Are you having fun? Prof Winston asked me this question several times during my visiting semester at MIT. I was shocked to hear his curious tone when he asked it-----what's the point of measuring one's fun when assessing if she was suited for graduate school? But after working with him for a year, becoming his PhD student, and living in the devastation after his passing, I now realized-----that is all the point. Episode 1 - A Special Beginning I came to know Prof Winston in 2016 by searching for "AI lecture" on YouTube. I was then a freshman, studying Information Engineering at Nanyang Technological University (Singapore) and having no plan for my future. I remember staring at his lectures, feeling awed and inspired. I liked that when he asked "Do you prefer a demonstration?" cool things would happen on the screen, then another cool thing. I liked that when he said "you may ask... you might have..." a big idea would follow that explained something so clearly in 5 seconds, then another big idea. I liked that he knitted his brows and he told little stories. I liked that he walked close to the students. Sometimes he wrote on the blackboard, and then he would soon come back close to me. Even though it seemed impossible for me to switch to AI at that point, I kept him in my dreams. In 2017, my opportunity came with the MIT Special Student program, which enabled students from other universities to take classes at MIT. I ended my application essay saying "(I) may have the opportunity to sit in his spring class 6.803 (Human Intelligence Enterprise). There is nothing that I dream and aspire to do more." Luckily, I was admitted. Because I drew better than I wrote, I emailed him for the first time on 1 Jan 2018 with a drawing of him: He replied with a comment: I note that people who draw pictures are smarter, because we humans think with our eyes as well as our special human faculties. What a complement! When most people look at a drawing, they only see a nice drawing. But Prof Winston saw me. I said I agree by sending another sketch about why I think drawing can make people smarter. Then we started chatting about what subjects I plan to take. I had known that it would be difficult to get into 6.803 because it is a lottery class reserved to regular MIT student (only a portion of registered students will be accepted into the class). I was told that if I take the class, "he'd be throwing out a regular MIT student who may also have that dream." Maybe, I could, still try? I replied to Prof Winston that my top wish was to take his 6.803 and I hoped the lottery could bless me. And because my plan was to gain a PhD in AI and to become a professor like him, I planned to take three related subjects and one research project 6.100. I then asked for his recommendations on subjects. His reply lifted me to the air: For you, I especially recommend Robert Berwick's 6.863, which luckily will be taught this Spring (see link). He is an excellent lecturer with highly relevant material, some of which is drawn from his recent book with Chomsky (Why Only Us).
If you drop by before registration day, I may have other suggestions.
You will win the lottery.
What do you have in mind for 6.100? Instead of recommending a subject that would be taught that Spring, he "especially" recommended a subject "for me" that "luckily" would be taught that Spring. Instead of sending me a link to the list of recommended classes on 6.034 course website, he invited me to meet him in his office and to tailor suggestions after seeing me. Instead of telling me that I could take the class because I travelled ten thousand miles for it, he said that I would "win" the lottery because I might have proven myself good for the class. All this felt very special. So I couldn't wait but confess that I wished to contribute to his vision of "developing a computational theory of intelligence" by working on the Genesis System. The first step towards that vision was to get an Internet wire for downloading the code from Git. So at the end of my first meeting with him, he walked me to the infrastructure group's office. He watched until the staff found me the right wire and converter. He walked very slowly. He had been standing the whole morning preparing for and giving the How To Speak talk. He could have just given me an instruction or referred me to a graduate student. But he chose to lead me through the building. All this felt very special. Episode 2 - Give Students a S atisfying Experience Title: Please come see me now if you can...
...interrupt if I am with someone. Days later, I received such an email from him. I went there to be informed that he could not offer 6.803 that term because he would be out of action for a month. He said it with a heavy tone, not for complaining about his medical conditions but for caring about my feelings. He asked if I would decide to go back to Singapore. I said certainly not because I came for him. Why "come see me now"? Six months later, I learned that he was deciding to give me a private version of 6.803. He then soon concluded that if he was going to go to that much trouble, he might as well let another 20 or so students of the most positive attitude join. That became 6.yyy, the same Human Intelligence Enterprise with a small room of "special" students and two weeks without him. PS: He called the subject 6.yyy because the original subject had been called 6.xxx, which was called 6.xxx because he believed he could not remember 6.803. But he remembered all the students in 6.034 (Intro to AI) by their names each year. So who is actually having problems remembering the name of a subject? I guess it is us. To honor this fashion, I started introducing myself to other people as Yang or Z, instead of Zhutian. The next day, the students registered for 6.803 were informed through email that the subject was cancelled. I imagine that most teachers in this situation would just say that their medical conditions prevented them from giving the class, but Prof Winston was feeling sorry for something else: I deeply regret that I am unable to offer 6.803/6.833 this Spring. Unfortunately, I will be out of action for several weeks in March, and I have determined that there is no way to fill the hole in a manner that I think would produce a satisfying student experience of the sort provided in past years... Students who were truly interested in research on story-understanding systems signed up for 6.yyy. We sat in a small room with him sitting in the front. We read one paper and wrote a one-page communication assignment before each class. Each 1.5 hour class would be 1/3 communication lesson, 1/3 stories about the author, and 1/3 our opinions on the paper. To give each of us an equal chance to speak, he drew from cards of our faces to decide who would answer his next question. On 14 Feb, Prof Winston announced that he would be out of action until 5 March. He told us that amazing speakers would come to discuss important topics and he felt sad about missing the talks. So he asked for our help in summarizing the talks into broken glass diagrams (similar to mind maps). He needed my help? Wow. I felt motivated and in charge, barely noticing what he would be really feeling in the next few weeks. Prof Winston thought from our perspective and communicated through our perspective. Because he took all that trouble to be so considerate, I felt that he cared about his students and he cared about me. Episode 3 - Answers, Advice, and Action Items Switching my role from a remote student who watches his lectures to a mentee who learns from interactions with him, I was able to experience how he helped students. And most of his help comes in the AAAI format----Answers, Advice, and Action Items. Over the year, he told me several rules about communicating with him: Rule 1: Do not be afraid to ask questions and give suggestions. How are you going to learn anything if you are afraid to make suggestions ... Do not stop make observations. I am not god. Rule 2: Do not waste time by not asking questions. If I can answer a question in a minute, do not wait around to talk to somebody else and certainly do not blow a day trying to figure it out yourself Rule 3: Ask one question in each email, so he wouldn't delay responding because he doesn't have answer to one of the questions. Rule 4: Do not be afraid of asking too many questions. If there are too many, I will say no. When Rule 4 conflicted with Rule 3, he would respond me first and follow up in future emails: 3) ... Let me think about it, along with Question A. He had been advising students since 1972 and everyone whom I talked to after his passing said they valued his mentorship. What made him so admirable? I can only gain a little glimpse of his capacities by reverse-engineering the ways in which he crafted his advice for me: I recommend against trying to ... I worry about the first paragraph because ... Ok, but note that ... otherwise agree. Maybe a bigger part is learning ... Marr would accuse you of putting the mechanism before the problem ... I think a good warm-up project would be ... If you find yourself with nothing to do, read ... It would be instructive to go through ... First reaction is that ... would be better. Perhaps mobile robot could ... In research matters, I couldn't recall a single case that he imposed his opinions on me. He made it more like discussing with an intelligent friend. He was able to change my mind without saying "not" and "don't". If some step was really necessary, he would usually put it as: Be sure to ... Apart from answering questions and giving advice, he also gave very clear and specific instructions on what I could do or expect: One good project would be to A) ... B) ... C) ... Step: Received.
Step: Will respond soon. Help me understand (/≥▽≤/) Do you have CSAIL access on your ID card so you can get in early, late, and on weekends? If not, ask me about it on Monday. PS: One explanation for this action-item habit involves connecting our communication pattern to our thinking pattern in general. At that time, he was developing problem-solving systems that follow steps and adapt to conditions, and I was developing knowledge acquisition systems for learning by reading instructions and learning by talking with experts. So through our email exchanges, we seemed to be generating data for training our systems. But I felt more like he was generating data for training me. When it comes to giving specific and adaptable action items, here is my favorite: I suggest:
Project > clean to rebuild system
If that doesn't work, copy new code to a safe place first and then
Right click on Genesis > Replace with > head revision
If that doesn't work, find the bin directory where your operating system is putting compiled genesis files. Delete everything in bin directory and try again.
If that doesn't work go to bed. I ended up telling him that I could only go to bed ... Then he helped me to fall asleep easier by replying: ... I know you are going to school, so not a priority. Episode 4 - Hack Mode Once I proposed to add a feature to the system that we are developing. He said he would do it when he was in Hack Mode. And that seemed to happen every other day... What would you expect a 75-year-old famous professor to do during spring break? 29 Mar, Thur, 3:44pm: ... Very promising. Hope you are having a good break. See you Monday. 29 Mar, Thur, 5:36pm: I have debugged problem with ... 30 Mar, Fri: The ... code has bugs --- I am working on it now. 31 Mar, Sat: I am debugging and refining ... I will show you new features on Monday. 3 Apr, Tue: I have thought of a way to ... I will explain tomorrow. Imagine you were cracking your mind around a wicked bug on Sunday, 28 Apr, and received the following series of emails from him: 10:30am: ... I am looking into this. 10:55am: It is a nice day. Go outside.
... worked great yesterday, so it will take a while to see what's wrong and I am sleepy. 10:57am: On second thought, wait.... 11:01am: Ok, I see what happened... He asked me to go outside to enjoy the sunshine while he fixed the problem... What can I say. He really loved making his Genesis System. He really meant it when he said "The goal: To develop a computational theory of intelligence within my lifetime." So it felt heroic to get my hands dirty with him together. Episode 5 - Forgive and Educate Prof Winston treated mistakes through a calm and conductive way. Once I didn't realize why my mistake was severe, here is what he said: You are telling yourself the wrong story and missing an important question. Here is what you missed: ... Once I asked for his advice about a difficult life situation. He made himself available quickly: ... that was a blunder. Before you do anything, Skype. Once I screwed up the codes of the Genesis System due to carelessness. He gave me quite a lesson: I have decided on a punishment: When you get back to Cambridge the Genesis group will parade you around MIT wearing a placard around your neck that says "I did not check test cases." You will be dressed in rags and wearing a dunce cap. There will be crowds of students, all shouting insults and throwing garbage at you, especially in building 10 under the great dome. There will be photographers.
That all was a joke.
So, you need to be more careful, as such breaking could have been a very serious disaster, and as it was, I blew about 2 days debugging. Minimally you must: 1) Let me know when you touch files outside your package. 2) Run the test stories and make sure they all work properly The next morning, around the time I usually replied to his emails, he called me through Skype: I call just to make sure that you know it was a joke. I would never forget the lesson and his fatherly tone that day. Episode 6 - Care and Attention At the end of July 2018, I went back to Singapore (12-hour time difference from Boston) to apply my Genesis programs to robotic applications. That period involved me writing bugs in my day and Prof Winston's fixing the bugs in his day. This is a joke. But he did take charge of implementing many features and drafting papers. That was when Prof Winston realized my troubling sleep schedule: You looked very tired on Skype. How much sleep are you getting? Since then, I had received many emails and calls from him that lit up my whole sky: Send me a note after you wake up with 8 hrs of sleep. Tell your mother that you are sleeping too late. I'm thinking it (new group meeting schedule) is in the middle of the night for you (to join through Skype), but you can sleep in on Saturday. Correct?
Will you be able to attend from you residence? I don't want you running around town in the middle of the night. Remember, do not ever ride a bike at night in Cambridge. We have invested too much in you. I don't like you to get killed. Now you see, he was not in his usual Advisor Mode. My mother had been hugely grateful for how he had taken care of me. So I figured a way to introduce her to him: "I passed your words to my mother about my sleeping too late. She wholeheartedly agrees. With domestic and foreign force, we are confident that we are able to get Zhutian on bed by 11 pm. This is to come into effects on 3 Jan, 2019." His reply soon arrived: You have 15 minutes remaining to be awake. Episode 7 - Encourage I had known Prof Winston for 18 months, and never have I heard him using the same expression to encourage my progress or my idea. It must have taken a beautiful mind to produce the following: Splendid. Amazing. Very nice. I especially like your ... Thank you for joining yesterday. I value your comments. All sounds great. By the way, this is all terrific. Very, very interesting. A dimension of ... that had not occurred to me. Very cool. <-- very, very cool. This is extremely good stuff. He nurtured me with powerful words like these. Month after month, I grew into a different person. From being afraid of making a sound during group meetings, I started to share opinions. From being stressed over revising every email at least three times before sending them to him, I started to make jokes. From identifying myself as a student with a certain major, I started seeing myself as a theorist with promising ideas. I started criticizing and welcoming criticism. I started singing songs on my bike ride home. But one of the most powerful words of encouragement he gave me is not a compliment; it is a promise: As for the future, as long as you want to do what you want to do now, I will do whatever I can to make it happen. He did, all the way into his last week when he told me "let's make something impressive and here is the first step", until the future stopped on 19 July 2019. Except that the future didn't stop on 19 July 2019. As his ideas, writings, videos, stories and words of encouragement remained, he had already done all that I need for my dream to happen. Am I Having Fun? Yes, nothing have been more fun in my life than working under Prof Winston's wings, quoting his big ideas every day, and sharing my thoughts with him about almost anything. Because of the fun, I did not feel that I was working, but living my life to its fullest. Because of the fun, I did not feel defeated when I did not succeed at once----I was just learning and learning is fun. Unfortunately, I've lost him. I thought I had forever lost that source and sharer of my fun. But on second thought, let me be the one who creates and encourages fun, just like he did.