Technology Day 2021: Pathways to the Future of Computing

Technology Day 2021: Pathways to the Future of Computing

Show Video

WHITNEY ESPICH: Hi, I'm Whitney Espich, the CEO of the MIT Alumni Association. And I hope you enjoy this digital production created for alumni and friends like you. CHARLENE KABCENELL: Hello, everyone. My name is Charlene Kabcenell, a Course 6 alumna from the class of '79, and I have the honor of serving as President of the MIT Alumni Association. I want to offer a special note of welcome to my fellow alumni who are celebrating their class milestones this weekend, and I also want to thank the thousands of alumni from all classes and all schools who have made the time to participate in today's in-depth Technology Day Program.

I hope that you're enjoying it. I know I am. Tech Day is a popular cornerstone of the MIT Tech Reunions experience brought to you by the MIT Alumni Association each year. While our present pandemic related circumstances prevent us from being over there in Kresge, one of the silver linings is that our online setting enables us to engage all alumni, no matter where you are in the world, in this meaningful institute tradition.

This being MIT, we find ways to mark the past while committing to help in the present. And as always, keeping our eyes to the future. That omnidirectional way of thinking is very much at the heart of the Institute's Schwarzman College of Computing, and we're fortunate to have with us today the college's inaugural Dean, Dan Huttenlocher.

As many of you know, Dan is a fellow alum from Course 6, earning an SM in electrical engineering and computer science, and a PhD in computer science from the institute, after earning his bachelor's degree from the University of Michigan. As the inaugural Dean of the MIT Stephen A. Schwarzman College of Computing and the Henry Ellis Warren 1894 Professor of Computer Science and Artificial Intelligence and Decision Making, Dan is back at MIT following a distinguished career at Cornell. Most recently, he served as the Founding Dean and Vice Provost at Cornell Tech.

Dan also has extensive industry experience having served as a scientist and lab director at the Xerox Palo Alto Research Center, and also helping establish a financial technology startup, Intelligent Markets. Dan's research in computer science is broad and interdisciplinary, spanning algorithms, social media, and computer vision. His research and teaching have been recognized by a number of awards, including ACM Fellow and CASE Professor of the Year. And today, we have him here to take us through the pathways to the future of computing. Welcome Dan, and thank you for joining your fellow MIT Alumni today. DAN HUTTENLOCHER: Good morning from here in Cambridge, or whatever time zone you're in.

It's a real honor and pleasure to be here with you today. Let me just quickly share my slides. So I wanted to talk to you a bit about what we're doing with the Schwarzman College of Computing here at MIT.

And then we'll open it up and have some time for questions and discussion. So the mission of the college really is to address the opportunities and challenges of this age that we're living in, an age where artificial intelligence, algorithm software, computing hardware underlie almost everything that we do. As an academic institution, of course, we're largely focused on academic capabilities of academia to support this new era. And there are three pillars, really, underlying the activities of the college.

One is supporting the rapid growth and evolution of core areas and computing, most notably, computer science and artificial intelligence. And in particular, two aspects of that. One is the tremendous growth in scale and breadth of core computing.

And the other is the very rapid pace of change in the content. And in fact, sometimes people say to me, well, gee, computing is used in everything. But so is math. What's different between math and computing? And the math that most of us use hasn't changed a lot in the last 50 years, maybe even century or two, depending on the math that we're using.

Whereas if you look at the machine learning that people are using, often those are results that are literally in the last months or couple of years. And so how do we really address this very rapid pace of change, particularly in AI, but in computing broadly? The second piece is that because computing really is a part of investigation in almost every academic discipline, it's not like the traditional sort of school-based and department-based structure necessarily can serve all of the needs in computing. And so the college cuts across all five schools.

And particularly with a focus on educating what President Reif has called computing bilinguals-- people who know computing and another discipline in which they're using it. And then the third core pillar of the college is social and ethical and policy issues in computing, where we really bring together humanist social science policy and technical expertise. Looking both at how are we responsible in deploying and developing computing, but also what are the opportunities to do good with computing? So the college has a very brief history. And one of the things that's really been sort of delightful about being back at MIT-- as Charlene mentioned, I had kind of a three decade hiatus from campus-- is that incredible problem-solving orientation of people at MIT. And so even in the COVID pandemic, with all these new things to address with getting the college off the ground, people have really rolled up their sleeves broadly, and we've gotten a whole new college up and running in the last couple of years, mainly in a time period where we were remote. So I started back in August of 2019, after the provost and others had started doing some planning.

We did some reorganization of some of the existing departments and labs and centers. And in particular, a big reorganization was in electrical engineering and computer science where the Department now has three components in it. Not just electrical engineering and computer science, but also artificial intelligence. Artificial intelligence and decision making or AI and D, as we call it. And then in January of 2020 we put this new organization into place.

And then we all know what happened in late February, early March of 2020. So we've been building this largely remotely. So the college, just from an organizational perspective at MIT, things that you're either familiar with or maybe you've heard a little bit about. Things like EECS, of course you're familiar with. It's a long-standing, major department at MIT. So EECS is jointly in the School of Engineering and in the College of Computing.

And as I mentioned, now has these three constituent faculties. IDSS, Institute for Data Systems and Society is in the college. Operations Research Center is a joint between the Sloan School and the college.

And then the Center for Computational Science and Engineering is in the college. And those are all academic programs with graduate and/or undergraduate degrees. And then on the research side, CSAIL, the largest research lab on campus, LIDS, the Laboratory for Information Decision Systems, and then some newer things. The Quest for Intelligence, MIT-IBM Watson AI Lab and the Jameel Clinic, all of which had started in the last several years, although predating the College, all come together under the college umbrella.

But in addition to those units, we created two new cross-cutting program areas that really go across MIT departments, labs, and centers across all five schools. First of those is the Common Ground for Computing Education. Each of these I'll talk about in a bit more detail. The focus of the Common Ground really is multi-departmental collaborations around computing education. Around, how do we really educate these computing bilinguals? And if you think about the structure of academic departments, they're fantastic at the depth that you need to really get a degree in a particular discipline. But when you look at things that cut across disciplinary boundaries, it's often more challenging for departments to come together and work on them.

And the Common Ground is really about, how do we do that for computing education? And then the Social and Ethical Responsibilities of Computing. There was a wonderful phrase, which has really stuck with me, which is, how do we incorporate multiple perspectives? Those of users, those of people who think about the broad implications of these computing technologies. And also, how do we use computing to actually improve lives of people? One of the main thrusts of the college is expanding the size of the MIT faculty by about 5%, by about 50 new positions. And those are split.

25 of those are core positions, which primarily are in EECS. And 25 of those are what we call shared positions with departments. In the first two years of recruiting, EECS hired a dozen faculty, and eight of those were new college positions.

And in this current recruiting season, so for people who will start next fall or sometime thereafter, because they might defer for a year, there are five searches in EECS, four of those on new college positions. The process of those recruiting searches is still ongoing. So we're not completely certain how many people are joining, but several of those searches have concluded with new faculty who will be joining in the future. And then the 25 shared positions, I'll talk a little bit more about.

But they're really about bridging computing. They're about this computing in every discipline on campus. And they're shared with departments from across MIT, from all of the schools at MIT.

And they're shared between some unit in EECS, at least one of these three faculties, or IDSS, and the Department at MIT. And this year, there are five of those searches that have been underway. And two of those have hired new faculty who will start over the coming year. So in identifying areas for these shared searches, we went through about a year long process to identify areas where computing was important, and that somehow tied together across the institute. So that we weren't just hiring 25 faculty that were in unrelated areas, but we had broader themes that mapped across the departments of MIT, but also had a big computing component.

So those are social, economic, and ethical implications of computing. And networks, and the things highlighted in red here are ones where we are currently undertaking searches. So those who are with SHASS and with Sloan. Computing and natural intelligence, looking at both the cognitive perceptual and linguistic aspects of that with School of Science, SHASS, and SAP.

Computing in Health and Life Sciences, Computing for Health of the Planet, Computing and Human Experience, and Quantum Computing. And again, in the Health of the Planet with Engineering and SAP, there have been ongoing searches. And every department was asked to submit proposals for potential search areas. And then all of the five school deans and myself worked with the provost to figure out what these six thematic areas are across the five schools. And so the searches that we do for shared positions all align with that.

And you can see that these areas are things that are often big priorities for MIT elsewhere, like around health and life sciences, around health and the planet. And some of them are more particular to the computing arena. Another aspect of the college that's been very important is looking at diversity, equity, and inclusion, and how we create a more belonging environment. Computing is not well known for necessarily having the most diverse of science and engineering student bodies and faculty. I'm very happy that one of the things we did this last year was to hire a new Assistant Dean for Diversity, Equity, Inclusion in the college, done in close coordination with EECS, because they're a huge department.

About 45% or so of the MIT students are majoring in some variant of a major that involves Course 6. And also at the institute level with John Dozier in ICEO office. And she'll be starting later this month and really working on how do we improve representation at all levels? But most importantly, how do we create an environment that creates a broader sense of belonging? And one of the things we did this last year is a colleague and friend of mine, Claude Steele, who served on the MacArthur Foundation board with me for about a decade, is one of the world's experts on really building diverse communities, as he studies what he calls the science of building diverse communities. And so we had a really wonderful session with Claude, again, here in the virtual world that we're living in these days. But it really, I think, helped raise a lot of constructive ways that we should be moving forward here.

Computing infrastructure across campus is also something that's critically important, and where the college plays a role. But really a sort of intellectual guidance role. The Vice President for Research, Maria Zuber, is really responsible in her operations and her research operations for computing infrastructure. But the college is co-leading on the academic coordination and on fundraising to help support research computing infrastructure. Two Petaflop-scale machines have been donated to the college for broad use at MIT. A Satori machine from IBM, and more recently, a Petaflop-scale AMD machine, which has just come online in the last few months.

And then, of course, space is always the final frontier in an academic environment. So we're also working at how to expand the green power supply space we have off-campus for more computing space. So returning to the computing bilinguals, I wanted to touch a bit on education, and talk about both the blended majors at MIT, blended between computing and other departments, and also on the common ground for Computing Education. But really what we're trying to do is infuse computing across the disciplines in two means. One are majors that bring together computing, usually the computer science part of Course 6 with another department or program. These are at the undergraduate level.

And then the Common Ground, which is focused on individual subjects that bring the forefront of computing together with problems and methods and understanding from different domains. And the Common Ground is aimed both at undergraduate and graduate level. And then the social and ethical responsibilities of computing research, which I'll also touch on, part of its mission in addition to research and engagement is a teaching component where they're looking at the social and ethical aspects of computing and how to integrate that in with all of the computing education that we do. So I want to start with the blended majors. The blended majors, which we're now referring to in that way, but they predate the college by many decades. In fact, 18C dates back to that time period that I was a grad student at MIT, which is math with computer science.

And just to give you a sense of the scale of these, it's a very good size major. There were over 100 students this last year in 18C. And then about a decade ago, CS and Molecular Biology, 6-7, was created.

That's been on the order of about 60 or 70 students at a given point in time. So also a pretty good sized major. And then three new ones got created, really, in that time period right around the launch of the college or after the launch of the college. So 6-14, which is CS, Economics and Data Science started in 2017, again, at about the scale of 100 students. Computation and Cognition between Course 6 and Course 9.

So EECS and BCS just launched in 2019, and already had over 100 students in it. So that's a really rapidly growing program. And then 11-6, Urban Science and Planning with CS, which is a good size major compared to the Urban Science major, and also very, very new. Just launched at the same time.

And you can see in the i-chart at the right, this is the craziness of 45% of MIT undergraduates in some variant of a computing major. But one of the really interesting things is that the blue and the orange and the light gray at the bottom are 6-1, -2, -3. And the things at the top are all of these blended majors. And you can see that the blended majors are growing, and in fact, the 6-1, -2, -3 are starting to decline. And this is one of the things that was sort of a hypothesis underlying the development of the blended majors is that there are a number of students ending up in particularly 6-2 and 6-3, because that's the easiest place to get the computer science background that they want. And if we start building majors that blend computer science with other areas, that will give students the ability to get the depth of computer science that they want, and the depth in the other discipline.

We're seeing that already getting proven. But there are some significant buts there. I think it's very important that we do this in places where there's real student and employer and graduate school interest, and not just create blended majors all over the place. And there are really also have to be faculty research and education collaboration across that boundary. Because if the blended majors aren't blended, and that's why we're starting to use that framing. If they're really just sort of like a double major, or a major and a minor, then there's no real reason for students to take them.

In fact, a major and a minor, or a double major in some senses, is more flexible, because you can always drop one of the two majors or drop the minor. And some peer institutions have tried some of these sort of CS plus x for various x types of approaches, and have not been so successful, because the students sort of voted with their feet and said, well, it's just as easy for me to double major in CS and something. Why would I take this? So we've put a lot of attention into blending these. And in fact, we put a hold on new such majors until we really understand more about how the ones that we have are working.

But I anticipate in the next few years, this is an area where we will be developing more blended majors as we see the demand. And the interest certainly seems to be there. So the Common Ground, now, is about individual subjects rather than undergraduate majors. But again, focusing on computing bilinguals. And the common in Common Ground is really addressing this issue that I mentioned briefly at the beginning, which is that classes and curricula tend to get created by individual departments. That's the structure of an institution.

And there can be things like the GIRs. But there, again, some department owns them, even if they're there in service of the institute broadly. At least with computing, there's a huge opportunity, and we think, need, to develop subjects that involve multiple departments.

And that's what the common ground is there for. There are three current focal areas under development in the Common Ground. Fundamentals of programming and computational thinking, machine learning, artificial intelligence, data science, and algorithms in the sciences, engineering, and social sciences. And fundamentals of computational science and engineering, what gets called CS and E, which is a lot of the large scale modeling and numerical techniques in the sciences and engineering. And there's a standing committee of about 30 faculty from all five schools, plus Open Learning, which sort of three subcommittees, each of 6 to 10 that are focusing on each of those three areas in addition to coordinating across the areas. And that group is charged with identifying pilots and then working with departments to pilot subjects, and then get the longer term commitments for those subjects.

And the role of the common ground is helping bring departments together, helping resource pilots. But the departments long term really have to be committed to these, because that's where the faculty are, and that's where the students are. They're majoring in programs offered by the departments. So I wanted to just briefly illustrate three of these. So one is Introduction to Computational Science and Engineering. It's an alternate version of 6.0002,

which has been developed by Dave Darmofal in AeroAstro and Laurent Demanet in Math and EAPS. And really, it focuses on, and the illustration here is when they first offered this, it was during the Mars landing. And so it's a perfect example of even at the most introductory level, in a class that a lot of first years take, in computational science and engineering, you're really not concerned just with the computing but with the physical manifestations of that. For example, in controlling the complicated multi-stage process of landing a vehicle on Mars.

Oh, and just one thing I should highlight, I love this student quote, because it's sort of to me what the objective of this class was. Balance of physics, math, and computer science. It's, again, this is the thing where physics does physics, math doesn't math, CS does CS. And how do we bring these together in ways that departments individually aren't as well set up to do? Another one that I'm personally super excited about because of my computer vision background.

This is the kind of way I wish I had learned linear algebra, and certainly my students had learned it when I was working with them, which really brings together linear algebra and optimization. The sort of computational side of how linear algebra is used at the core of much of machine learning, high dimensional data representation. And so, this subject has been co-developed by Ankur Moitra in Math and Pablo Parrilo in EECS.

And takes this very computationally oriented view. But again, it serves as an alternate to 18.06. So for students in 18, or in Course 6, and I think other departments will be adding this as an alternate, they can use this in place of 18.06. So it covers the same core mathematical concepts, but does them in a much more computational setting, including some problem sets it involves, sort of mini programming projects. So not just conceptually. And then the third one I wanted to touch on, and there are another three or four that are currently in development for this next year or so, it'll be sort of six or seven of these subjects by next year.

But this is the third one that was done this last year, is modeling with machine learning. And there, EECS was faced with this huge challenge in the sort of 2013 to 2020 time period where the enrollment in machine learning went from a couple of 100 students to like on the order of 1,000. But the distribution of those students changed radically from being 80% or 85% Course 6 students to being barely a majority of Course 6 students. And the students coming from outside really wanted to use machine learning as a modeling tool, rather than learning the guts of machine learning. And so this is a new set of subjects that have been being developed. And you can see it's a broad set of faculty from a number of different departments.

And it's an interlocking set of two six unit subjects, one of which is core, Modeling with Machine Learning, which all the students have to take if they're taking this. And then there are a set of options which they pick one of, whether they're interested in Sustainability or sort of Molecular Engineering, or Systems Modeling. They pick one of those. And that has had about 125 students in it this semester. And is a structure for these shared kind of common ground classes that we're looking at more broadly. So I also wanted to touch briefly on the social and ethical responsibilities of computing, or SERC.

SERC, as I mentioned, has this, it's the second crosscutting piece in addition to the Common Ground that's new in the college. And really focusing on the social and ethical issues in teaching and in research, and in engagement of the institute more broadly with the world. And on the teaching front, I just wanted to call your attention to something, which is a case study series that we started.

In that the first set of cases were published last month, I guess, in May, and they're online and they're available for everybody. And they're already being used in a number of MIT subjects. But the idea of these case studies, which are at and PubPub is a new publishing platform that MIT Press has just started running-- the idea of these case studies is to pick topics that are important around, and you can see these are all things we've seen in the press.

Right? Computing and privacy around data and mapping and neighborhood applications, facial recognition, risk prediction in the criminal justice system. And they dig into these more in-depth than you can ever see in the way the press covers things. But at the level of depth of an undergraduate subject. And they're designed to be used either as a case study in a technical subject that might be working on the relevant technology, or in a social science and humanities subject that might be looking at the social and ethical issues around these kinds of technologies.

Another thing going on in sort of coordination with this is development of active learning projects, where we take existing technical subjects like 6.170 Software Studio, 6.034, Intro to AI. And we have people in SHASS, postdocs and graduate students engage with the teaching staff of the technical class in developing materials that really give students the ability and the need to actively engage in questions around social and ethical issues related to the problem sets and to the projects that they're working on in the classes. So the idea is really to embed this in the technical curriculum.

On the research front, one of the things that SERC has been doing is developing what they call an ethical computing platform that research groups can use to ask and answer questions about the design of their research, how they're collecting data, how they're protecting data, what potential unintended social consequences might be of what they're doing. Right now, this is really just at the level of a protocol that gets followed. But they're actually implementing this now, also, as a software platform that can be used to help research groups in tracking these things. And that software will then be made available under open source for anybody to use.

So SERC is really looking at the full gamut of research, external engagement and teaching, and the social and ethical responsibilities cutting across at MIT. So I wanted to touch briefly on two things that are a little more recent at MIT than the things that have been around for a long time, like EECS or the Operations Research Center. One is the quest for intelligence, where Jim DiCarlo has just stepped in as the Director of that, stepping down as the head of the EECS Department. And this is really, the Quest now is making really a substantial bet on this premise that artificial intelligence, sort of engineered systems, and natural intelligence, sort of scientific understanding, are really interlocking aspects of a single collaborative grand challenge that brings together science, engineering, and application. And they're in the process of developing a set of missions with faculty that span across these areas to pursue things that bring together artificial and natural intelligence. The Jameel Clinic is another thing that Regina Barzilay is heading, which has really developed a lot in the last couple of years, like the Quest has.

AI in health is around MIT in many different aspects. But when you really look at AI in health, and particularly machine learning in health, the Jameel Clinic serves as an epicenter in tying academic research to the practical impact, to working with hospitals and doctors on how it is that we can both translate problems from their domain to ones that can be studied at MIT, but also work that's done at MIT back into practice. So the thing I wanted to close with is a little bit of more on the eye candy front, which is the new home for the Schwarzman College of Computing building on campus.

And so along here is Vassar Street at the bottom. These are the railroad tracks. This is 46, the [? PCS ?] building.

And this is where the new Schwarzman College building is going. It actually goes out over the tracks. This is a picture of the site from a few weeks ago. So this is looking from the sort of building 36 kind of perspective over toward the Albany Street Garage on 46. This is from Albany Street Garage back over to 34th [INAUDIBLE] the data center there in the background.

Gives you a sense of the size of the site and the location of it. This is where the old Cyclotron building had been. And this is the south facade facing onto Vassar Street.

And you can see that it's a very open, welcoming space, because this building is supposed to be bridging all of MIT. And it's also on the north end of campus. And so it's also aiming to draw people over across the chasm of Vassar street.

But at the same time, it faces south. And so one of the things, it has very innovative design for south facing facade, where there are multiple layers of glass. And these glass shingles on the outside, actually, this is a construction technique that's used quite a bit in Europe and other places, and less so in the United States.

But this outer facade of glass is separate from the main exterior facade of the building, which gives you a lot of daylight and very low solar heating load. This just sort of shows coming out of the building, and the sort of entry sequence-- oh, sorry, wrong direction-- entry sequence. This just shows you the lower floor. The first two floors are much smaller, because of the railroad track behind them. So this is mainly a classroom and some convening spaces. Just some renderings of the lobby area and the stair going up to the second floor.

Kind of a nice kind of seating area in the lobby, looking back over to 34, 6 and 8. This is the convening space down on the first floor, overlooking Vassar. The classroom on the first floor. And the second floor has a bunch of tutoring and project space room for student projects, which is something that's really been missing particularly in all of these big computing majors. About a 250 seat lecture hall.

And then as we go into the upper floors of the building, they're more open and raw space, which is the research space, with some finished areas for collaboration that face out, again, onto Vassar Street. And then just interior offices. And then on the top floor, there's a big event space, sort of analogous to the space at the top of the Media Lab and the Sandberg Center space. And because of the height of this building, it actually looks across the whole MIT campus [INAUDIBLE] to Boston and beyond. So the view from up here on the eighth floor is tremendous.

So with that, thank you all for your attention here, and we'll turn it over to questions. Thank you. VICTORIA GONIN: Thank you, Dan, for a wonderful presentation.

My name is Victoria Gonin, and I'm the Executive Director of Alumni Relations here at MIT, and I'm delighted to be here today as your moderator. Please continue to submit your questions on the Q&A tab. Now for our first question. What is the leading edge research progress in quantum computing at MIT and around the world? DAN HUTTENLOCHER: So quantum computing is a field that, there are a few fields like this.

One of which I work in, computer vision. So quantum computing, since the time I was a graduate student in the 1980s at MIT, has been about to happen. And I do feel like it's different now. It's about to happen, and I think that it's more likely to really be about to happen. Computer vision has actually happened in the last five years, but it was about to happen for about 25 or 30 years as well. But quantum really is starting to advance, and across the whole stack.

So At MIT, we have people working in the physics and the chemistry of quantum computing. We have people working in the quantum computing hardware, in the software and the algorithms. It's really that whole stack from fundamentals of quantum physics, quantum chemistry, properties of quantum materials, all the way up to putting that into machinery. And then, what are the algorithmic and software consequences of that? One person I would really call attention to, who just joined the MIT faculty, but has been at MIT-- the main campus faculty, but has been in and out of MIT for quite a while-- is Will Oliver, who's in the EECS Department, who really works across quite a broad range of that stack. Many people are more specialized at one level or another, because there's so many different domains involved there. Will, in addition to being a professor in EECS, is the senior member of the research staff at Lincoln, and has been at Lincoln for a long time working on quantum.

But in addition, is now, and still does some work at Lincoln Labs. But also is on campus. And so he's a perfect example of working across that entire stack.

VICTORIA GONIN: Thanks. Next question is from Harvey, the class of 1958. Why no Course 15.6? After all, computers dominate business processes these days, for better or worse.

DAN HUTTENLOCHER: Absolutely. Terrific question. And really, the existing blended majors happened in a sort of organic manner. And what we're doing now is, as I mentioned briefly in passing, we've sort of put a hold on new blended majors while we sort through the fact that we created three of them in a sort of two year time period, and we really need to understand how they evolve over time and how we create them.

But I would agree with the assessment that a 15-6 or 6.15-- and that's another question, like there's 11-6 and there's 6-14 and 6-9, Why are they numbered the way they are? There are lots of things. But I think that to me, personally, my Cornell, before I became the Cornell Tech Dean, my faculty point was actually joint between the business school and computer science.

So I'm a firm believer in that interface. And in MIT, we have good collaborations between people in 6 and in 15 on the faculty front. So I think that will be a natural one. But I would guess probably for the next year, or even two years, we won't be starting new blended majors until we-- we really don't want to fall in the trap that some other institutions did of launching a whole bunch of these and having them not really meet the expectations of the students. Having them essentially be a double major. And so that that's something that we're working hard on.

That's a great question. Thank you. VICTORIA GONIN: Question. Are these new blended majors more popular at the undergraduate level or the graduate level? DAN HUTTENLOCHER: So the blended majors, they're all undergraduate. They're all undergraduate majors. They are supposed to all have an [INAUDIBLE],, because there's at least the ones that have a 6 in the, the intent is to get all of those to have an [INAUDIBLE] degree as well, because Course 6 has an [INAUDIBLE]..

So that's just a fifth year for MIT undergraduates, not a separate graduate program that non-MIT undergraduates can apply to. That's not true of all of them right now, and so that's, again, one of the things that we're working on before we start rolling out more of them. Is that something that is going to be a property of all of them? If it's not, why not? Et cetera.

But what I call terminal graduate degrees-- the PhD, and in some fields, the Masters, depending on the field-- those really are things that are at a sufficient level of depth that blending doesn't make as much sense. So it's really the undergrad and the [INAUDIBLE] degree that we're looking at there. VICTORIA GONIN: OK.

How are your faculty thinking about ransomware in the current wave of attacks? Do you think banning cryptocurrencies will help address this phenomenon? DAN HUTTENLOCHER: So this is a terrific question, and it's also a question that I think has two very different answers. So one answer is an answer around what's going on in the sort of world at large today, and the other is the mapping of that back onto the kind of research that happens in an institution like MIT. So when we look at, there are definitely subjects at MIT that look at the practical issues around securing software, around not just ransomware, but all kinds of software security. Ransomware is only possible because the software is riddled with insecurity. And so there's a big focus, both in education, and in research on cybersecurity, which is not specific to ransomware. When you get to a particular sort of use-- I hate to call it a use, but it's become one-- those tend not to be studied as much in the academic realm as the fundamental things like flaws in software that allow things like ransomware to exist.

So cybersecurity is a big area, both educationally and research-wise at MIT. And it really has to do with securing software and making software more trustworthy. And it spans both the computer languages that are used to develop software, what's called programming languages, and also the development of large computing systems. Then on the cryptocurrency side, that's something that, again, the cryptographic basis of cryptocurrencies is something that MIT has a long history in leading in cryptography, and what's called public key cryptography, which is used for all online transactions to protect transactions and information. And so that's again, a fundamental piece of work that MIT has long been leading in. And then work on cryptocurrencies is also done to some degree at MIT, but tends to be, again, researching new cryptocurrencies and less around sort of the policy issues around cryptocurrency.

And that's a place where we're starting to change things some at MIT. So there's a new initiative in the college called the AI Policy Forum, so it's not focused on cryptography, it's focused on artificial intelligence. But that's a place where we're getting much more focused on policy issues, because AI policy is so important. And then there's something that's been ongoing at MIT for maybe a number of years now, maybe in the five to eight year range, called the Internet Policy Research Institute, IPRI, which also has been looking at some of these broader cryptocurrency and other sorts of internet policy issues.

But I don't think cryptocurrencies will or can be banned. They're just, they're there. They'll be used.

The question is how far underground. You ban them, they'll just be completely underground. They're not going to go away. So my personal belief there is that there are a lot of regulatory issues that are important around cryptocurrencies, but I don't think outright bans are going to be practical of sort of almost any technology once it's out that widely used. VICTORIA GONIN: Interesting. Next question, are there research papers coming out of the MIT Quest for Intelligence group? If so, can you identify a few of them? DAN HUTTENLOCHER: I think my internet connection is doing something strange here.

Hopefully, hopefully this will get better. VICTORIA GONIN: Yeah. Did you hear the question? About-- DAN HUTTENLOCHER: Yes.

Sorry about that. The Quest for Intelligence has had some research papers come out, but not with this newer thematic approach of really combining the science and engineering of the applications of intelligence, because that's something that's really just been developed in the last semester or so. But there are extremely relevant pieces of research about components of that.

So CBMM, which is the National Science Foundation, Science and Technology Center, the Center for Brains, Minds, and Machines, has been very much motivation for the science side of the science engineering and applications of intelligence in the Quest. And that's been ongoing for about eight years, and a number of very interesting papers that have come out of that. And then, of course on the engineering side, there are a lot of pieces of research out of CSAIL and LIDS, the Laboratory for Information and Decision Systems on the applications of AI.

VICTORIA GONIN: Good, good. Regarding the Circle Project, do you think that science fiction authors should have a role in examining the ethics of technological advancements in computing? DAN HUTTENLOCHER: That's a fascinating question. I do think that our understanding, and therefore, often, our misunderstanding of technology and what it does and doesn't do, can and can't do, what impacts it might or might not have, does come from fiction, science fiction being the most important when it's futuristic kind of technologies.

But it's also, what makes fiction is that it doesn't necessarily have a societal objective or aim. So should is probably too strong a term from my point of view. I don't think we should be prescriptive about what should be in fiction. But I do think the more that people understand, including science fiction authors and the world at large, that a lot of our societal preconceptions about and understanding about technologies does come from fiction, and often is therefore, can be very misleading. So I think one example of that is that if you look in sort of popular culture today, I would say that one of the biggest worries that people have about AI is that somehow, AI will take over and subjugate humans. Or killer robots will arise.

And that is very firmly in the realm of fiction. There is zero basis for that in any sort of scientific or technological work. This is the problem. As we've all seen with science, scientists will never say something's impossible, because that's not the nature of science. But there is nothing to suggest that will happen in the science or the engineering. And everything about the science and engineering points in the completely opposite direction from that.

But that's not our understanding, because it makes such great science fiction. So I do think that these are issues where, I would reframe it to say, who are any of us to say what a fiction author should or shouldn't do? But I do think that society broadly should be aware of the fact, and we really do have a collective responsibility as a society to differentiate fiction from the way these things really, what are the real issues to be worried about. Because there's a lot to worry about AI, and if people are spending their time worrying about stuff that pretty much won't happen, they're not worrying about the things that they should be.

VICTORIA GONIN: Yeah, interesting. OK. As a grumpy old alum, I'm still trying to understand what the purpose was for making CS a formal school. Is it just that we've gotten too large and pervasive to manage it as a department? Fundraising? Other reasons? DAN HUTTENLOCHER: Well, I'm going to I'm going to take the mantle of a grumpy old alum myself, since I got my PhD in '88.

Not a spring chicken. But I think there, that's a great question. I think that there are a few reasons underlying this. So one is that it's quite deliberate that the college is not called a school.

It's called a college, and it's a pretty different structure from the schools. So when you think about a school, in some sense, it's about the depth in a set of academic disciplines, right? In the sciences, in engineering, in management, in architecture and planning, in humanities, arts, and social sciences. And when you get deep, you tend to get siloed. That's just a fact. It's very hard not to.

But the flip side of this is that-- and so the way academic institutions tend to be organized is schools, which have departments in them, which are deep and fairly siloed, and then there are interdisciplinary things that cut across. And it's very hard to connect those to each other. So the college is really trying to have the depth of computing and the breadth of these cross-cutting programs in the same organizational structure. And now I'm going to hazard trying to draw a bunch of diagonals here, connecting the depth and the cross-cutting piece. And that's what the college is about. And if I'm successful as a Dean, if SERC is successful, if the Common Ground is successful, that's what they're doing.

They're knitting together the depths of computing with cutting across all of the schools and all of the departments at MIT. And so that organizational structure is there for that reason. Now, are there also practical issues about something that has, on the order of half of MIT undergrads and 20% of MIT graduate students? Yes, there are. But that alone to me would not be a good reason for creating something new.

I think it's that we're really, it's back to this issue that I mentioned at the beginning. That especially at a technical institution like MIT, almost all of us use math. But the math that we use is not recent math. And I don't see this as going away. Maybe at some point in the future of computing, we'll stop changing so quickly.

You can decide that we don't need a college of computing, or maybe it becomes another school, or who knows what. But the last 50 years, that certainly hasn't happened. It's just gotten more and more rapid. And I think the next 50 is going to look that way, and I'll be dead by then anyway. But so I think this is not a short term issue, because I also wouldn't want to create a part of MIT for something short-term. So I hope that's helpful.

VICTORIA GONIN: Next question, artificial intelligence gets a bad rap. Eliminating jobs, abuse by law and border enforcement, et cetera. Are you bullish on its prospects despite all this? DAN HUTTENLOCHER: Yes. I'm bullish, but I'm bullish with real concerns about how we proceed.

And you know, I think that the bad rap that AI is getting, and frankly, tech is getting more broadly, the tech sector is getting more broadly, there was a lot of Kool-Aid drunk starting before the dotcom boom. And then there was sort of a bust, and then there's sort of been this AI boom. But I think people have looked to computing technology, the internet, and then AI to somehow be panaceas.

And they're not. No technology is. All technological advancements are, so to speak, a double edged sword. They have positive and negative impacts, and we really have to pay attention to both of them. And so I think that that over-optimism, that overclaiming, overset of assumptions about it solving the world's ills, that there's a natural backlash in the other direction. That certainly hasn't been happening.

And so I think now, in my view, an over-focus on the concerns. And I think really, the right place to be is in the middle, where we're looking at the opportunities. There are a lot of jobs that can be made much better by AI. And in fact, if we look at the history of automation, which is all that's going to take jobs away, over the last 150 years of modern automation, whole new types of jobs that no one could have envisioned have been created.

And they're often higher-paying and more interesting jobs than the ones that were replaced. But that change is extremely disruptive. Right? People who are mid-career and losing their job because of automation aren't the ones who are going to get the new jobs, because they need a whole different type of training. So there are big societal issues to address and to grapple with the kinds of changes, for example, in jobs that happen.

On law enforcement, I'm a computer vision guy. I know a lot about facial recognition. And I can just say that to ban or throw out the best technology we have for solving human trafficking, child exploitation, missing persons is a fatal mistake in my view. What parent are you going to say, well, we know your child was abducted, but we're not going to use the best technology to try to find them.

But the flip side is that the privacy issues and the misuse of this technology and the potential for bias in this technology are very real. Affect real communities, do disproportionately affect people of color. And those are real issues that need to be addressed. But this absolute view that it's going to solve all the world's ills or that we should ban it are, frankly, both beneath people with an MIT education, in my opinion.

We need to be the ones to educate people in the how to use these technologies to really address human needs. So a great question. Thank you. Sorry, I got on a bit of a soapbox there.

VICTORIA GONIN: That's good. So our last question, which I think is a perfect ending question. When will the building be finished? This is coming from Joel, from the class of 1967. And he wants to know if the event space will be available for his reunion in May of 2022. DAN HUTTENLOCHER: I wish I wish I could answer that in a positive way. So the building starts construction this summer, and it's a two year construction project.

So summer of '23 is when the building will be finished. And then, who knows when spaces can be booked. I think, rightly, we will be very conservative about not booking event spaces till we're sure they fully are operable. So even like if somebody was doing something in fall of '23 for the big event, we would not probably book that till we're sure the space is really working. But the objective is for people to be moving in before the fall of '23. So thank you very much.

VICTORIA GONIN: Good. Exciting. So thank you Dean Huttenlocher for joining us today and for sharing your work and your vision for the College of Computing. It's very exciting.

And I just want to close by saying these talks are going to be made available by the MIT Alumni Association YouTube channel as soon as possible. And for those of you who are attending reunion events live, we encourage you to take a quick spring break, and then we're going to be offering an opportunity for table chats immediately following this session so that you can go deeper on the discussion with your colleagues attending today. So thank you all, continue to enjoy Tech Reunions, and we look forward to seeing you. Thank you. WHITNEY ESPICH: Thanks for joining us.

And for more information on how to connect with the MIT Alumni Association, please visit our website.

2021-07-05 08:12

Show Video

Other news