53. Why do computer graduates need to solve problems before they can find jobs? Do students who often question their advisors do better research? How do ends and means affect what we do?

Original link: https://www.lidingzeyu.com/asking-why/

53. Why do computer graduates need to solve problems before they can find jobs? Do students who often question their advisors do better research? How do ends and means affect what we do?

In the last episode of the program, Jiaming and I started from his experience of studying at South University of Science and Technology, and discussed why domestic undergraduate colleges and universities have the strange phenomenon of disconnecting the Internet from time to time. such a point of consideration. At the end of the last issue, we discussed the mismatch between talent training and employment needs in universities. Listeners who missed it can catch up on the previous episode.

In today’s episode, we will continue our last discussion. Why do you still have to quiz to find a job after studying computer for many years? What challenges does Jiaming bring to recruiting people without taking exams in his startup?

Then we also discussed why people don’t ask themselves “why” more often, such as why do people do Ph.D., why do they do research? Many times we have to make the so-called major decisions about manpower when we don’t know anything. When we were ignorant in high school, we didn’t know the university, but we had to choose a university and a major. Choose your ideal industry and job.

Garmin and I also discussed the difference between “means” and “ends” and what do we lose when one thing we do changes from an end to a means?

Finally, we explore the difference between students who like to question their mentors and students who are less likely to question authority. What is the difference between the research projects made by these two different students?

Well, I like the discussion with Jiaming in this issue, and I hope it can inspire some of your ideas.

Youtube: https://www.youtube.com/watch?v=YRPkyotaAEo

Bilibili: https://www.bilibili.com/video/BV1LY4y1s7m8/

Text Version: Scroll down

Audio version Search “Li Ding Chat Room” in your podcast app

for example:

– Apple Podcasts https://liding.page/apple

– Podcast Casts: https://liding.page/pocketcasts

– Overcast: https://liding.page/overcast

– Spotify: https://liding.page/spotify

Guest: Jiaming Sun

https://jiamingsun.ml/

Image Derivative

https://idr.ai/

Apply for a job at Jiaming: https://liding.page/jiaming

Li Ding’s job search list: https://www.liding.page/jobs

Chat with me for half an hour: https://www.liding.page/1on1

0:00 Notice of this issue

1:22 The strange phenomenon of brushing the question

10:26 Ask why

20:23 The paradox of changing majors

25:20 Optimize GPA

28:23 Student questioning tutor

35:43 Garmin’s Startup ad

39:49 Preview of the next issue

Production team

* Li Ding Zeyu

* Pengdi Zhang

* Guo Yu

* Lin Xuhui

* Zhang Tianjin

Next program preview:

In the next episode, I interviewed Cao Xiang on Bilibili. Cao Xiang is currently in charge of the AI ​​Platform and next-generation editing tools at Station B. Before this interview, I had heard about Cao Xiang’s unique personal experience from many listeners and friends, including his experience at MSR, Lenovo and his own business.

In this issue, our starting point is how to proceed with the career path after the PhD in HCI Human-Computer Interaction. Cao Xiang believes that the important point of career planning is to constantly find the direction with the smallest angle between it and his own goals. And he thinks that he is very lucky that he knew what he wanted very early on. Once the goal is determined, the so-called career planning can be simplified into searching, or waiting for the emergence of such a minimal opportunity.

We also discussed, how to make real impact? How to combine scientific research with making money? Of course, if it cannot be combined, how to make money? We also talked about how hard it is to change the DNA of a company, and thus the car that software companies overturned trying to enter the hardware industry.

Finally, I still have to ask a topic I am very interested in: traffic password, how can I make this program recorded by Cao Xiang and I more popular at station B? There are no unexpected answers, but at least I tried.

Interested listeners can subscribe and turn on notifications to be notified as soon as possible.

Text version

(Please see the text version of the first half of the issue here )

Li Ding: When you graduate, the mismatch between the knowledge learned in school and what he needs to find a job after graduation is getting bigger and bigger. I also have the same observation that it is me. For example, don’t say no Yes, just say myself. When I was looking for a job when I graduated, I had to answer the questions. In theory, I had already learned C++, and I had learned python for more than 10 years at that time. I started using it in high school, but at that time, in order to meet the requirements of the interview, I had to brush LeetCode, and I also brushed it. After several months of LeetCode, I passed all the high-frequency questions. Like you said, this is a very mismatch. You have been studying in this professional industry for 10 years, and it turns out that you have to rush to work for a few months before graduation. This so-called so-called exam-oriented education for work. And then it doesn’t actually exist in the CS industry, it exists in all industries. For example, I was talking to an American doctor before, and after getting a doctor’s MD diploma, you have to take a separate doctor’s license, and it’s not exactly the same as what you learned in MD. I don’t know how big the mismatch is, but it’s obviously different. If it’s the same, there’s no need to study it. Then the same thing, such as being a lawyer, is the same. You have a JD when you finish your JD. I also did a chat with a JD before, a law student, after you finish your JD, you To test the bar of each of your states, each US state has a bar exam, and you have passed the bar exam, which has another mismatch. If you say that you have completed a 4-year undergraduate degree, or even a JD with a JD and N years, it turns out that your degree is actually useless, and we still need a professional vocational exam. There should be a lot of examples of this, I just happened to think of these two here, I think this is very interesting, but I don’t understand why, why it is getting more and more diverge.

Jia Ming: Let’s not say anything else, that is, our own field is CS. Why do you want to find a job to brush LeetCode? Actually, I still can’t understand this matter. This is like when I interview others, I never face arithmetic questions, I think this is a waste of everyone’s time. I would ask him to give a specific question, for example, if there is a bug in this code, you can solve it for me. I can just observe how he solves the bug, how he searches for information, and how to check documents. How to see other people’s solutions to similar problems. I would think that this matter is more worthy of investigation and is more related to his actual work content. Of course, this may be better. It also confirms that this may be more during the undergraduate period. I just said that “learning by doing” is in the form of “learning by doing”. , one of the most repeatedly trained abilities. I don’t quite understand why, for example, in the United States, especially in the United States, there is actually a trend of finding a job based on the theme of brushing. In fact, there are often some things that exist. Although you don’t understand them, there are often some principles behind them, and they are reasonable. But I haven’t thought about it so far. The only thing I have is seeing other people’s opinions, that is, he is testing your obedience, that is, giving you a goal, you can accomplish it well in a short period of time, It shows that your obedience is good. So you can come, maybe because I don’t think it’s really because, this is a very puzzling. Because the United States may not need this kind of obedience, if you say that China thinks this way, there may still be a little possibility, right? But I don’t think the United States is an employee recruited for obedience, so I don’t know if you have some profound experiences and opinions. That is to say, why do you want to set up, is this the collective unconscious or what? why is that?

Li Ding: Actually, I don’t have a deeper opinion, but I have a different opinion. I also read it on the Internet before others said that I read it and wrote it on the Internet, why the United States converged to do LeetCode. Especially for junior interviews, now I feel that many senior / staff interviews are also brushed by LeetCode. They say that the culprit is the so-called big company. At that time, for example, Google’s so-called big factories started to standardize this thing, because once your company’s scale is large, you want to ensure that the recruited people are compared and can be quantified. But if you want to quantify, because everyone’s academic background is different, for example, a so-called Ivy League school, or a community college, or a school with a lower ranking, how to standardize these things. You can’t always talk about those people in famous schools, how can they be? So they come up with some standardized tests, and then slowly evolve. Because the interviewers have to have the same questions, they can’t say that every interviewer asks questions that are completely imaginative, so you can’t do a calibration among multiple interviewers. So slowly, the interviewers who are similar to Google have only 20 question banks. At the beginning, for example, dozens of question banks, everyone found that we can brush this name, brush this question, and as a result, Google will increase Question bank, slowly evolved to the current situation of LeetCode where we see thousands of questions. And then Google has already imitated this, and other companies will follow. Look at the startup. You see a lot of interviewers. They prepared LeetCode to meet Google. In order to reduce their burden, we can come to meet us. . Although we don’t understand it, it certainly makes sense for Google to use it. Then I personally guess like this, it has snowballed a lot now, I feel like any government company policy, once it snowballs, I feel it’s hard to stop.

Jiaming: Yes, it is actually difficult for me to understand. I think, for example, interviews in the United States, and some design interviews, including behavior interviews, right? I think its existence is something I can understand. But it is the coding part. The test is for this kind of question, which I can’t understand so far, and many people don’t understand it.

Li Ding: I don’t understand either. I can only say that our company is a research engineer. During the interview, we will meet coding, but we will not meet coding such as LeetCode, but it is more like what you said, that is to say, to give a real answer in our research environment engineering challenges that will be encountered. Of course, this has advantages and disadvantages, just like you said, you can ask some real abilities to reflect the real world. But the disadvantage is that, for example, the interviewer will think that your company is a little strange. All the other companies I ask are the kind of standardized questions. In fact, other companies are using the college entrance examination. Why do you engage in self-enrollment? Is there something tricky about your company? Of course, because our company’s reputation is better, it will not be affected. But you can imagine that our company has accumulated so many years to dare to do this. If you ask these so-called bizarre questions from a startup, I feel like you might scare off a lot of job seekers.

Jiaming: I have some experience with this incident recently, because we were interviewing like this recently, so we used this method to interview. Yes, there will be some people who are completely uncomfortable with such a way of inspection. Maybe this is a good screening period. I think this is a feature, and I think I really hope to recruit people who recognize this interview method very much. He can also solve these problems very well, because of this matter, I believe everyone will have more or less experience.

Li Ding: Yes, you mentioned intrinsic motivation just now, can you be more specific? We went off the track just now, saying that this coding interview, we took back the intrinsic motivation.

Jiaming: Actually, it also ran through me, but I may be a sophomore or junior in my undergraduate degree. After I slowly figured out this matter, it ran through it. It is still going to often chat with others until now. This thing is to say, this is during the undergraduate course, this thing is actually a very worthwhile question in itself. But it can also be generalized and extended to all aspects of life and work. In all aspects of things, that is to say, this contradiction is that some things are what you want to do in the first place, you yourself know why you want to do it, and Some things are what the evaluation system expects you to do. For example, for the platform system, it will use a quantitative standard to evaluate you, which is the same as LeetCode, right. That is to say, it needs a standardized thing, and everyone is at the same starting line to be fair, and to evaluate everyone’s ability in all aspects. Not to mention whether this thing is reasonable, not to mention whether this thing can be done, but at least this thing exists, and this contradiction is even very serious. That is to say, during the undergraduate period, in fact, I just said that there is a mismatch between the ability that the job you are looking for expects you to master, and the ability that the school expects you to master. So for research, there will be similar. If a student wants to do research after graduation, he will also face this kind of contradiction, even more. That is to say, his common contradiction is undergraduate. For example, everyone knows that sophomores and juniors want to join a certain laboratory and start to conduct scientific research training. If they are particularly strong students, they hope that when they apply, they will be able to apply. If you apply for a senior year, you can have a paper, and if you can send it out, you can at least vote for a paper. For this field of CV, that means he has such an expectation. And the so-called senior and senior also told him that this is the so-called optimal solution. This is if you want to go abroad, not to mention that no one asks you why you want to go abroad, but if you want to go abroad, and use the word “going abroad” instead of reading the word “PhD”, going abroad is the purpose, not “PhD” is Purpose, these are some very strange phenomena. If you want to study for a PhD or go abroad, you should do these things in your second or third year, and he will go. I will contact the laboratory in my second and third year, including myself. At that time, because of the environment of SUSTech, we also encouraged students to participate in scientific research during their undergraduate years. So in fact, at that time, I was in a state like joining a laboratory in my second year of sophomore year, studying and doing some things. But there is a contradiction, that is, the laboratory or the things you want to join in the laboratory to learn things and do are actually different from your so-called school evaluation system. Because he will spend the only time you have, you may, for example, spend more time in one place, and the other will be affected. But for example, when you apply for a PhD, there are actually requirements on both sides. This contradiction is actually a very, very sharp contradiction when I was a student. Maybe for some very talented people, they may think that I can handle both sides very well, but I think this is very difficult for most of them. You can’t do both, and you can do research very well. Well, on the one hand, the school curriculum can be handled well. I think this is actually a very high requirement for a person’s ability, if you want to do both well. So one of my experiences at the time was that the contradiction in this matter was very sharp. This extends to you can ask a lot of why, why do you want to brush your GPA, why do you want to meet the school’s evaluation standards, why do you want to join the laboratory, why do you want to study for PhD, why do you want to do research, why do you want to study , there can be a lot of interesting questions behind it. In fact, in the final analysis, it will involve one point. In the end, it is your motivation to do everything, such as the external evaluation system. Even the senior and senior I mentioned earlier told you to go, you want to join the laboratory, this is also an external evaluation system. As a third-year student of SUSTech at that time, one of my luck was that I didn’t have such a senior or senior to tell me anything, because it was a blank sheet of paper, that is to say, the whole environment, I had no previous experience. I can learn from it. The people behind me have my experience to learn from, but the front of me is almost equal to none, especially in computer science. I am the first computer student. So in this case, I will naturally ask a lot of why such questions, why am I right for anything, I am now these things that others tell me to do, why do I do it. I remember thinking about it at the time, and I also discussed a lot of such issues with some of my good friends at the time, and the conclusion I reached was that why do these actually come back to one point in the end, you still have to find out Your own intrinsic motivation is intrinsic motivation. Why do you want to do anything, why do you want to join the laboratory for scientific research, why do you want to learn the course well. I really touched on the last topic, which is the process of “learning by doing”. It may be easier for you to find why, and “learning by doing” may be more difficult. I just think that at such a stage as an undergraduate, many people may even be on the verge of graduating with a Ph.D., and they may not understand why they want to study for a Ph.D., or even become a faculty. That is to say, this question should be asked at a very early stage, and it should be thought over and over again. If you want to understand it before doing it, I think it is more suitable for me, at least in terms of my personality. mine. Instead of not trying to figure out why to do anything to take the time to do it, it was a very torturous process for me.

Li Ding: I very much agree with the point of intrinsic motivation. Because I just recorded an episode with my PhD tutor Changxi a few days ago. When the audience hears this episode, it should be the last episode (51 episodes). After he took the tenure, he was searching for his inner why, Why am I doing what I’m doing? Just to rethink all the paper he has published. That’s how I sounded to me, kind of like criticize Changxi, but it’s not that I just say It takes him that long to think about why he’s doing this, right. Changxi is so smart and an established researcher. It took him many years as an undergraduate, as an intern at MSRA, then as a Ph.D. he’s doing. I feel that I am no exception. I am also likely to start thinking about why very late. I feel that this is an undergraduate, high school, or even a Ph.D. No one prompt will come to ask you why. For example, during PhD qualification, he will ask you how well you know your field, but he will not ask you why you are doing this. Of course, this question is often the implicit professor who will ask you why you want to do Computer Graphics. We will give some answers, for example, I like beautiful rendering, I feel like I want to do Pixar. But maybe I didn’t think a little bit deeper at that time, why did I make an animated film like Pixar, why. If you keep asking why why why, you may be able to help some people. Then I would like to add one more thing. You mentioned just now that “learning by doing” is difficult to discover intrinsic motivation, which I may not have thought about before. But after you said it, I was just thinking that I agree, because if you accept this setting in the framework of “learning to do”, the first subconscious is to optimize the GPA. Because you are studying and you are a student, right, in the course, your lab only takes up very few points, and the project has very few points. The most important thing is midterm and final, right? How to optimize these two, the final optimization result is to learn right. If you learn it, it will be difficult for you to find out what you really like to do and what you don’t like to do, because you are only writing the answers in words. This is actually something I haven’t considered before, because I don’t think there are too many such project-based courses that may be part of the HKUST and Columbia University, but for example, I may take at most one per semester. It doesn’t go up when it’s time, but maybe one or two times a year. It will be the kind of open-ended professor who will give you an idea, and then a few students will do a project like this. But there are very few of them, most of which are like you said, “learning by doing” is the vast majority. I may have a course that I am particularly impressed with, that is, Operating System. I don’t know about SUSTech or how education is taught in China. Anyway, when I was in HKUST, I felt that it was the most rewarding course after I finished it. Talking about my gains, I I don’t want to do research on Operating System, and I don’t want to do anything related to operating. Just because I didn’t know what the domestic course was like. At that time, the design of HKUST consisted of a lot of projects, one small project every week and one big project every two weeks, so basically I was in the lab that semester, sitting in the computer room. spent the entire semester there. Because you have to debug all kinds of things, and then group projects, you have realized how to teamwork with others, and realized a lot of things that I did not experience in previous courses. Like you said, you can know what I don’t like to do, so I choose some other direction. But you said why I chose Computer Vision for my undergraduate degree and Computer Graphics for my Ph.D., but I don’t have a lot of signals to tell me that I really like these two fields, but say that I happened to do it, I Not very annoying, then why not. Since I have already started doing it, I can have more capital if I continue to do it, right, and then I will continue to do it, there is an inertia.

Jiaming: I think how to say it, I don’t think the process of practice can be said, you can really find the so-called true love. That is to say, if you can find the one, in a certain direction, in a certain field, the meaning of this may not be here, but more can be used for you to go to is equivalent to trial and error. He can tell you what you want to do, but he can also give you negative samples and tell you what you don’t want to do. I think you gave a good point just now, so I don’t necessarily say that you have to figure out what you want to do when you make important life choices, such as when you graduate from an undergraduate degree. , in fact, it can be a very long process, because what a lifetime is very long, a person’s career has decades, so it may not be too late to figure out this matter, at least you It’s not too late to figure this out before your career is in a stable state. I think I am very surprised about this matter. It is said that very few people will put this matter on the table to talk about it. It does not mean that you have really reached a certain state during your undergraduate period, a certain direction of convergence, etc. . It is about this matter, why this question is not treated as an undergraduate main degree. Just like you said that there is no prompt, there will always be someone asking you such a question, whether you have reached such a state, all of you are not even clear about this goal, and even very few people will find themselves very inner. Motivation to do things as their own, such as freshman and sophomore goals. What’s more, for example, you are assigned to a major, if you are not assigned to a candidate major like SUSTech, you must do a good job in the major. The choice you make is based on your performance as a high school student and your college entrance examination results, that is, a number of factors that are totally irrelevant will determine you to do this, I think it’s ridiculous, the whole process. This is what I think is often the case, but perhaps the most absurd thing about the transition from high school to college is that even the choice you made when you didn’t know anything about it, you can’t even change him. . Let’s say that you can get a divorce if you get married, but even if the cost is very high, you rarely hear about someone in your undergraduate degree, for example, because I don’t like this major, so I want to drop out of school, and then I will take the college entrance examination again. I don’t think I can say that there is something wrong with the design of the whole system, at least to say that even if you choose a major that you don’t like in the end, you should have enough motivation to find what you think should be the right choice. There should be a similar mechanism that allows people with such thoughts to do some of this switching, I think this is one.

Li Ding: I think you say this is actually very interesting with the ratio of majors in universities. I don’t know much about the ratio of majors in China, but I have chatted with a few very limited friends. It is a difficult process to change majors in domestic universities. , you can turn away. So this is a paradox. If you want to leave a department you don’t like, you must first get into the top of the department in the department you don’t like, and then you can transfer to a department you think you like, but You don’t actually know if you really like that department, you just want to give it a try. Then this is not criticize China anymore, it was the same model when I was in Hong Kong. At that time, many people were studying very well in mathematics, for example, and suddenly he was going to transfer to chemistry. We were very puzzled that you were not the best in mathematics, why did you suddenly change direction. They said that I just liked that major, and I changed it when I was fed up. But in fact, I didn’t understand at the time, because I felt why you gave up such a good foundation. Looking back now I feel like they were brave and they made one of their attempts at the time, which is very encouraging. I haven’t actually contacted them now, but I feel like they should be thankful that they made such a decision, or at least tried it.

Jia Ming: Yes, I think that when it comes to SUSTech, our point at that time was very good, that is, no one restricts you from transferring to any major. Of course, this will obviously lead to, for example, CS Finance being full. In some majors, it is inevitable that no one will learn this. Voting with your feet has always been true. I think this is very interesting because it involves many factors. For example, when you choose a major, there are many people in computer science. Does it mean that you need to increase teaching resources and support so many people. He will pull one hair and move the whole body, affecting many other things. It may be precisely because of these, and because of these other constraints, that this opening cannot be opened, and everyone can have such a free choice of major. Such a right, so I mean the complexity of the matter itself, not to say why no one is doing this now. He can do it all at once, but it is not so simple, but there are more behind it, even for example, some systems of colleges and universities and so on. Including why some colleges and universities must concentrate research and teaching on some people, and why these two roles cannot be separated, it involves a lot of deeper levels of some status quo. And I mean this contradiction may also be difficult to reconcile.

Li Ding: This is not a very simple answer, and this is not our purpose in this issue. Our purpose in this issue is to discuss and discuss, and then share some of our views and talk about some of these thoughts. It would be great if it could inspire some deeper views of, for example, educational practitioners.

Jiaming: I also very much want to hear how practitioners in this field think about us discussing these issues, because we are not professionals in this direction after all. Also, maybe I want to make a point about what we said just now as a student, and maybe it will have more instructive meaning for our maybe ordinary audience. That is to say, especially for undergraduate students, it may be more worthy of attention to what you do, right, whether it is an end or a means. If it is a purpose, it means that you want to do this thing yourself, or that your purpose for this thing is something and you have something that will become a means. Let me give an example. For example, you don’t actually want to give an example. Your undergraduate degree is for example, machinery, and machinery to CS is a very common operation. The undergraduate major is mechanics. After graduation, you want to do programmer-related work. In fact, after graduation, you can understand it as a means, because your ultimate goal is to engage in, for example, a computer-related major. , or to apply for a PhD in this direction, etc. That is to say, when the nature of a thing changes from an end to a means, because many people often think that it is an undergraduate, I often think that the meaning of the whole process of undergraduate is that its purpose is to get a diploma. To graduate, but there are many things that you think are ends, it may end up being a means to a farther end. If a thing changes from an end to a means, you can think of some ways to overfit this evaluation system, so that your goal can be achieved faster. This is something I think is worth mentioning. For example, if your purpose is to apply for a PhD, not to mention why, of course, it is even better if your purpose is to do scientific research, it must be that applying for a PhD will become you. means. But if your purpose is to apply for a PhD, then in fact, these undergraduate evaluation systems may become an intermediate goal. So once it is an intermediate goal, you can think of some ways to overfit an evaluation system, and then achieve your goal.

Li Ding: I very much agree with what you said. Once a thing turns from an end to a means, there will be various short cuts. This overfit is very common, especially people like us from the roll country can definitely roll up. Me too, just now you mentioned applying for PhD optimization, and I thought of a case of my own. I actually don’t proud anymore, but this was what happened to me at the time, because when applying for a PhD, I would have to pay the transcript of the GPA, and the higher the GPA, the better. At that time, many students said that some courses with relatively low GPA should be put into the last semester, because at that time your application has been approved, so it doesn’t matter. At that time, we needed to take two courses in humanity, one in sociology, two in humanities and one in social science, and the amount of reading in these courses was huge. With a pure engineering science background like ours, at least I was not exposed to the kind of non-science and engineering knowledge that I read at the time, and I put it off until the last semester. That semester, of course, the results were good, because those classes I took were the ones I liked very much, because I didn’t care about the grades anymore. As long as I am interested in this topic, I choose it. I chose East Asian History at that time, and I learned a lot of things that may be missed in high school history textbooks in China. Because the main Chinese history that may be studied at that time, the history of East Asia is a history of the entire East Asia, which I find very interesting. Then the teacher was a professor in Taiwan, and I thought it was very good. What about Introduction to music theory? Although I don’t know any musical instruments, but I have learned a lot of this kind of stuff. At that time, I felt that I had a headache to memorize those music rules. Then, because there was no pressure of grades, I felt Instead, I learned very well. I’m really learning for the sake of learning. I didn’t say that I took any short cut for the sake of brushing grades. The same is true of sociology, because I really liked it at the time and signed up for that course. I didn’t like it and I didn’t even want to watch it. Because it’s the last semester, I don’t care about grades anymore. So he has pros and cons. I can only say that if you say that I want to take this course in my freshman year and sophomore year, I will definitely be more utilitarian, because I think it will affect my future application.

Jia Ming: I think this may also be, for example, your school in Hong Kong and a big one in China, including, for example, I don’t know about this matter in the United States, for example, you observe the United States, during your PhD, you are around The people at the school and the undergraduates at the school at that time, I don’t know if you have such feelings. That is to say, domestic students may be more inclined to go and expect an external evaluation system, and then they even take the initiative to find such an evaluation system, and then they can work hard to overfit it.而不是说就是,比如说他自己有一个非常明确的想做的事情,然后有非常明确的主见,然后怎么样就是说我不知道。你在这件事情上有没有一些观察,就是说在香港和美国的环境,以及你在国内上的体会的一些不同了。

李丁:对我之前跟很多合作者,包括教授,包括公司的研究员也都聊过。我们的观察到的一个点就是在纯中国教育体系下,大陆体系下出来的学生,像你刚才说的就会寻找一个这种外部的评价的标杆。同样的在相对应的。。

佳明:他甚至变成一种安全感的来源。

李丁:他就希望你能够告诉他,要怎么做什么样才是好的。相对应的话,如果一个学生是在一个比较偏西式的这种教育成长的话,他就是比较不服气,很多时候我或者说比如其他这种我的collaborator,他们给出任何意见,那些学生的第一反应就是说why,why am I doing this,我不想做这个东西能不能不做,我能不能做点别的。与之相对应的话很多见过的一些中国大陆的一些学生可能就会说,你说这个idea 挺好的,我来试一试,他们做得非常快,很快就试出来了。但是他们做出来好和坏之后,他们就会在等着adviser 们,包括我的collaborator 等着adviser 们来说这个结果好坏,下一步我们该怎么做。他告诉他之后,他又说good idea,又接着去做,他们的反馈特别快特别好。当然这个是一个相对而言的,我不是说国内的学生没有critical thinking,我只是说跟我听说过的或者接触到的美国体系成长下的西方体系成长的学生,他可能比较少的会去质疑所谓的权威了。我也不是什么权威,我只是碰巧比你们早毕业几年而已的一个人而已,但就是他们会很不太会去质疑权威。相反的我如果是在一个这种比较西方的一个小孩,他们就会时时刻刻在质疑我的任何的事情。当然有好和不好的,比如说一个outcome,之前我记得我们就像上一次聊天的时候我也提到过,就是说很多时候在一个project,在学生他不太质疑我的情况下的时候,这个project 做的跟我的vision 会很像,因为essentially 就是我的vision,我出去给talk 的时候都很轻松,因为这就是我的vision,学生实现的非常好。相对应的作为一个天天被质疑的一个project,被学生天天质疑说为什么你要这样做,你不该这样做,应该跟着我的想法走,这样子的project 出去我给talk 的时候,或者跟别人分享的时候,我会要花很多时间去准备,因为其实这个idea 已经不仅仅是我的了,他很多都是学生自主创新想出来的东西。我就在这里对比,我比如说过三四年五年的,哪个idea 我回头看会更喜欢。包括我有一样的问题,我也问了我的collaborator,大家的一个感觉是大部分这种被challenge 的想法,最终会是在大浪淘沙里面剩下的一些好idea。而自己当时比较独令断行的这种idea,这种类型的想法,反而可能不太经得住时间的考验,因为没有人质疑,所以我们就在这条弯路上面越走越远。虽然最后都两个的outcome 都是一篇paper,甚至都可以是best paper,但作为一个researcher,我自己回头看,或者collaborator 自己回头看的时候,我们会感觉这个有点意思,这会inform 我们接下来的研究该怎么做,我们怎么带学生,怎么去mentor 学生。比如就会鼓励说,如果我们看到一个学生不老是质疑我的话,我觉得鼓励他就是质疑我,挑我的刺,这才是因为我们发现这样子的研究,更经得住时间的考验。如果没有人批评我的话,反而我天天就在活在自己的bubble 里面,这个是我感觉我不想看到的,我想成长,所以我希望跟我合作的无论学生也还是教授也好,能够指出我的不足。

佳明:是的,我觉得其实作为一个学生来讲,在这种以跟一个比你更senior 的人合作的时候,可能还是有不同的strategy。就是说对我是觉得有很多人会选择说,我作为一个很好的执行者,会把目的达到,比如说我要get published,可以把一个paper 发出去,或者说我让自己的简历当中多一个paper,这个其实是比较常见。或者说绝大部分人,可能都是这样去看这个事情了。至少从我之前看到过的一些例子来看是这样的。就是说我觉得说到底还是在于你是否有,你是把publication 当成了目的还是当成手段。我觉得有一些人可能觉得我跟这样一个senior 的人合作,我可以很快的把paper 发出来,特别是本科生,他很快会很习惯会这样想。但是可能这个时候他就是这个只管干了,我是一个很好的执行者,可以把这个事情很快的实现出来。但是很多时候真的就错失了把一个research 做好这样的一个,我认为是更值得追求的一个目标。这其实还是一个目的与手段的不同,或者是intrinsic motivation,满足intrinsic motivation 和满足评价体系的这样的一个不同了。我第一次是可能我观察的也是很多我见过的国内的学生,可能更擅长去找到这样一个评价体系,去满足外界的目标,把一些本来应该是手段的东西当成了目的。比如说到底,你发paper 是为了让publication 加一,还是为了你把research 做好对吧,这也是一个手段和目的的不同。所以我是觉得这件事情也是值得去思考和讨论的。

李丁:关键真是太巧了,因为这录制这期跟录制Changxi 那期是完全分开的两个话题,其实没有特别大的overlap,但是最后讨论到我们看我们两个人converge 到这个点,和我跟Changxi converge 到那个点出奇的一致。就是什么是目的,什么是手段,发paper 到底是什么。对我感觉就是听众可以把两期(51-53期)联系在一起听,因为是两个perspective。Changxi 是一个tenure 的教授,他肯定是作为一个教育者,他可能不是教育,他可能并不是教育的改革者,他是教育的的一个从业者,这是一个角度来分享的。我们两个纯粹是属于更多的教育的旁观者和经历者来去给出了一些comments。

佳明:对我觉得这个非常有意思,但是证明我们碰巧想的东西是比较一致的。我也很期待看一下,对作为一个tenure 的professor,他是怎么看待这些问题的。

李丁:因为刚好今天下午我在剪辑他那期,就是我们刚录制那期节目,所以一切就好像我在听echo 一样,下午刚听完,结果晚上又听了这个声音。最后的话,我们这期聊的时间比较长,我不知道最后要切成两期还是什么的。你现在要给你的startup 做个广告吗?

佳明:好啊,就是说其实非常有趣。就是说我们其实刚才讲了,很多重要的人生节点的选择对吧。我最近其实就做了一个我觉得非常不同的选择,我加入了一家创业公司,或者是co-found 这家创业公司,是做三维数字人的内容生成的,我们主要是用neural representation 就是neural rendering,和基于neural representation 的这种reconstruction 来去做人体的这种静态以及动态的建模。你可以理解为我们可能是全世界非常少有的至少目前我只听说过一两个的这样的这种以NeRF 或者是neural representation 为吃饭的本领的这样的一家创业公司了。所以在技术上我们是有非常先进的国际领先,我们团队其实是我现在所在的浙大的CAD 实验室的团队和中科大的一位老师的团队联合一起创立的。我们其实在这个领域有很多的论文发表,包括各种顶会,其实加起来快近百篇论文了。我们其实出来创业也就是看到了这样的这种neural rendering 然后就是NeRF 的这一套技术进入了一个相对比较成熟的状态。所以说我们觉得是一个很好的时机,可以把这些事情真的自己做了这么久的学术,可以把它真的产业化去落地。所以是正好看到这样一个机会了,这恰好同与此同时也是我们产生一个想法的同时,元宇宙的浪潮也兴起了,所以说各方面的因素在一起,我们去决定来做这家公司。所以也欢迎有类似想法的同学加入,我们也招我我们肯定是招这种算法相关的,特别是3D vision,graphics,以及人、人体的数字化等等,和各个方面的一些算法的人员。这个应该也是我们李丁聊天室的听众,应该是理工也会比较多的。我们同时也会招开发人员,比如说移动端的开发iOS 和安卓包括前后端,包括一些云服务的一个运维部署等等。其实我们作为一家创业公司,各个方面的职能都是很稀缺的,所以说希望感兴趣就可以投简历,另外一点想说的就是说我们其实也是个非常不同的创业公司,其实非常有趣的topic,我们其实更多的像是一个就是把自己的原创技术去落地的这样一个非常少数的国内的团队,我觉得是一个非常稀缺的体验。我们在落地自己原创技术的过程中,也会做出更多的原创技术,相当于是我们把自己做research 这个能力,去继续的在以一个明确的工程目标的方向去继续的发扬光大,去在做出很多甚至是学术上也都非常有价值的创新。所以整个的journey 本身也是,至少目前我干了大半年,我会觉得非常rewarding,非常有收获的这样的一段journey,所以如果对neural rendering,对这种人体的三维数字化digital human 有兴趣的同学,其实非常欢迎,至少加入我们可以跟我们聊一下。我可以详细的介绍一下我们做的技术,看看是否有兴趣加入我们,我们也招实习招全职都招,也非常欢迎实习,地点在杭州。

李丁:各位有识之士,我的听众里面如果感兴趣的可以考虑。我发现我之后是不是应该开设一个李丁聊天室的jobs page,因为基本上很多来这边嘉宾都会有一个posting,我感觉有一个中心化的地方可以把它全部show 出来,这样的话听众可以有个地方去看之后,我来看看能不能建一个这样子的网页,很简单的网页。

佳明:这是非常好的一个平台也非常感谢能给我宣传一下的机会。

李丁:今天谢谢佳明来这里的分享,我们讨论了从南方科大讨论了很多发散性的这些东西,这期的节目时间比较长,然后但我感觉是很有意思的一期。我们下期再见,拜拜。

本文转自: https://www.lidingzeyu.com/asking-why/
This site is for inclusion only, and the copyright belongs to the original author.

Leave a Comment

Your email address will not be published.