on just one thing, and that is improving undergraduate education, and not just for the top students, but for the tens of millions of college students in China.