Science Fair Series: Demystifying Quantum

Quantum technologies are redefining the landscape of science and policy. National leading experts Spyridon Michalakis and Gorjan Alagic explore the foundations of this emerging field and its implications for innovation, security, and global governance. 

The Science Fair Series is a new meeting series highlighting cutting-edge developments in emerging technologies that will impact foreign affairs. This event is made possible by the support of the MacArthur Foundation, Rockefeller Philanthropy Advisors, and the Hewlett Foundation.

FROMAN: Good evening, everybody. Welcome. Welcome to our first session of what we’re calling our Science Fair Series. This is a(n) effort to have CFR be at the forefront of informed conversations about key emerging technologies that have ramifications for foreign policy and national security. We’re very grateful to the MacArthur Foundation, Rockefeller Philanthropy Advisors, and the Hewlett Foundation for making this series possible.

And tonight’s event is all about demystifying quantum technology—using the strange roles of quantum physics to supercharge computing, secure communications, and revolutionize sensing. Already in the conversations in the hallway I’ve heard various remarks which really reflected the value of tonight’s event and this series, which is everyone in Washington is talking about quantum computing and very few people are willing to admit that they don’t really know much about quantum computing. So this is intended to be a safe space for members and others to ask questions.

First of all we’ll hear from two terrific experts led by Kat Duffy, our senior fellow here, and then you’ll have an opportunity to ask questions. And rest assured we are all learners here, and so it’s a great opportunity for that.

It’s timely because CFR just published a task force report—task force was chaired by Gina Raimondo; Jim Taiclet, CEO of Lockheed; Justin Muzinich, the deputy secretary of treasury during Trump’s first administration—on economic security. And I was pleased to read the National Security Strategy released Thursday night if for no other reason because it mentioned that there were three key technologies—AI, biotech, and quantum—which happen to be the three technologies that the task force focused on. And so whether we had foresight or they read the report I don’t know, but clearly quantum is very much at the top of everybody’s list as a key technology that we need to excel at and, in the current context, work to be a leader in. And I hope you’ll look at the task force report and some of the very specific recommendations, including calling on the government to be a purchaser, a demander for the first commercial-scale quantum computer to try and bring that to the market and create market forces to support this going forward.

We’re very fortunate tonight to have two very accomplished scientists in this area, Spiros Michalakis and Gorjan Alagic, to discuss this emerging field. They’re going to be led in conversation by Kat Duffy, as I said, our senior fellow for digital and cyberspace policy. We’ll kick off the conversation, then there will be an opportunity for those of you in the room and online to ask questions.

So welcome to the first Science Fair. We’re going to do this, Kat, quarterly, is that right? Quarterly, each one—each time taking a new emerging technology. But we thought we’d start with a really easy one first. (Laughter.) So, please, welcome to the stage. (Applause.)

DUFFY: Happy Monday, everybody. Thanks so much for joining us here and for the folks online. Mike, thank you for those remarks.

So, yeah, as Mike said, this is our first-ever Science Fair Series meeting. We’re really excited to be doing this.

One of the reasons that we chose quantum as our—as our first session is because it’s been in the news so much lately because it’s been such a clear priority for the administration. Side benefit that we have a wonderful task force report that came out that talks about it as well, so I really encourage everyone to check it out.

I’m Kat Duffy, senior fellow for digital and cyberspace policy here at the Council, and I’ll be presiding over today’s discussion. As a reminder for everyone, this conversation is on the record and is being recorded. So, with that, I want to say we’ll have about twenty-five minutes of a sort of fireside chat, and then from there we’ll go to an open Q&A.

Before we get going, so I can get a tenor at least of the room, who here would self-identify as quantum-fluent? OK. We got a couple. I see you, Tim. All right.

Who here would—who here would self-identify as quantum-proficient?

ALAGIC: See, I was waiting for that one. (Laughter.) (Inaudible)—that.

DUFFY: Who here would self-identify more as quantum Duolingo? OK. That’s perfect. That’s exactly what we’re looking for.

I think one of the reasons that we were really excited about launching this series is having watched the shock that took over the world when generative AI tools became publicly available; and how big a gap there was between so many policy thinkers, and policy leaders, and decision-makers and those who are the cutting edge of that science. And so part of what we’re really hoping to do with this series is give our incredible members and then the general public as well the ability to truly ask some basic questions to understand how to phone a friend, and then also the ability to ask some tougher questions if you want during the Q&A for those of you who identify as proficient or fluent.

Before we start, one last thing that I do for all of these meetings: Can everyone please do this for me? I need to make sure everyone has the capacity to do it. This is the jargon giraffe. (Laughter.) I have used the jargon giraffe for many, many years to great success. If you hear a term, an acronym, a topic, and you have absolutely no idea what it is—we have all experienced this at various times—I guarantee you that there are other people in the room who feel the same way. If you would do us the courtesy of jargon giraffe-ing, we will watch for that and then we will explain whatever was confusing. OK. Fantastic.

And so, with that, I’m so excited to introduce Gorjan Alagic and Spiros Michalakis. And they are incredible experts in their field, but I really wanted them to be able to say in their own words what they do. So, Gorjan, can we start with you?

ALAGIC: Yeah. So I’m a research scientist at the University of Maryland, and I study kind of the intersection of two fields. One of them is quantum computation, specifically quantum algorithms; and the other one is cybersecurity, and cryptography more specifically.

DUFFY: Spiros?

MICHALAKIS: And I’m Spiros Michalakis, and I am a mathematical physicist, a research scientist myself at the California Institute of Technology. And there I study the foundations, I guess, of space and time, quantum gravity, the theory of everything, but also at the intersection of what is known as quantum many-body physics, which includes quantum computers and a bunch of other things like that.

DUFFY: Fantastic.

And so, again, starting with sort of 101, there are, as Mike said, a lot of people throwing quantum in front of whatever these days—quantum mechanics, quantum physics, quantum computing, quantum cryptography. From your perspective as experts, when you hear “quantum,” what are you thinking? And do you have a sense of what you think other people are thinking—(laughs)—when they say that? What’s a misconception, I guess, that you are sometimes seeing as people come to the field?

ALAGIC: Well, maybe one misconception is that quantum technology is a new thing. And there’s a sense in which—and we were talking about this before. So there’s a sense in which quantum technologies are already all around us. I mean, it’s a science that’s, you know, a century, over a century old now. And we’ve been building devices using our knowledge of quantum mechanics, including our cellphones, for—just to give one example—for a long time. So in that sense, you know, quantum technologies are not new.

So when we talk about quantum technologies now, what do we mean? Well, we do mean something new, something that’s different. And what is that thing that’s different? It’s the way in which those devices actually process information or do computations. So, you know, my cellphone, in order to build it you certainly needed a knowledge of quantum mechanics, but the things that the cellphone itself is doing are in some sense kind of classical physics, sort of boring old computations that don’t actually require any quantum mechanics to understand. So that’s what’s really kind of new. So when you say “quantum,” I always want to know, well, do you mean like quantum mechanics or do you really mean kind of quantum information, quantum computation, these sorts of things?

MICHALAKIS: Yeah. I want to add to this. So I was responsible for getting Paul Rudd, the actor, to say in one of his movies, like, do you guys really put “quantum” in front of everything? And I chuckled when I saw it in a movie, but it’s true now everyone is putting “quantum.” It could be quantum AI, quantum computing, quantum advantage, quantum supremacy, quantum everything.

And as Gorjan said, I think what is new now is that we are in this quantum 2.0 revolution that is qualitatively, not just quantitatively, different from what we’ve done before with quantum physics. What is special about now versus even, you know, twenty, thirty years ago is that we are learning how to harness the power of connecting quantum pieces of the world together so they can start talking to each other. And that uses a resource that is almost like energy but maybe even more fundamental we call quantum entanglement. This is the thing you may have read, right? You know, you bring two particles together and then you can send them apart after they’ve interacted with each other for a little bit, and no matter how far away they are from each other somehow they maintain this memory of each other. But it’s something much deeper than that. It’s almost as if, like, they exist beyond space and time itself. They will always be connected. It’s something even more fundamental, a link that, you know, you cannot break unless you do specific things.

And that power, to be able to, like, put together a bunch of these things and have them interact, and then allow them to work in concert like a massive orchestra, to create tunes that we’ve never heard before, this is amazing, right? And this is the foundation of quantum computing. But we couldn’t do these things—I mean, the Nobel Prize for the first lab to ever, you know, even get, like, a few of them—three or four of them to talk to each other, without, like, just disappearing, like kids running away from each other, that was not long ago. That was David Wineland from, like, maybe five, six years ago. And so that is a big deal. And we’ve come so far from the world where we’re thinking, OK, even Einstein thought it was too weird. It was spooky action at a distance, is what he called quantum entanglement, to now using it again and again in what we call the entanglement frontier.

At Caltech specifically, and part of the Institute for Quantum Information and Matter, and we, literally, like, you know, our motto is “exploring the true final frontier.” It’s not even space, right? It’s the entanglement frontier. It’s the place where you have very high complexity quantum information processing, quantum states that then you can squeeze incredible amount of, you know, solutions and information out of, if you know what you’re doing.

DUFFY: So to take it down a level maybe, there’s—some people would say that, you know, they would think about quantum or quantum particles as being unstable. Other people would talk about superposition. What are the difference—what is the difference between those two things? Are they inherently unstable, or are they just not corralled yet?

MICHALAKIS: Those are two—right, like, you know, in—yeah.

ALAGIC: You can—I mean, you can—you can have some—you can have a quantum state that’s in superposition and is unstable. You can have one that’s not in superposition and is stable or unstable. And so in some they are—

DUFFY: And what does superposition mean? Like, does it have a cape? Like, what—

ALAGIC: (Laughs.) Sure. Yeah, so that’s a great question. You know, so when we say “superposition” what do we mean by that? So let’s think about—for example, so I think—you know, one good way to try to understand principles of quantum mechanics, as a layperson, I think is to kind of compare them to kind of probability—like flipping coins and drawing cards, things like that. So, for instance, you know, we all—we all know what it means to flip a coin. If I were to flip one right now and now show it to any of you, you might then decide, well, what—you know, what is the state of that coin? Like, I might know that it’s actually heads, but you guys don’t know what it is. Maybe you just know that it’s a fair coin. So how should you describe that coin?

You might describe it by saying, well, it’s heads with probably a half and it’s tails with probability a half. That’s going to be my description, OK? But notice that this is—this description only—it’s really a description of your knowledge. It’s not a description of the actual state of the thing. The actual state of the thing is heads. I’m looking at it right now. I know that’s what it is. Superposition is a little bit like that. It’s the ability to be in more than one state at once, except, unlike in the case of probability where it’s just sort of modeling limited knowledge, it’s actually inherent. So if I were to put a coin in superposition of in an equal superposition of heads and tails, even if I told you the entire history of the universe and you had complete knowledge you would not be able to say, oh, actually, it’s heads, it’s actually tails, or whatever. So it’s really inherently in that—in that combination. And, you know, there’s no—it’s not a matter of perspective. It’s really, actually simultaneously in those two states.

MICHALAKIS: Actually, I do want to add to that. For me, the simplest way of thinking of superposition is a shift in perspective. Like, if I were to ask you, you know, to move one step forward, that would be different for every single one of you. For some of you, it would be like, you know, this way, this way, this way. And if you think of quantum physics as a theory of knowledge, as it actually is. It’s not just a theory of physics—it’s something way deeper in epistemology—then it asks you to look at the world from different perspectives. And if I were to tell you, OK, I want you to move forward one step and to the left at the same time. how would you do it?

In space it’s not as hard. You just rotate yourself a little bit and then just take one step forward, OK? But in the realm that goes beyond space and time, if you were to try to understand the world from, like, this is forward and this is to my left, but somebody says, like, no, this is forward for me, and this is to my left, again, it’s two different ways of understanding the world is just shifted, rotated perspective. You’re now in a place where you don’t know what forward is relative to the previous, you know, point. You’re, like, in between forward and to the left. So this is what is going on.

Internal to this quantum devices, internal to, like, the way we understand the world, we only understand the world from specific points of view. And we have this illusion that we all share the same point of view, the human point of view, and that there is no other point of view. Quantum computers and quantum physics comes inside to tell us, like, you’re very, very wrong. There are much more complex, high dimensional point of view you don’t have access to, right? And if you broke away from the human point of view, you could get to answers or understand things immediately, much faster than, you know, being—it’s like going in Manhattan and trying to actually fly over Manhattan to where you need to go, or just go the Manhattan distance, as we call it, around every single step.

DUFFY: Basically. I think one thing we were—(coughs)—excuse me—saying earlier is that if we’re all—if we’re all fish in a quantum ocean, we don’t understand that we’re swimming in quantum, right? And so part of what the really interesting exploration is right now is getting to become fish that would essentially understand what we are actually moving through, and how we—

MICHALAKIS: How do we open—that’s right. How do we open the curtains? Like, again, I’ve worked with Marvel introducing the quantum realm. And when they sent one of their actors to, like, you know, ask me a bunch of questions at Caltech, one of the questions was, you know, what is the quantum realm? How small is it? And how do we get there? (Laughter.) And I was, like, hmm, we are in the quantum realm. This is the quantum realm, right? All the things we think are not possible are actually possible, we just haven’t found a way to open the curtains to see the invisible, make it visible.

And specifically, what this means—and I talked about the entanglement frontier—is the ability to create macroscopic, stable, reliable, on demand quantum phenomena. One of which we call quantum computers, that are large scale and can exist long enough internally into the quantum realm, right, to do computations that we wouldn’t be able to do otherwise. Quantum sensing is another one, right, to use properties—fundamental properties, pre-existing but invisible to us for tens of thousands of years.

By the way, this is something that is incredible to me. The quantum realm and quantum physics has been the law of the land not for the past 100 years. It has always been the way that things have run. We just never understood that. We never saw any of that, right? So we’re trying to undo our primitive intuition and say that we can ask many more interesting questions of the thing that exists, as Plato put it, like, you know, the thing in itself, reality, you know, outside of what we see. And that’s what we’re trying to uncover. And we can see a lot of applications come through that.

DUFFY: So Spiros is either famous or notorious, depending on your perspective, for having worked with Marvel to create the phrase “quantum world.” So Dr. Strange, and The Avengers, and Antman all exist in a quantum world. And that’s Spiros’ fault because I think—

MICHALAKIS: Guilty.

DUFFY: Was it microverse?

MICHALAKIS: It was the microverse. They couldn’t say that.

DUFFY: Because copyright. (Laughter.) So—

MICHALAKIS: Tiny people.

DUFFY: So we got “quantum world” because copyright, because microverse wasn’t available. But so I want to—I want to move to the world of both the theoretical and then the, like, this for sure is going to happen, right? Or this for sure is a distinct possibility. So if we are in a, let’s say, a quantum world, like, we’ve had some really advanced quantum breakthroughs of some sort, Spiros, what do you think is the coolest thing—theoretical possibility in physics—like, it is a theoretical possibility. It’s not science fiction. What is the most interesting part of a quantum world that we could be in?

MICHALAKIS: You asked two things that are almost contradictory to each other, because I believe that everything is possible. Like, that’s the most important realization coming from quantum physics, not quantum mechanics. It’s a special flavor of quantum physics as an epistemology that tells you, like, every law of physics, everything you’ve ever thought was, like, settled, it’s just really persistent illusion. But if you know how to zoom in and zoom out properly and shift your point of view, then teleportation, time travel, anything we’ve ever thought. Like, you know, even crazier things than that, traveling left or right or up or down in time, not even forward and backward, whatever that means, jumping to different realities, I think it’s all possible.

But if we’re going to be more specific about what is, you know, not potentially science fiction kind of advances you can see through science, quantum simulation is a very good. Richard Feynman at Caltech, why we have now the pursuit of quantum computers beyond just Shor’s algorithm, which has to do with cryptography, is the idea that nature is quantum. And we need to understand the language that nature speaks to itself, right, as it is doing, everything it is doing. And by the way, this is amazing. No supercomputing simulation can even do the simplest thing that nature does within milliseconds. So we’re not in a simulation. That’s actually pretty simple because we can tell apart whether we’re at the root of the simulation argument or not because we have a quantum universe, and we need quantum computers to simulate even the simplest classical processes when you zoom in enough.

And so to me, like, you know, this says that you can use quantum computers to create mini multiverses, like—or, mini-universes. We become, like, the lords, the overseers of it. And then we tell it what to do, instead of telling us what to do, right? And that is very cool, because you can simulate chemistry, drug discovery, material science, like, to understand the right energy landscape for connections between atoms, molecules, and so on and so forth, that then give you amazing new materials. For example, superconductors that superconduct at room temperature. You know, Back to The Future. I mean, that would be very cool, right, floating around, and all that stuff, so.

DUFFY: Which raises a huge range of possibilities in terms of use cases, where it can exist, how it can exist. But so, Gorjan, I want to go over to you, because cryptography. So in foreign affairs and in national security, one of the biggest concerns that I hear, at least, is what quantum could do to our existing encryption. And for those who don’t think about encryption a lot, it’s basically our entire world right now. Your text messages, your emails, your login to your bank, our government records, you know, our intelligence community—all of these things involve public key encryption. And so can you—can you speak to the very real sort of near-term potential that quantum has, and how it’s—how it’s being addressed in terms of protecting encryption?

ALAGIC: Sure. So, you know, there’s a lot of uncertainty as to what these sort of large-scale quantum computers will be able to do. And when I say large scale here, I’m talking about devices that we don’t yet have. So far away these devices are, that’s very hard to say, but, you know, they’re probably at least five to ten years away. But one thing that’s for sure is that if we had those devices right now, if anyone had them right now, they would be able to easily break most of the cryptography that we currently use. And specifically, they would be able to attack the kind of cryptography that’s used to establish secure channels. And this is a fundamental building block of the internet, you know, the ability to be able to connect at a distance to different parties without having to, you know, agree in advance on some shared secret. This is fundamental to, essentially, the entire information infrastructure system that we have around the entire world.

And so if quantum—if these large-scale quantum computers were available right now, that would be a huge problem for that—for that entire system. And I want to emphasize that, you know, this is not the kind of—it would not present the same kind of problem as these kinds of hacks that we see on a daily basis, or, you know, leaked hashed password databases, or, you know, whatever—or sort of social engineering attacks, or these kinds of things that are sort of limited in scope, relatively easy to patch. We can kind of address them and keep going. The reason that we can address them and keep going is because the fundamental algorithms were never really under threat. And these are exactly the algorithms that quantum computers would be able to attack. So there’s in some way—you know, if it happens to you, there’s kind of very little you can do to stop it.

Now, people have, of course, known for some time that this is coming. In fact, you know, this—Shor’s algorithm, this quantum algorithm that we know can break cryptography, was one of the earliest, and maybe to this day the most important, discovery in quantum computation. So that happened in the early ’90s. And people have thought a lot about how to protect our cybersecurity systems from quantum computers. So we now have cryptographic schemes these—we have algorithms, alternative algorithms, they’re called post-quantum cryptography, that believe would not be—

DUFFY: You might have led a process for standardizing—

ALAGIC: Oh, I didn’t lead it, but I did participate in it, yeah. And these algorithms are believed to be secure against quantum computers. And, you know, NIST has standardized some of these algorithms, and they’re in the process of being deployed. And, you know, this is going to be a very difficult and challenging process because, you know, I mean, cryptography is really everywhere. And kind of ripping these things out and replacing them with something else is not a simple matter. So this is, you know, a very long process. I think the current sort of federal government plan is for these things to be—for the existing public key algorithms to be deprecated in 2030 and eliminated by 2035. That’s sort of the hope. And, you know, it’s a long process. And it’s very challenging. And it’s definitely a very real threat that we have to worry about.

MICHALAKIS: I want to add something to this, if I may, because it’s very important. And we have private conversations, but maybe they should be on the record, that we need a lot more cryptanalysis, even for the new post-quantum cryptography standards that Gorjan helped standardize. And one of his, you know, nightmare scenarios is that we literally move—because that’s what the NSA says we should do—and remove Shor, right, you know, we just move to these new lattice-based cryptographic protocols. Might as well just do that. And then some kid appears and is, like, huh? Because we haven’t had forty, fifty years, right, you know, and a lot of money and high stakes to make sure that these things work. Then we have nothing to fall back into. And Gorjan loves to even say, like, and it turns out we couldn’t even make big enough quantum computers to break Shor, so now we are in real trouble because we substituted this, like, thing that worked with something that may not work anymore.

And that raises the bigger question of, like, why haven’t we created cybersecurity, like, level agile cryptography, right, where you can plug and play as needed? Because if that turns out not to be right and we need to substitute something else, we don’t have to go and rip again at the very DNA level, right, you know, of all of these communications, and try to do it over five—there is no five or ten years, right? If you’re—this is a zero-day. Economies around the world just—it doesn’t matter if you’re Caterpillar and you’re, like, oh, I have no, you know, whatever, presence. Of course you have presence. You’re talking to each other, you know, about physical inventory. You have a website. You have—you’re doing—

ALAGIC: Your bank account.

MICHALAKIS: Your bank account, everything. Imagine going back to sixty years ago.

DUFFY: Well, and I think this is, you know, something Gorjan and I talked about. Everyone has different, sort of, risk thresholds or probability thresholds where they’re going to pay attention, right? So if you’re an investor, and you may be thinking, well, there’s a 50 percent possibility of return here, it’s worth it because the return would be huge. But if you’re in cryptography, and there is a 0.01 percent chance that it could be broken, like your hair should be on fire. Like, that’s a—that is an immediate and urgent situation.

And so I think one of the things that’s really interesting that I really take away from this conversation is both that for all of the—we hear a lot of increasingly kind of concrete statements around exactly what quantum will be able to do, and why it is so important. And what I hear from both of you, who are really working in the field, is that we are really—we are really working with an unbelievably theoretical world, and trying to prepare in one paradigm for a paradigm we might not understand. And that’s just a really challenging thing to do.

MICHALAKIS: I mean, we are experts, and we do not have deep enough intuition to feel confident that, you know, any of these things is going to pan out the way that we want it to pan out. Like, we need a lot deeper research, everything from curiosity-driven all the way, you know, full stack, to how do we deploy this securely. Very few people are doing this, including, like, quantum, like, AI Lab at Google and IBM, and all the ones you may have heard IonQ. All the ones doing really good work to build, like, hardware, it is so freaking hard to do this work. It is, like, seriously hard. And we need a lot more brainpower, collaboration, partnerships to be able to put this together.

DUFFY: And this is my last question, and then we’re going to go to Q&A. You know, one of the things that’s really struck me with some of the administration’s focus on key technologies, or critical technologies, right, on AI or on quantum, is that I don’t tend to think of those as technologies so much as laterals that could inform all sorts of different technologies, areas of science, areas of medicine. And so when you all think about what investments should look like in quantum, take as an assumption—just take as a critical assumption that there is a national security and, say, global imperative for the United States to stay ahead of the curve in terms of quantum. What are the investments that, as scientists, you feel are most mission critical in this moment to make sure that America is positioned for whatever it is? Gorjan, let’s start with you.

ALAGIC: I think algorithms research. I think, you know, I’m going to strongly agree with Spiros here, the, you know, hardware is very difficult. It’s a big challenge. But without interesting software to run on that hardware, you know, what is it for? And the interesting software for quantum computers is going to be algorithms. It’s going to be, you know, computational methods for solving some problems that we either don’t know or maybe can’t solve using our existing machines. And, you know, that’s a very—it’s a very difficult area. Like Spiros was saying, it’s very challenging. Parts of it are theoretical research, maybe significant parts, but there’s also a lot of—a lot to do with, you know, the stack that Spiros was talking about. Thinking about, you know, error corrected algorithms, thinking about running these on different platforms, and so on. So I would say quantum algorithms is a major area.

I would also add, you know, as Spiros was saying—was talking about cryptanalysis. Cryptanalysis is a subfield of algorithms. It’s looking for algorithms that can break cryptography. Which might seem, kind of, like, why would you do that? Well, you would—you want to do that because you want to get there first. You want to figure out whether your thing is broken or not, before you go out and deploy it and somebody else figures it out. So it’s a critically important thing to do, which is sort of—it doesn’t have this, sort of, like, sexy payoff, but it’s super important that it’s done. And I think that’s a sub-area of quantum algorithms that needs more support.

MICHALAKIS: I think of it as preventative.

DUFFY: (Inaudible)—it’s probably not immediately revenue-generating. Any time—

MICHALAKIS: But it’s—like health care, though, it’s preventative, right? You know, preventative health care may be all you ever need until, like, you know, at some point it’s going to save you a lot more money than having to treat, like, you know, symptoms afterwards.

DUFFY: How about you, Spiros?

MICHALAKIS: What Gorjan said. And I want to do you all a solid and ask you series of questions, it’s going to create maybe a lot of clarity at the end of this, right? So here’s, like, you know, the first question. And the context is that you’re all familiar with GPUs right now. They all started with video game cards, right? Graphics, all that stuff. And now they transformed the world because you can train AI on them. I mean, it really is a big deal. So from that point of view of, like, going from gigahertz, like, for your processors and your CPUs, to now terahertz, for this NVIDIA H100s and A100s, and they’re $40,000 apiece, and you need 25,000 of them to train, like, you know, GPT4 and whatnot. They’re just faster, right? They allow for many more computations per second, and all that stuff.

So here is the question, all right? How many of you think that a quantum computer will be a thousand times faster, a thousand times more operations per second, than your laptop? Raise your hand. OK. What about a million times faster? What about, like, a billion times faster? I can do a billion times more operations per second? What about a billionth times faster—one over a billion? (Laughter.) That is the correct answer. Sorry to burst the bubble, but a quantum computer will be able to do about a couple of million operations per second instead of trillions of operations per second, which is what a good laptop can do these days, or your phone can do millions of operations, right? So what is going on here? Why are we up here? Why are we talking about quantum?

This is a way to help you clarify what is special about quantum computing. What is special is not the quantum computer. It is the quantum algorithms that can only run on a quantum computer, because what I didn’t mention is that even if you’re, like, able to only run a million quantum operations per second, the quantum algorithms that can leverage quantum operations, not just classical Lloyd logic gates, and all that stuff. They can allow you to—it’s almost like you’re in a little tricycle against NASCAR drivers, right? And they have to go—the Daytona 500 have to go 500 times around. I think that’s right, right? And you only have to go around once on a smaller track, right, with your little tricycle and training wheels. Who is going to win? You’re going to win, right?

And, by the way, the training wheels are like kind of error correction right now to make sure you don’t fall over as you’re trying to even do that little circle. So that’s the correct analysis you have in mind, right? It is something that is actually much slower, right? The number of operations it can do, it’s way fewer. But each of these operations depending, like, you know, if you have the right quantum algorithms, like, the quantum magician, a quantum software developer on hand, they can extract immense, immense value from that, right? So instead of you having to take—do you know how many operations you need to train GPT4? I think was, like, ten trillion trillion operations. That is one with twenty-five zeros, OK? And required, like, these terahertz machines that can do, like, ten to twenty trillion operations per second, and you needed a 25,000 of them running for four months straight.

That is very expensive from an energy perspective, from a money perspective, right? Microsoft spent $1 billion, because 25,000 times 40,000 per each of these GPUs is $1 billion to create Azure, like, the cloud where OpenAI trained these things. And you can imagine that if you could do the same training, because synthetic data, quantum data is much more powerful, potentially only using like a million quantum operations, that would take you one second, right? (Laughs.)

DUFFY: I think of it sometimes as—and then we’re going to go to Q&A—I think of it sometimes as a—I’m a cyclist. Like, I’m a bike commuter. There is no way that on my bike I could beat a Ferrari in an F-1 race. But I would 100 percent take on a Ferrari getting from Capitol Hill to CFR. (Laughter.) A hundred percent. I think that I would smoke them, at least on my e-bike.

OK, we’re going to open up to questions. Tim. (Laughs.)

Q: Yeah. Thanks very much to the panel. Kat, thanks for organizing this, and to the Council. I’m Tim Persons with PwC.

Just in—this is a contextual thing. You are all really talking about one of the big risks is the harvest now, decrypt later. So one could argue, depending on the when—the big when question—we’re well into the red zone on this. What we know is that it takes about a decade, it’s a surprisingly long amount of time, to do a full crypto system change out. So, again, if you’re—somehow if we’re in that red zone already, I just wanted to ask the panelists your thoughts on are we in the red zone? When do you think the big, scalable, scary thing is going to come, and put all commercial encryption systems at risk? Thank you.

MICHALAKIS: That’s a great question.

ALAGIC: Yeah, no, I think that’s an excellent question. You know, there’s a sense—so maybe, just to back up a little bit, you know, there’s a sense in which quantum computers, sort of the view from Mars, kind of affects cryptography generally. Kind of almost no matter where you’re using it or what you’re using it for, quantum computers will have something to say about it. Now, you brought up a great point, which is that there’s a big difference between encryption and authentication, right? Authentication is—like, you can think about it as logging into your bank account. And encryption you can think about as the algorithm that makes sure that nobody can read your communications with the bank once you’re logged in.

And so, you know, if—so quantum computers, if they were to exist today, could break either one of those algorithms. They could break both of them, actually. Now, of course, you don’t have to worry about the first problem. You don’t have to worry about somebody trying to break into your bank account today with a quantum computer because there is no quantum computer of that kind. But what you might very well worry about, as you said, is that somebody might be storing those communications, those encrypted communications, between you and your bank, and then ten years from now, fifteen years from now, using a quantum computer to decrypt them. You can’t protect yourself against that possibility right now, unless you switch to these new algorithms.

Now, when will that happen and how much you should worry about that depends on two kind of big question marks. One of them is, you know, what is your—what is your application? You know, like, I don’t maybe personally care all that much about keeping my bank transactions secret for twenty years, but there’s plenty of other people who care a whole lot about keeping their communication secret for twenty years. So that’s sort of the big question mark. It’s very application dependent. The other big question mark is when is this cryptographically relevant quantum computer that can decrypt things going to come? And that’s a very difficult question to answer. You know, I think it’s, it’s hard to imagine that it will happen in, you know, less than ten years, for me personally. But, you know, these estimates vary a lot. It could be a lot longer.

One thing I can say is that, you know, we can’t really eliminate the possibility that, you know, some sort of quantum analog of a transistor won’t be invented tomorrow. You know, there might be—very well be some major breakthrough that sort of really changes the game, that the completely transforms our approach to even building the thing, you know, maybe skips the whole physical qubits things, go straight to logical qubits or something like that, and kind of overnight kind of brings these things to within kind of a year or two away. That’s possible. And it’s very difficult to assign kind of confidence levels to something like that. Yeah.

DUFFY: Mike.

FROMAN: Excuse me. This being the Council on Foreign Relations, there’s been one word notably not mentioned yet, which is China. Where do we stand vis-à-vis our research with China? It’s thought that China is spending about five times as much as we are on quantum computing research, perhaps, or quantum research. Is that significant? In a different world, would you be cooperating or collaborating with a lot of Chinese scientists, and are—how worried—if in the AI world people talk about us being ahead of China towards AGI by some number of months or a few years, how would you answer that question on?

MICHALAKIS: That’s a light question for quantum physicists. But what I will say is this, I will tell you, from my experience, that not in a different world. In this world. We have at Caltech brilliant, you know, Chinese students. And we’ve had them there and around the United States for decades. And, in my view, this is one of the best things we can do for our relationship with any other country, because the best ambassadors I’ve ever seen are scientists. Because we’re trying to collaborate and understand how to beat a much bigger opponent, which is nature itself. Like, these are, like, human-level threats. And whether we get first or they get first there and all that stuff, this is—you know, if you have a quantum algorithm that can help you fight climate change, or find higher energy sources, better density sources, all this stuff, that helps everyone.

Having said that, of course, yes, it’s naïve to say, well, there is no threat then, and they’re not going to try to, like, do something first, and use it adversarially. But, again, I mean, it’s probably above our pay grade, export controls. And, you know, one of our mentors, John Preskill, is literally responsible for dealing with some of these things. But I think he’s also of the opinion that science transcends basically everything else. There’s just something like that journey, that adventure requires all hands on board. I don’t know, like, how this translates, right? But, yeah. They’re taking a lot of great brains, not just China, but Europe and others right now, because we are closing off the gates. And I think—I mean, I grew up in Greece. Gorjan, I assume—

ALAGIC: Bosnia.

MICHALAKIS: Bosnia. Like, here we are. And some of the best minds we’ve ever met also were immigrants. They came in and they were looking for some, you know, exciting adventure, something that they couldn’t get where they grew up. Even if they were to go back, the more the merrier, I say, right? Like, we’re very proud at Caltech for being a catalyst for some of the world leaders who are either now running the biggest companies around quantum or the biggest labs around the world. And even some of the ones that were trained in the U.S. are now running things in China. There’s still collaborations, right? China was collaborating with Austria on satellite, like, you know, and quantum networking.

I don’t know. I’m worried that at some point we’re going turn off that spigot and the communications are going to stop. I hope not. It’s, like, you know, the three body problem says, you know, try to keep things open across different partners. But I try to look at science, even science that has, you know, potentially, like, important applications for national security, as maybe a bright spot for connecting, instead of, like, separating.

DUFFY: And I think, you know, we had a conversation earlier that won’t be online, but I think something that you both mentioned as well, which—and it’s an important consideration. You know, in the task force report we point out that the United States right now lacks—is fourth, basically, for quantum talent, behind China, India, and the EU. But that which inspires scientists to want to come here and want to build here has to do a lot with the cultural openness to science for the sake of science, not for the sake of a national strategic advantage, not for the sake of dominance, not for the sake of a giant revenue return. That, for many of the most brilliant minds in the field, that isn’t going to be the thing. That’s not going to be the magnetic force. The magnetic force is the space to really think big, to dream, and to do so kind of free of other constraints.

MICHALAKIS: Every single breakthrough, I think, that has—like, is transforming even quantum science into something that could be useful for the world, all the big ones, and especially ones that then transform directly into resource allocation and cheaper ways of doing things in industry, basically come from just curiosity-driven research, with enough guidance to say, hmm, wouldn’t it be cool if we developed a quantum algorithm or an error-correcting algorithm, a fault-tolerant algorithm that required only one-hundredth of the qubits, or the fidelity, or whatever between the quantum operations and the quantum entanglement to do the same thing?

And all of a sudden that makes a huge difference between, like, whether it’s going to take you $1 billion to build this over the next few years, versus going to take you $10 million, and start prototyping on top of that. Think—this is very important point—that, and we mentioned it a little bit in passing before, quantum algorithms and quantum software is what makes quantum computers special, right? Just to get all the glory, because without quantum computers you cannot run an algorithm that leverages quantum operations, quantum logical operations. That’s the key. If nothing else you take away from this evening, this is it, right? It’s not that these are just faster GPUs or more parallel GPUs. That’s not what’s happening. And it’s not like—

DUFFY: Gorjan, I see—because I see Gorjan’s got a thought as well.

ALAGIC: Well, I just wanted to jump in and say that, you know, like, one way—if you want to think about these things sort of adversarially and as a competition, you know, one way to beat a bigger pot of money is to work on the right thing, right? I mean, you know, like, and that’s where I think—and actually, up until very recently, I think that’s kind of been happening in the quantum field. That because of the history of science, and the institutions, and the freedom, and the amazing talent, kind of, the U.S. scientific community had a better sense of what are the things that we really should be pursuing in the realm of quantum technologies, and what are the things that are, OK, interesting, we should keep them in—you know, we should work them out on our whiteboards and, you know, keep them in that room until some later point? And I would say, in my view, China kind of didn’t really get that quite right. Now, of course, they’re catching up. They’re putting in more money. But, you know, this is one of these places—these ways in which the U.S., I think, still has an advantage, provided we kind of continue investing in that advantage.

MICHALAKIS: That’s right. Anyone who develops the environment for talent like us, to want to say, like, that’s where I want to be. Not because it’s even, like, even more money or whatnot, it’s just this feels like that would be amazing environment to think and, like, create for the future and for the human race, they’re going to get, like, all the talent. They’re going to get all the individuals who are going to come up with the quantum algorithms that are actually going to be useful when you actually run them on these super-corrected qubits.

DUFFY: It’s interesting to think about getting to this hard science through what some would call a softer or more creative opening.

We have a question online.

OPERATOR: We will take our next question from Steven Hellman. Mr. Hellman, please accept the unmute now button.

Q: Hey guys. Really fascinating conversation. Thank you.

Quick question, you kind of touched on this, but what is the relationship between AI and quantum computing? Given that the big innovation in AI came from parallel processing, and my understanding of quantum computing is that you’re effectively parallel processing at almost an infinite scale. Is there a synergy there?

MICHALAKIS: This is a great question. And I will take it. The conversation—

DUFFY: And I want you to do it in, like, one minute or less. We have a lot of questions.

MICHALAKIS: Yeah. I’ll be quick. So the truth is that quantum computers don’t process information like in massively or infinite parallel scale. This is a part of a quantum algorithm that is allowed within, you know, the computational framework. But what we need is to be able to bring back the answer from all these different computational universes back to ours, right? And when you think of it from that point of view, understanding what makes quantum software so much more powerful than classical software is something we are still missing. We don’t have as much, you know, expertise and wisdom in that experience. When it comes to AI, it actually turns out, when you zoom in, that the training part for machine learning models, for, like, you know, in general machine learning, is something that we can already do very fast, through what is known as stochastic gradient descent. This is what we call a linear process.

What is much more expensive, and potentially where quantum computers can really revolutionize things, is the amount of data you need to train to pass through this, like, you know, feedback loop. And with quantum computers, you may be able to generate very deeply, like, textured information correlations, right? Connections between things that otherwise you need a million pieces of data. And you can have it in one of them, and then that can reduce the amount of training time by a million times, or a hundred times, whatever it is. That changes the game.

DUFFY: And Gorjan?

ALAGIC: So I would—yes, I completely agree that this—I’m not sure why it’s still around, but this unfortunate description of quantum computing as sort of trying many solutions in parallel is not correct. And, you know, it’s much more interesting than that. If we had more time, we could get into the details of how it actually works. And that would be fascinating. What I would say, as far as the connection between AI and quantum, I would say that it’s really not clear what the connection will be and what the interplay will be like, except insofar as that it seems, you know, LLMs can help us in a wide variety of settings, technologies, and science. So to whatever extent they can do that kind of generally, they will certainly have an impact on quantum technologies as well. Whether it will go beyond that, and in what way it will go beyond that, I would say at this point it’s somewhat unclear.

DUFFY: Well, and we have LLMs, because of the transition—like, architecture breakthroughs. And, you know, and that’s sort of the current AI that we’ve got. That’s not the AI we’ll always have, necessarily. So I think one of the things that’s really interesting, with your discussion on we don’t know when the breakthroughs will come, is that there may be breakthroughs that come powered by AI, and they’re also—if there’s a breakthrough in quantum, that may power new breakthroughs in AI. And so it can be a—

MICHALAKIS: One very quick thing I’m going to add in the way that AI is actually helping already quantum, is to help us decode—once we have a sense of what the errors are and where the errors are, it’s actually very difficult to try to decode from that how to apply error correction, right, during the quantum computation. And we’re developing right now in neural decoders that are basically understanding the kind of noise that is in the device and helping us run quantum algorithms to correct the errors much faster. And this is something without which, like, we wouldn’t be able to get any of the results you get from Google every now and then, or Amazon, or IBM, saying, wow, we were able to do this and it was fault tolerant.

DUFFY: Next question. Oh my gosh, we have so many.

Q: Thank you so much for really educational presentation. My name is Jaime Yassif. If I’m at NTI.

I have a question about the beneficial applications and the sort of broader implications of the fundamental breakthrough that we’re anticipating when these computers eventually come online. So certainly it has implications for encryption, but it’s going to just change the game generally about what’s possible computationally. So I’m just curious to know if you’re imagining that future what are the ways in which it’s going to change the game more broadly in terms of fundamental capabilities? What is that going to mean for national security and economic competitiveness? And, for example, can we apply that for an arms race on encryption? Can we apply quantum computing to help strengthen encryption algorithms as a counterweight to what we’re already concerned about? Thanks.

ALAGIC: So what I would say is, you know, as we just said a second ago, you know, the idea that quantum computers solve problems by trying many solutions at once is misleading, I would say. And this—and then Spiros earlier talked about the fact that, in terms of clock cycles, quantum computers are actually going to be much slower than the computers we have now. So it’s all about the operations. And what that should kind of be pointing you towards is the fact that, as far as we can tell, the areas where quantum computers can help us are not universal. There are certain problems for which quantum computers can help us and can solve—can produce solutions to computational problems faster. But at least as far as we currently know, for the majority of computational problems, there doesn’t appear to be an advantage to using a quantum computer over a classical computer. So one really has to think about—carefully about, what exactly is the structure of this problem? Why do I think that, you know, quantum mechanical operations should somehow help me with solving this problem better than with classical ones? So one really has to get into the nitty-gritty of how these algorithms work. And that will inform, you know, let’s say, where you might potentially expect certain applications.

As far as the—as far as kind of applying quantum computers kind of in the positive sense for cyber security and cryptography, there are ways in which quantum technologies can be used as kind of in a positive sense for cryptography, but they don’t actually require quantum computers. So there’s—for example, there’s this quantum key distribution technology that’s been around since the ’90s, where you can use certain quantum devices—which are much simpler than a quantum computer, we’ve known how to build them since the late ’80s, early ’90s, to do certain cryptographic tasks that we can’t do with our existing—with our existing hardware. So there’s some promise there. There’s also quite a bit—quite a lot of limits to what they can be used for. But, you know, there’s some—there could be something interesting there.

DUFFY: I want to—I’m cognizant of time, because we are always so careful about punctuality at CFR. I know that you all can probably stay for about ten minutes right at the end if people want to ask questions. And so I wanted to wrap with one final question. You know, as Mike said, it’s the Council on Foreign Relations. And our membership, you know, is comprised of so many different people, you know, in business, in media, in foreign affairs, in policy. From both of your perspectives, if there is one thing that a CFR member could take away from this conversation to sort of keep their—to keep their eye on or to just have in their toolkit when they’re reading about quantum in the news, or they’re seeing sort of grand proclamations about it, what is the one thing, if you could tell them anything, that you want them to have in their head?

MICHALAKIS: Maybe I can start. Think of a quantum computer as a boat, right? You’re not going to use the boat to go to your neighbor’s, like, you know, gala or across town. You’re going to use it to try to explore new places you couldn’t have gotten with the fastest car we have. But we don’t know—we need explorers to figure out where there is interesting things to explore. We literally need to—like, to have individuals who are willing to go the distance, to get on that boat, and try to imagine where you can take them—completely new lands where you can have, like, miracles happening, right? So it’s not that this is, like, an infinitely powerful vehicle that can take you anywhere and everywhere.

And, as Gorjan said, it’s not just slower, right? Even the applications, even the places where you could potentially find something new, we are discovering them as we go. And in fact, even when we have small devices, which is why it’s so important to be working, right, you know, on this right now, it will help us understand if some heuristic algorithms, right, some ones for which we don’t have proofs that are going to be massive advantage. Even if you have, like, a 20 percent lower cost to do something that was just impossible before, that could be transformative. Let alone exponential, like, breakthroughs.

ALAGIC: So I would go back to something that you said earlier. I would say, you know, there’s a lot of debate about when quantum computers, when these large-scale quantum computers are coming, and kind of with what probability do we have confidence in those estimates? And what I would say is, you know, it’s one thing to think about those estimates as an investor, as you said, where you’re interested in much more concrete numbers and you’re much higher probabilities, otherwise you might not invest, for example. And on the other hand, being somebody who’s interested in policy and wanting to know, let’s say, should the federal government transition to post-quantum cryptography, or should my industry do that, and how quickly?

So in order to understand this latter question, well, actually, there is no question. I think everybody agrees that the probability is high enough and that the likely timeline is short enough that this has to happen, OK? So there’s lots of uncertainty as far as, like, what the applications of quantum computers will be, how much economic output they’ll generate, all these other things. What there isn’t uncertainty about is that we need to do something about the problem of cryptography.

DUFFY: Fantastic. Well, with that we have to wrap. Thank you all so much for your time. Thank you to the people who joined us, our members and others who joined us online. I want to say, if—you know, we talked about demystifying quantum. And to some degree, you might walk out being, like, I’m more mystified. (Laughter.) And others may feel demystified. I think, for our perspective and for Science Fair, if you walk out of this room or out of this virtual meeting feeling a healthier sense of engagement and curiosity and capability in entering the world of the unknown, then that’s a win. And if you potentially feel a healthier degree of skepticism in entering a world of claimed certainty, that is also probably a win. And so with that, let’s all go be Magellan. Thank you so much. (Applause.)

(END)

This is an uncorrected transcript.