Episode 144 - Fred Chong: "Never been a day I thought this was the wrong job."

Professor Fred Chong is the Seymour Goodman Professor of Computer Architecture and is also a fellow of the Institute of Electrical and Electronics Engineers. In this episode of the Course, Professor Chong shares how he witnessed the evolution of computers and A.I., while finding a niche, quantum computing and computer architecture, for himself to dive into as a research pathway.  As a faculty member at the University of Chicago, he finds joy, meaning and impact in teaching and guiding students. Tune in to listen to Professor's career story.

Transcript

Stephen 00:00
Hello, and welcome to The Course! I'm your host, Stephen, and today I'm speaking with Professor Fred Chong of the Department of Computer Science. Professor Chong is the Seymour Goodman Professor of Computer Architecture. He's a fellow of the Institute of Electrical and Electronics Engineers, a distinguished member of the Association for Computing Machinery, and has won Intel's Outstanding Researcher Award in addition to awards for both graduate teaching and undergraduate teaching.

He's here today to tell us about the nuts and bolts of the machines that power our world, describe a few serendipitous moments that have brought him to where he is, and tell us how he became a University of Chicago professor.

Professor Chong, thank you for joining us on The Course. How are you this morning? 

Fred Chong 00:41
Very good. How are you? 

Stephen 00:42
I'm doing well. Thank you. Before we get too deep into anything just wanted to ask you to please quickly tell us your position at the university and just a little bit in a layperson's terms about what you study. 

Fred Chong 00:54
Great. Yeah. I am the Seymour Goodman Professor of Computer Science, in the Department of Computer Science. I also happen to be a chief scientist of a company called Inflection, which acquired a software company, a student of mine started two years ago. 

What do I work on? I work on quantum computing, and in particular, I tend to focus on the software that runs quantum computers, and we'll probably get into this a little bit more later, but, so quantum computers basically are a little bit different than the computers we're used to.

They use quantum physics to do their computations and can do certain specialized things better than conventional computers. And what I do specifically is build software that is really tailored to the physics of the machines, so we can get the most out of those machines and hopefully solve interesting problems soon.

Stephen 01:53
Awesome. Well, yes, I definitely look forward to hearing more about what you're up to on that front. But going back into history, when you were a kid, maybe middle school, early high school age were there signs that you were going to end up studying quantum computing or what did you think you were going to do with yourself?

Fred Chong 02:12
Well, I pretty much knew I was going to study computer science. I'm, you know, sort of one of those kids, you know, when I was whatever, seven or eight, I started with a computer. I'm old enough that that computer had these little buttons that look like pieces of gum, something called the chiclet keyboard.

Stephen 02:29
Oh, wow. 

Fred Chong 02:30
Like back when they didn't have real keyboards, they were just buttons. And in fact, it didn't have a disk drive or a hard drive or a flash drive that had a tape deck that's like you put like the ancient tape in it and it played little beepy sounds to load the program in. So I started with those and it was, pretty much an amazing thing to me, you know, sort of this the ability to create software programs that did things you wanted. 

And you know, so that's how I started. I went to school, I went to MIT and my parents thought I shouldn't be a computer scientist. They thought it was sort of a fad. And that I should be an electrical engineer. And luckily for me, there was a department of electrical engineering and computer science, so I could just sort of be a computer scientist within the department they thought was okay.

And then when I took a course in my sophomore year, where they taught you how to build a computer. And back then we built it with little plastic parts on this little briefcase we had with lots of wires. what I learned from that, which was this really amazing thing, that the software that we, that I've been writing all that time, it gets translated into little strings of ones and zeros.

That's how computers think, represent things. And what was amazing to me is the way we built the machines, those ones and zeros actually went all the way down to the wires to little switches. And it just said this switch should be on or off. That's what the zero and one meant. And that made the computer actually compute things. And so this like concrete connection between what I've been doing, you know, for many years, writing software and how the hardware actually works was like a huge deal to me. And that's, you know, my, I think my actual position here is like Seymour Goodman Professor of computer architecture. 

So, computer architecture is like basically how you build, design the hardware and how it works with the software. And that's the thing that I got really interested in my software here in college.

Stephen 04:45
That's very cool. Okay. So, you already had plenty of experience writing programs, but we're talking about actually, the literal tiny, but real, like physical things that are carrying out the programs. 

Fred Chong 04:57
That's right. That's right. Yeah. So to me, that was the, like this intuition of how machines work and why they're fast or slow and why they cost as much as they cost. And it's like that. 

Stephen 05:10
So what did you at that time think you were going to do with that? Did you kind of assume that you'd be like going into industry? Did you have a real desire to study computer architecture? Like what did that spark in you? 

Fred Chong 05:25
So I had a real desire to study computer architecture. I thought I would go to research lab, like AT&T Labs or IBM or something like that. That was incidentally also the time that AT&T got broken up from its monopoly and didn't, and then thus didn't have quite the resources to have the kind of research lab that they used to have. I'll tell, I mean, I'll tell you what happened when I went to look for a job. 

A friend of mine actually, who's also a professor now, he suggested, hey, you should try interviewing for a faculty job. You'd be good at it. And I was like, all right, I'll try that. And, you know, the reality is it's much easier to become a faculty member and then move into industry than to do industry first and move into being a professor. And so, it was sort of like, well, I guess I should. you know, interview and see what jobs I get offered and try it. And I discovered it was really the job for me. 

I'll tell you a little interesting anecdote, which is that about the first or second year that I was a professor, I went to visit a friend of mine at MIT who's very successful.

But I was talking to him and he was like, you know, I love this job, but some days I come in, I'm like, what did the heck am I doing here? And I thought, I looked at him and I said, you know, there's never been a day that I thought that I was in the wrong job. And now like 25 years later, still it's the same thing.  Never been a day that I thought was the wrong job.

Stephen 06:55
All right. Well, that's a great sign. That's always good to hear. I've never had someone come in and say, I definitely think I'm in the wrong job, but you know, it's good to hear the positive. Like what problems, or just like aspects of computer technology were interesting to you at that time. I mean, it occurs to me that we might actually be talking about a period before a lot of our listeners were born. 

Fred Chong 07:20
That's right. This is like

Stephen 07:23
Yeah, like what was the state of the field at that time? And what were, you know, what drew your interest? 

Fred 07:28
Yeah, this is like the 1980s and 1990s. Right. In the 80s, I think the hot thing was, you know, AI, but it was starting to go enter into a time of, you know, maybe disillusionment, which is interesting because now we're in a period where AI is a huge deal again. but it was at a time when I think we thought that the AI was going to solve all our problems.

And what we discovered at that time was we were only good at very specialized things like computer vision or, you know, certain kinds of expert systems that advise you of very specific things. So I think we learned that AI taught us specific techniques that we had worked on, or that we had developed things for that, and that it was a long time, pretty much until recently, that we got to solving more general problems with AI, partially because computing power is much more powerful.

And so, instead of going into AI, I sort of went into parallel computing, so, or multiprocessors, so, you know, your machine on your desktop used to be, it would run one thing at a time. And actually, these days, it runs many things at a time. At that time, it was one thing at a time. And what, you know, what people discovered was, well, there's going to be a limit to how fast we can make a single microprocessor, a single what we call thread of computing.

And instead we should work on, you know, doing many things at once with multiple microprocessors or what we now call microprocessor cores. So each chip you have as many processors in it. And so I can do many things at once. And that's what I studied initially when I was in grad school and when I first started my first faculty job. 

Stephen 09:14
As I always do when I'm, you know, we're, talking about like that period of history or whatever, I have multiple Google Chrome tabs, I'm running Slack. I got Zoom, of course, and a recording software. The idea of not having that capability is almost frightening now. 

Fred Chong 09:32
Absolutely, I mean, your iPhone is as powerful as a supercomputer of that age. 

Stephen 09:39
How does that make someone who's been in this field for that amount of time feel, or do you, does it make you feel old or does it mean you feel wow, we’ve came so far in time.

Fred Chong 09:50
I think it's great to see how far we've gone. What's interesting, okay, so we can start getting into, you know, why do I work on quantum computing now? There are a few reasons for that. One is the advance in conventional classical technology is slowing down. And so, you know, we have to find new technology drivers for computing.

And we're so used to computing getting, you know, improving so quickly exponentially really over time that if we stop being able to support that, you know, society and our industry will be very different, So one of the things about computer architecture is very driven by technology and for a long time was driven by, you know, this classical CMOS chip technology.

Now we're looking for new drivers, the other thing I'll say is you ask, you know, about, you know, different times of history and, you know, my view on my experience with it. I would say that when I was, you know, starting out in the 80s, I sort of felt like if only I'd been alive in the 1950s when computing first started, you know, everything I thought, I would think of would be new.

And, you know, I would have had so much impact in the world if I had just been here when they started. And what's exciting about quantum computing is it's basically analogous to the 1950s for classical computing. You know, we're just in the beginning and, you know, everything is so new and everything we work on makes such a big difference.

You know, so I think, it's a historic time, really, in computing, and I mean, of course, it is also historic in certain ways for AI, but for quantum computing, it's a historic time. It's like an exciting thing to be working on right now.

Stephen 11:31
Yeah. Could you tell us, well, first of all, could you, I, you did give a pretty good, I think, primer at the top, but just in you know, basic terms, what is quantum computing and also curious, like when it started to come across your radar, when it started to occur to you that was actually like something you were going to be spending a lot of time on. 

Fred Chong 11:53
Sure, classical computers, you know, sort of, you can sort of think of them as computing with switches or that are zero and one. Right. So they basically compute with zeros and ones and quantum computing is interesting. What you're really going to do is you're going to those zeros and ones in classical computers are represented, essentially on transistors that have zero and one. electrical charge, essentially. 

What quantum computers do is they use some sort of quantum physical phenomenon. So let's say we're looking at the nucleus of an atom, and we're looking at what we call nuclear spin. And so the spin can be in an up direction or a down direction. And we decide, okay. up is 0 and down is 1, for example.

So we decided we're going to use something else to represent the 0 and the 1. And what's interesting about quantum physics, sort of quantum mechanics, is that the 0 and 1 are not, you're not only in 0 and 1, you're sort of simultaneously in 0 and 1, it's called superposition. And so you sort of, have some chance of being in zero, some chance of being one.

And then at some point, what you do is you basically ask the atom whether it's in zero and one. And it forces it into zero or one, but with basically some probability of one or the other, some, you know, some fraction of the time it'll be zero and some fraction one. It's like coin flips, right? Except that it's like, you can think of the coin as sitting on its edge all the time until you look, until you make it be one side or the other. 

So, why is this useful? Well, it turns out if you have many, many of these quantum bits, zeros and ones, they can represent an exponential number of numbers. So when you have, you know, say 10 bits, you can represent a thousand different numbers. So two to the 10th. Okay. Now in a classical computer, you can only have one of those numbers at once.

In a quantum computer, you can have all of those numbers at once. And then you write programs to compute things on all those numbers at once. Now, you know, so there's the thing that makes, some scientists and nation states interested in this is, there's a problem in cryptography.

It's in what's called public key cryptography, so something called RSA. And the way RSA works is you need to be able to compute something easily, like sign a document, but you need to make it hard for people to forge the signature. And the way that works is you have what's called a one-way function.

So it's easy to compute in one direction, but hard to go backwards. Okay. So there's no such thing in the world as a one way function that's truly one way, but what we do in computer science is we use what's called, we use a mathematical function that's very easy to do one way and hard to do the other. Right? 

So, what we use is what's called the product of two primes. So you take two prime numbers. So, if people don't know, prime numbers are things you can't break into smaller pieces. Can't factor them into multiples of two things. So like three is a prime, five is a prime.

Stephen 15:06
We're talking significantly larger though, right?

Fred Chong 15:08
Yeah. So we would use really big, like, you know, a thousand bit numbers.

So really big numbers. So you use really big number and you multiply the two of them together. So easy to take three and five and multiply it into, and you get 15, if that number is really big, if instead of 15, you have like, you know, some, you know, value in the millions or something, it's hard to factor that to find the two numbers that make that, right? So you have the, so that's the one way function. 

So for classical computers, it takes an exponential amount of time, which means if you have a, you know, 1000 bit number, it takes two to the 1000 time. So a lot of time, like years and years, thousands of years at a time to look for the two numbers that make that number. 

So quantum computers do this much faster, arguably in hours or days or something like that. And so the idea is you can break cryptography with this. That's sort of why there's, there has been a lot of interest in building these machines from a national security point of view, but they're also really exciting for basically understand like simulating chemistry and molecules and materials, making better batteries, making  better photocells, things like that.

There's a grand challenge problem for Microsoft, which is about making a more efficient process to make fertilizers. So it's a problem called nitrogen fixation. You know, so there, there are certain class of problems in computer science, which are, they scale exponentially, which means as the problem gets bigger, it takes, you know, every little bit that you make the problem bigger, it doubles in it's how hard it is to, or how much time it takes to compute.

Those problems are things that we can't approach very well with conventional computers, even supercomputers. And so, the idea is that quantum machines will be hopefully be able to solve certain ones of those problems and allow us to work on some of these things that we call them intractable. They're intractable problems.

Stephen 17:12
Yeah. and I mean, I can obviously see how someone who's already studying computer architecture would be interested in that, but, can you just tell me a little bit about how you actually. 

You know, went down that path, like when this started to become something that people were seriously working on.

Fred Chong 17:28
Yeah. People often ask me, you know, in my career, how did I see ahead and pick the things to work on that end up being like sort of interesting, exciting later? I will tell you that it's mostly serendipity and opportunity like that, you know, you have to be interested in the opportunities, but they just come up serendipitously. 

So it's a combination of two things. Number one, my emphasis in computer architecture has always been, as I mentioned, technology, and in particular, new technologies that will shake things up, right? Doing research in computer architecture can actually be made at various times not so exciting if you're just working on the conventional thing, and you're actually a little bit of a slave to market forces and what the industry is already building because what they're building is what we're good at, right?

So like, you know, we build processors in a certain way. We build memories in a certain way. It's hard to change that, because they're, you know, billions of dollars invested. So I think it's interesting to look at new things that will change that. So what we think of as the commodity trends, like what are, what are the things that are being manufactured and sold the most?

Right. and then, the other sort of serendipitous thing is I happened to go to school as an undergrad with one of the pioneers of quantum computing on the physics side, his name is Ike Chuang. He literally wrote the textbook that everyone uses. And, in about 2000, 2001, you know, they had been building some small versions of these machines and he thought, oh, you know, we need architects to think about what happens when these machines get bigger.

How do we design them? You know, more than like three or four quantum bits. Once it gets bigger, what should it look like? And what does the software look like? And things like that. And so he decided he would build a team of people to look at this and he asked me to if I was interested, and then he personally sat down for like an entire day and tutored me in the basics of quantum computing. So people ask me, how do I get into this? And how did I learn, how did I learn any of this stuff? 

And I'm like, well, you know, I sort of had this serendipitous advantage. Where some of the where the person who is the best at teaching this sat down and personally taught me.

Stephen 19:54
That helps.

Fred Chong 19:57
Absolutely.

Stephen 19:58
Well, yeah, I'd love to hear I know we're probably skipping ahead, like a couple of decades, maybe, but I would love to hear a little bit about what your like experience is like teaching, this kind of thing and, and who are you working with? I mean, are you still doing a lot of kind of like basic, like intro to architecture type things? Are you mostly working with graduate students on like bottles for quantum computing. 

Fred Chong 20:21
Sure. I mean, I do have, you know, a pretty strong graduate emphasis. I have, you know, a dozen graduate students in my group. And we have one of the largest sort of quantum computing certain software systems kinds of groups and, you know, I produce a lot of graduates that become professors themselves, which is pretty exciting.

And we, you know, it does take a certain amount of background to work in this area. So my grad students most of them are actually have dual degrees in physics and computer science because that's what we do. We sort of sit at the boundary there and try to sort of connect the two, right? Create software that's, that's very optimized for the specific physics of specific quantum machines and devices.

Stephen 21:15
And sorry to interrupt. Am I correct that this was, I mean, this was sort of theorized by both physicists and computer scientists well before there was what was necessary to actually build one. Am I correct about that? Or I don't know.

Fred 21:28
Yeah, that's right. That's right. You know, sort of famously Richard Feynman sort of proposed this in, I don't know, the 80s or late 90s, maybe 1980, about 1990 or so. Can't quite remember it, but there was a, basically the suggestion was, hey, you know, we have all these problems in physics that we want to understand. 

Our computers can't model them. Maybe we should just build a quantum computer that could, you know, model quantum physics really well. And so that was the original suggestion. And then it took quite some time to sort of build these machines, you know, around 2000, 2001. My colleague, Ike Chuang, built a machine that demonstrated some of the concepts and then it took a, you know, 20 years, really, to get to the point where we were really building these machines in industry and having them, you know, you can log on to them on the web and use them now.

So but it turns out, that in some ways, what I teach about quantum computing is perhaps a little bit more accessible than the basic physics of the machines. It's, you know, we try to look at some of the physics that would help us, but it's really a computing model.

So, you know, I teach a graduate class it's an, a sort of an advanced class, but some number of undergraduates do take that class. And then I also teach undergraduate classical computer architecture, which is, you know, the thing that got me into my sort of field and research to start with.

So something, sort of close to my heart and I think that, I guess, I think that because I like what I do, it has a positive effect on my students. A lot of my students, as I said, become professors. I think because I, you know, I love my job. And I think I was very surprised the way things work here.

I was, my students secretly nominated me for the Graduate Mentoring and Teaching Award. And, so I was very surprised when I won it, which I think is really cool. It's probably one of my best awards because it means that my students, my grad students are actually happy. 

You know, they seem to be happy, but, you know, they, that it was sort of neat and then actually last year I won the Undergraduate Teaching Award, which is also a surprise to me. But you know, I guess it's because I, you know, I quite like the subject that I teach.

Stephen 24:03
Yes, I, that's certainly the sense that I'm getting at least. I, I know that this, this question can, can be tricky and feel free to take it as specifically or as generally as you want, but what advice would you give to someone who is considering, um, you know, following in your footsteps? And You know, some, someone who is curious about your field and like interesting and pursuing it at this level, what, what advice do you think you would have? 

Fred Chong 24:25 
Yeah. I mean, I think that at least the way I have approached my research and my career is really trying to think out of the box, you know, to sort of use a cliche. But I've really focused on disruptive technologies and ideas and what can be new. I think being a professor, being in, being a researcher, being at that edge of what is truly new is where it's really interesting.

Having, you know, looking at the intersection of different fields and ideas is a good way to think, to find what's new. I think a certain level of optimism and fearlessness towards, working on something that's as disruptive is useful.

I think, optimism really has helped a lot in what I do. And I think it's useful because it can be infectious for your students. And then I think that, you know, and for me, I think that, you know, working with my students and working with really sort of nurturing smart people is what leads to success.

Stephen 25:42
Just in the time before we run out of time, what would you say is most fulfilling about what you do? 

Fred Chong 25:49
Yeah, I mean, I do think that, you know, I've been privileged in my career that everything has always gone pretty well for me in the sense that I feel happy that I've always been able to put my students interests first. You know, it's never been like, oh, I have to finish this thing or deliver this thing, and I need to force my students to work on a specific thing.

I usually let them work on what they're interested in, and I really just want to do the best for their careers, right? And so I think that's the most fulfilling thing. I think earlier on, you asked me, you know, how I got into being a professor, and I sort of said, oh, you know, I, decided to try it and I, thought I'd go into industry and I think, you know, looking back, the reason why this is sort of like the right job for me every day of my life is that, you know, I get to work with my students and, you know, I get to, I think have impact on the world through them, right?

Like they go off and become researchers themselves and, you know, that's a lot more impact than me. Any, you know, specific contribution I will make, technically. 

Stephen 27:01
Thank you, Professor Chong, for your time today. And Course Takers, if you enjoyed today's interview, please check out the other ones. Leave us a comment, subscribe, follow, and share this episode with your friends and family. You can find out more about the University of Chicago through uchicago.edu or the university's campus in Hong Kong through uchicago.hk. Stay tuned for more, and thanks for listening.