CBMM Summer School
Date Posted:
December 5, 2022
Date Recorded:
November 4, 2022
CBMM Speaker(s):
Boris Katz ,
Andrei Barbu All Captioned Videos Advances in the quest to understand intelligence
Description:
Boris Katz & Andrei Barbu - MIT CSAIL, MIT CBMM
PRESENTER: I'm now going to introduce the last speaker of the morning session, Boris Katz and Andrei Barbu. And just to give you some context, they're going to tell you about the Center for Brains, Mind, and Machines, having led a summer course, a three-week summer course, focused on the problem of intelligence. And they're going to tell you about that course, where it's been, and maybe where it's going next.
BORIS KATZ: We are all really fortunate to be here today and witness this amazing parade of brilliant talk by our MIT colleagues and our distinguished faculty. These presentations clearly demonstrate that we made great progress towards the creation of this new field of science and engineering of intelligence.
But in order to achieve these lofty goals that's spelled out in many of the presentations today, we need a continuous influx of young, talented scientists to carry the torch, to build on these achievements, and to take our field into the future. 10 years ago, when we started CBMM, we decided that the best way for us to do it is to educate the brightest, young scientists in the world so that they will be equally comfortable in neuroscience, and cognitive science, and AI.
And to make this happen we created the Brains, Minds, and Machines Summer School, basically for mostly PhD students and some postdocs. The school is taking place every summer in Woods Hole, Massachusetts. And it has been tremendously successful.
Now I'd like to introduce Andrei Barbu, who is manager of the school. And Andrei is also one of our TAs. TAs, these wonderful post-docs and research scientists that spend their time with students at the summer school pretty much 24/7. They are at the lab all the time, at all hours of the day.
They have breakfast, lunches, and dinners every day together. They go to biking expeditions together. And I would say, and I would argue that most learning at the summer school happens during these continuous interactions between TAs and our students.
And so I should stop here, actually, and I'll ask Andrei to tell us about much more about the school, about its past, its present, and its future. Go ahead, Andrei.
ANDREI BARBU: Thanks, Boris. So virtually everything that you've heard today, and one of the core ideas behind the Quest, is to eliminate some of the traditional boundaries between different fields of science, to have a group of people that are conversant across neuroscience, cognitive science, and computer science. But unfortunately, this is not how the vast majority of scientists are trained today. They're trained in one narrow area. And this is not how the vast majority of conferences work today. Our conferences tend to be highly specialized.
So like Boris said, for the past eight years we've been trying to change the status quo by running a summer school for three weeks. We invite some of the most brilliant undergrad, graduates, and postdocs students, 35 of them. These are experts in one of the three areas, in cognitive science, neuroscience, or computer science. And we give them the tools to become conversant in the other two fields.
And like Boris was saying, it's not just that we tell them about those other fields, although they do get to hear amazing lectures from some of the people you've heard about today and we'll hear about for the rest of the evening, but we have them do a three-week long project. That's what they spend the vast majority of their time on.
So they don't-- so they get their hands dirty, and they do something, which is going to be, very often, the beginning of a paper. So when they go back to their institutions, they don't just say, oh, I heard all these amazing things. They have the confidence to say, hey, I can do this new thing in this new field, why don't we change our research in this way? And that happens very often.
This is how Will, for example, took ideas from computer science and wondered, well, how would biologically plausible versions of these algorithms look like? How Pramod took ideas of symmetry and wondered how might the brain exploit symmetry.
And today, Pramod is not just sitting in the back. He's a postdoc with Nancy. And Josefina took some abstract ideas from semiotics that were very lofty and turned them into practical algorithms, working with our group, that tried to shed light on the evolution of human language and what algorithmically sets us apart from animals.
This happens over, and over, and over again. And a different piece that sets the summer school apart from other efforts is that it's multigenerational. Students come in, then they come in as junior TAs. They learn how to supervise other students. Then they come in as senior TAs. They have their own projects at that point. Then they come back as faculty members, and teach students, and bring their own students to be indoctrinated.
For example, Leyla is a faculty member at Johns Hopkins now. Candace is a postdoc at Meta. Colin is a postdoc at Harvard. Gemma is a faculty member at Goethe University in Frankfurt. Ignacio is a postdoc at MIT. Yen-Ling is a faculty member at Virginia starting next year. Each of these people started in the summer school as students and went up the ladder. I want you to hear about the summer school and its impact on them in their own words.
LEYLA ISIK: I had a really amazing experience at the summer school. It's just such an intellectually stimulating environment. So I think the combination of the amazing course instructors, the great talks they give, all the fantastic students, all being together in the summer camp environment in lovely Woods Hole really makes it a fantastic experience.
IGNACIO CASES: I am an autistic person, and I have to say that there were no barriers. The contents are incredible. The people are incredible. One of those few cases where I felt no barriers in my learning process. That was something very important in my life.
COLIN CONWELL: And it was this world of people coming together with all sorts of interdisciplinary backgrounds, all there to address the question of how we could use machines to better understand the mind. And in vise versa, use the mind to better build machines.
IGNACIO CASES: It definitely changed completely the way I look at research. And it also changed my career in a very, very positive sense.
LEYLA ISIK: But this very directly led to new research projects, one of which even resulted in a Journal of Neuroscience paper. But then, even at a bigger picture, being in the summer course really changed my thinking. And hearing from faculty that I didn't directly work with, and that worked on different areas than I did, I think really shaped my future research program.
GEMMA ROIG: I think it was a wonderful experience for the students, themselves, to have this hands-on experience, but also for me. And with one of them, in particular, we continued the project, and we published the result in one of the most prestigious conferences in computational neuroscience and machine learning, which is NeurIPS. So I think it was just a great success.
COLIN CONWELL: Almost single-handedly, CBMM summer school opened the door to the vast majority of the opportunities I consider open to me now.
IGNACIO CASES: Oh, man, if somebody asks me about going to the summer course, you cannot miss that. This is a life-changing experience. And you learn a lot. And it's absolutely fascinating. And you get new friends. It's unique. It's incredible. Has a very strong and positive impact on the research life of every student that has gone to it.
ANDREI BARBU: So let me do a deep dive on one student, Candace. She came, actually, through a different program in BCS, from Howard University to MIT. She worked in our lab for a summer on problems in machine learning. We were so impressed by her, we absolutely had to have her as a PhD student.
And already, even when she came in, she was working on problems to do with brains, minds, and machines. How is it that children manage to acquire the structure of language? How do they acquire grammar from perceiving what's happening in the world around them and watching adults speak?
But then she went to the summer school, and she heard that, hey, children have bodies, and embodiment matters, and they have mental simulations of what's happening in the world. And so she was wondering, how is it that your mental simulation of the world can help you acquire language more efficiently? And, of course, she published on all of these.
But then, a few years later, she came back as a TA and watched how students were using these models and saw their failures. And began to wonder, well, these models constantly fail. Is there something systematic in the pattern of these failures? And so then we did some of the first work on measuring social and economic and racial biases of grounded models. They're the kinds of models where you show them a picture of a woman who is a doctor, and they misinterpret it and just assume that all women have to be nurses.
And now she's a postdoc at Meta. But she's taken the values from MIT and these ideas from the Quest with her. So one of her first papers at Meta is all about, how do you make fair systems in NLP? So this really has an impact on people.
So in the past eight years, we've had about 400 students or so go through the summer school. It's resulted in well over 100 papers or so related to their projects. But we want to really supercharge the summer school in the coming years. We want to expand it beyond the faculty members that were originally part of CBMM into the more diverse group that you're seeing as part of the Quest.
We want to really double down on its diversity mission, add new slots to the summer school that are reserved for students from our diversity partners, and keep the summer school free. We're really proud that it doesn't matter what your means are, from what university you are, from where in the world you are, you can always come to the summer school as long as you're interested in these kinds of problems.
We want to double down on our research mission, give students the ability to take their projects and come back to MIT for a semester or two, and really push them forward to even better publications. And start workshops at major conferences. Today, if you're a computer scientist, and you want to learn some neuroscience, and you go to a neuroscience conference, the terminology can be very foreign. The ideas can be very foreign.
But imagine that there is a world where there are 100, or 200, 300 people that are conversant in your field, and your target field that have a workshop dedicated to taking people from your field, and showing them what kind of problems exist in this other field, introducing you to those people, and having much more multidisciplinary work.
We want to double down on education mission. There are many tutorials out there for, how do you learn computer vision if you don't know much about it? But there are very few tutorials out there that are centered around, how do you learn new ideas, even at home, on your own as a PhD student, that are at the intersection of natural language processing, and say, neuroscience?
And to take our materials at the summer school, distill them into something that's portable to other institutions, so that other people can also teach these ideas. And, of course, we want to increase our impact on industry with fellowships and job placements. Because ultimately, that's the bigger impact on society that we're going to have.
So to get the resources to scale up the summer school, our long-term vision is that it's not just going to be the MIT Quest for Intelligence in the future, it's going to be the Quest for Intelligence everywhere.
And if you want to talk to other students that have been to the summer school, there are quite a few in the back. [INAUDIBLE] was at the summer school, Federico. I saw Sherry somewhere. I'm sorry if I left out your name. Happy to take any questions afterwards.
[APPLAUSE]
PRESENTER: Thank you, Boris and Andrei, and thank you for all your work on that school, and the others that I know that participate. You guys are changing the world, not just in the science, but building the future of the people that are going to do it. So really kudos to you. That's wonderful to see.
have an interactive transcript feature enabled, which appears below the video when playing. Viewers can search for keywords in the video or click on any word in the transcript to jump to that point in the video. When searching, a dark bar with white vertical lines appears below the video frame. Each white line is an occurrence of the searched term and can be clicked on to jump to that spot in the video.
