Into the Impossible With Brian Keating - John Preskill – Part 1 – Quantum Computing, Artificial Intelligence, and Encountering Richard Feynman (#111)

Episode Date: January 18, 2021

Part 1 of 2 Learn about the exciting promise of quantum computing and how it may solve problems in fundamental physics. Join my mailing list to get slides from this conversation: briankeating.com. We went deep…discussing Artificial Intelligence, the simulation hypothesis, lessons from Richard Feynman and more! You don’t want to miss his answers to my patented Thrilling Three final questions! John Preskill is the Richard P. Feynman Professor of Theoretical Physics at the California Institute of Technology, where he is also the Director of the Institute for Quantum Information and Matter. He is one of the most prolific and influential scientists of our time. Preskill is a leading scientist in the field of quantum information science and quantum computation, and he is known for coining the term “quantum supremacy.” Preskill studied magnetic monopoles in Grand Unified Theories. This work pointed out serious flaws in the then-current cosmological models, a problem which was later addressed by Alan Guth and others by proposing the idea of cosmic inflation. He’s the Director of the Institute for Quantum Information at Caltech. He is known for coining the term “Quantum Supremacy” in a 2012 paper. Preskill has achieved some notoriety in the popular press as a party to a number of bets involving fellow theoretical physicists Stephen Hawking and Kip Thorne. Preskill was elected a member of the National Academy of Sciences in 2014. Watch my most popular videos: Jim Simons, the World’s Smartest Billionaire  Bill Perkins: DIE WITH ZERO:  Patrick Bet-David YOUR NEXT FIVE MOVES  Sheldon Glashow Sir Roger Penrose, Nobel Prize winner  Frank Wilczek Jill Tarter  Eric Weinstein Sir Roger Penrose  Juan Maldacena’s First Podcast Interview Sara Seager Venus Life Noam Chomsky Sabine Hossenfelder Sarah Scoles Stephen Wolfram ‍♂️ Find me on Twitter at https://twitter.com/DrBrianKeating Find me on Instagram at https://instagram.com/DrBrianKeating Buy my book LOSING THE NOBEL PRIZE: http://amzn.to/2sa5UpA Subscribe for more great content https://www.youtube.com/DrBrianKeating?sub_confirmation=1 ✍️Detailed Blog posts Learn more about your ad choices. Visit megaphone.fm/adchoices

Hi, everybody. Welcome to this episode of the Into the Impossible podcast. I'm your fearful host, Dr. Brian Keating of the University of California, San Diego. And I am enjoying these pandemic podcasts tremendously and never more so. Then when I get to interview one of my friends, one of my heroes, one of my inspirations and mentors, such as today's podcast, you'll hear really an exclusive interview with none other than John Preskill, who is the Richard Feynman professor of physics. at Caltech. Today you're going to learn about Richard Feynman. You're going to learn how he inspired a nine-year-old John Preskill, who later took the name of the very person who inspired him, namely Richard Feynman. You're going to learn about Feynman's blunders, if there were any. You're going to learn about quantum computing, the simulation hypothesis, artificial intelligence, and even impact on things like cryptography, blockchain, etc. This is really, to my knowledge, the first podcast of this type, not purely about scientific contributions made by John and his group. John's been an inspiration to me since I met him in the year 2000 when I was up at that little

technical college up in Pasadena known as Caltech. And ever since, he's been so generous and gracious with his time and his energy. He's working on a lot of things. I want you to stay in touch with me so that you can get these resources like get notified when his book on quantum computing comes out. This is a book you will not want to miss. He's one of the founders of this field. He'll talk about how Feynman influenced him as well as answering the thrilling three questions

that we always talk about on the Into the Impossible podcast relating to his ethical will, his monolithic wisdom that he would leave on a monolith, and also his advice to his younger self. You don't want to miss it. So please subscribe to my mailing list at bryankeating.com. You'll get resources from John, from Frank Willis. check from Michael Saylor. We've been doing so many phenomenal interviews. But for now, sit back, enjoy. And please, before you go, take a second to go down to the app reviewing section of, or the podcast reviewing section of this app. And please do leave a tiny constellation,

asterism of stars, ranging from one to five. I hope five, but maybe one. Let me know what you think of these episodes of the Into the Impossible podcast. We are getting so much great feedback from you, my listening audience that I just love so much. And I hope that you will do me that honor. It's really the only thing I'm asking for. And I do read every single one of them. Here's one that just came in just today from Jason Werner. Professor Keating manages to bring an impressive mix of brilliance to bear within this podcast through vocal and bleeding edge guests from countless sectors and philosophical Realms. The podcast is a must subscribe, but definitely check out the series on YouTube for its live and event-based programming. Cheers. Yeah, so please do that. Another review, I can't believe this,

I can't read this. A horrible podcast is on the air. He's a terrible, Barbara Keating. That's my mom. Oh, my God. How could she do it? No, she didn't do that. But anyway, please do subscribe and leave a comment because it will really help me out. Thank you so much. Now, enjoy this podcast. Any sufficiently advanced technology is indistinguishable from magic. Since I can't believe John, it's been many years since we first met, and it's none other than John Preskill, who is joining us today from Pasadena, California. John, how are you doing today? I'm doing great. It's really good to see you, Brian.

It's great to see you. I can't tell you how much it means that you're on the show. you're basically my most requested guest that I get on a frequent basis. Although we're branching into alternative topics now, we just had on our first show about Bitcoin, and we will talk about quantum cryptography, and we will talk about cryptocurrency and all sorts of things. So just for those out there, what are the promises and pitfalls, perhaps, of things that John has made foundational contributions to. But I really wanted to start off by talking about what the nature of your research is and how people can learn more about your research.

We figured that we would kind of discuss the implications and the applications of quantum computing to fundamental physics. My audience is very astute. They know a lot about quantum mechanics and about issues related to quantum mechanics. But can you give a quick update on what, or quick summary? How do you think about a quantum computer? What is the essence of a quantum computer to John Presco? Well, it's a device which makes use of quantum mechanics to achieve big speedups compared to classical computers for solving certain classes of problems.

And the types of problems for which a quantum computer is particularly well suited, is of characterizing the behavior of quantum systems. So, you know, from my point of view, as a physicist, and with the background in particle physics, the way I think about this whole field of quantum information is that we are in the early stages of exploring a new frontier of the physical sciences. You know very well, Brian, that if I want to go more, deeply into the early history of the universe.

As you do in your work, then we build more powerful instruments. Or if we want to explore the structure of matter on shorter and shorter distance scales, we build more powerful particle accelerators. If we want to explore matter as it becomes more and more highly complex, that's what we can do with the quantum computer. Because when we have many particles interacting quantum mechanically, states become very entangled. That's the word we use for the characteristic correlations among parts of quantum system. And with ordinary computers, we just can't simulate the behavior of highly entangled matter efficiently.

And that opens opportunities for discovery. So sometimes I speak of the entanglement frontier or complexity frontier. because we would like to go more and more deeply into the behavior of these very highly entangled states. And a quantum computer is the instrument, if you like, which will allow us to do that. And when you think about the fundamental paradigm of biology, you know, it might be evolution progressing by, you know, encoding of information via DNA. And in cosmology, it's characterized, the universe characterized on large scales by, you know, scales by isotropy, homogeneity. Is there a fundamental theorem or catechism of quantum computing,

or is it still, as Cho-in-Lai said of the French Revolution in the 1970s, too soon to tell? Well, there's sort of a fundamental principle of computational science, which sometimes people call the Church-Turring thesis, which is that we should be able with a, computer that's universal, that can do any computation, to efficiently simulate anything that happens in nature. And that's not something that we can prove. It's a statement about physics, really, but it was a widely accepted principle until the advent of quantum computing. Because we don't think, with the standard model of computation, the Turing machine, model that we can efficiently simulate very highly entangled matter. But with a quantum computer,

we think that is possible, that with a quantum computer, as far as we know, we should be able to efficiently simulate any process that occurs in nature. And so we need to update that church touring thesis, replace it by a quantum church touring thesis. And so I would say our operating principle, when we think about using quantum computers to explore nature, is that if it happens in nature, we have the hubris to believe that we can simulate it with computers that we understand in principle how to build. But if we want to do those simulations efficiently,

they'll have to be quantum computers. So I've said actually getting a couple of comments in the chat that your audio is still a little on the low side, John. Is there a way you can either move closer or is that that's an external microphone? Is there a way, do you have an internal microphone if you just disconnected the, let's see if that would work? Well, that's already better, actually.

Okay, well, I don't think I did anything. I turned my head. The microphone is. Yeah, the closer you can get to the microphone, the better. All right. I got it a little closer to my face now. Oh, that's much better. Yeah, people are.

asking if this is quantum audio, they don't want it. So the future better be not this level of audio quality. Apologies for that, everybody, but we're here with John Prescoe. Boy, they guys miss some really great stuff. Yeah, I know. We'll never repeat it. No, we got all of it. And it's just the audio is just a little inconsistent.

There's a little bit of background. But anyway, we're here with John Prescoe into the Impossible podcast. Please do subscribe to the channel. exercise your finger. John will tell you that carpal tunnel syndrome is rampant. The best exercise is to press every couple of seconds, the like button, the thumbs up button, and subscribe, but also subscribe to my newsletter at briankeating.com. I'll send you some resources from this conversation with John. We're going to talk about really some exciting developments in fundamental physics. So we talked a little bit about the fundamentals of quantum computing, maybe what the

fundamental theorem of quantum calculus is in comparison to cosmology and, and, and, and, and and other branches of science. I've heard it said, and this is all with due respect intended, that the only thing that quantum computers are good for is cryptography and simulating quantum computers. What would you say, after you punch someone in the face who would say that to you, what would you say to such a person? Well, I would say we have a very limited understanding

of what quantum computers will be able to do. We don't expect quantum computers to be able to efficiently solve all the problems we might care about, far from it. In particular, the problems that the computer scientists call NP hard, or NP-complete, the hardest problems among those for which we can verify the solution once we find it. Quantum computers can speed up finding solutions to such problems just by speeding up the exhaustive search for a solution, but they don't speed that up so dramatically that it will really, you know, change the world. So it is special problems with the right structure for which quantum computers have a very large advantage,

for which the speed up that you can achieve is essentially exponential. You know, you can have a problem whose solution takes a time, which is exponential in the size of the problem on a classical computer and in some cases solve that far more efficiently in a time that just scales like the power size of the problem. So that's where quantum advantage is really spectacular.

And you mentioned the two examples which are most prominently known. The application to cryptology and the application to quantum physics. Now, it's likely there will be

many others. But I think partly because we have limited imagination and limited understanding of the power of quantum computing, we haven't, as theorists, done a very good job so far in identifying a much broader class of application areas. And in a way, maybe that's not so surprising. in the case of classical computation, there are many examples of algorithms which theorists weren't able to promise would be powerful, but by experimentation,

it was found that they work quite well. When we have quantum computers to experiment with, which are of sufficient scale that they can start to solve hard problems, we'll have the opportunity to try out heuristic ideas for quantum speedups and see how well they work. And the hope, which we can't guarantee, is that that will open a lot of other potential applications,

which can be very impactful on society. But for me, as a physicist, even if what quantum computers turn out to be best suited for is simulating quantum physics, that's already a very powerful and interesting application, not just for advancing science, but I think for practical applications, at least in the long run,

we're not sure how far off in the future I'm talking about in discovering new materials and new types of chemical catalysts and so on. So those applications are part of what propels the interest in the field, but realistically, that's going to require a much more powerful quantum computer than we currently have, and it might still be decades away before we'll see that kind of impact. As for cryptology, well, of course, you know that what generated great excitement already over 25 years ago was in 1994 when Peter Shore discovered.

that with a quantum computer, we can speed up greatly solutions to some number theoretic problems, like finding the prime factors of a large composite integer, and because the public key crypto schemes, which we now all use in our daily lives to protect our privacy when we communicate over the internet, since those are founded on the assumption that certain computational problems are too hard to solve in practice, that's going to be upended when quantum computers are widely available. I think, you know, so the world is going to have to adjust to that, and we can talk about that more if you like, I think as the world adjusts, that application will sort of

recede an importance. It's sort of a historical accident, I expect. that quantum computers came along at a time when these public key schemes were in widespread use, and there are alternatives that will use in the future to replace those schemes, will still be able to protect our privacy in ways that are not vulnerable to attack by quantum computers. And then the application which will really impact the world that we are confident about, and understand reasonably well is the application to characterizing and simulating quantum systems. And when we look at quantum systems, I find most remarkable, although I'm a complete neophyte in this field, and I just look at it with admiration and respect for both the enthusiasm of the people working in this field,

which is quite different than your early work, is a grad student with Stephen Weinberg, and your contributions you made to topological defects, the magnetic monopoles, and cosmology. And we'll get to cosmology, I hope, in black holes in your relationship with Stephen Hawking and the many bets you guys engage with. But I do want to point out that from this novice's perspective, one of the most remarkable things about quantum computing

is that it can be done at all. Because either we use the properties of single quantum systems, you know, individual electrons, their spins or some, or collectives in persistent current. and phases and quantum incubates. We use Joseph's injunction, superconducting quantum interference devices, squid amplifiers in our microwave background research.

What's the most remarkable thing to you? Is it that we can manipulate both mesoscopic, macroscopic, and the microscopic? Or is there something else that really betwixtes you and really stokes your curiosity about this particular field of quantum mechanics? Well, it was also important that when the idea of quantum computing

came along and when the recognition grew that quantum computers would be able, in principle, theoretically, to solve hard problems we couldn't otherwise solve, the developments in experimental physics were poised to take us into the era where real quantum hardware could be constructed. You know, when decades ago, when one talked about doing, experiments with quantum systems, they were almost always ensemble experiments, where, say, if you were trying to describe the behavior of an atom or doing nuclear magnetic resonance or something like that, you were looking at a signal from many atoms or many electron spins and so on. But, you know, by the 1990s, we had the tools to manipulate quite accurately to control single quantum systems like single atoms. And that was technology was developed in part because it was motivated by the desire to make more accurate atomic clocks, like trapped ion technology, for example. But it's also very well-suited.

for doing quantum computing, where one needs to be able to control individual qubits, or even more importantly, the interactions between pairs of qubits, qubits, meaning the quantum analog of bits, quantum bits, which we call qubits, two atoms, for example. It's possible with lasers and using the vibrational modes, for example, of ions in the trap, to control quite precisely the interaction between two trapped ions and therefore to do an elementary step in a quantum computation by putting together many such operations. You can build up in principle computations of arbitrary complexity if you can perform those operations with sufficient accuracy with good enough control. So it's nice that technology came along or was starting to be available. just at the time that we recognize the potential of quantum computing. And the promise of quantum computing has continued to drive the technological developments,

which have become more and more sophisticated, with the goal in mind of scaling up quantum computers to devices that can really solve interesting problems. But that technology can have other spinoffs as well, like better purchased to metrology, more sensitive sensors of various kinds, and potentially to the cryptographic applications as well. Yeah, so these are all phenomenal applications, and I think some of them are less well-known than others. I am getting another request to have you move a little bit closer to the microphone,

and I think eventually you'll give up as you start crashing into the screen of your computer. But if you can move a little bit closer, I think that would be great. I will get closer to the microphone myself, although I am pretty close as it is right now. So people are asking about, yes, the implication for things like quantum supremacy that has been a term that you coined and used. I think probably misunderstood. First, before we get into applications or perhaps the prospects for quantum supremacy, both inside of physics and maybe outside of science altogether in cryptography and banking, etc., can you explain what you mean by quantum supremacy and how it should be used, maybe how it shouldn't be used, what are your thoughts for the audience about this term that you coined many years ago? Well, I think one of the most remarkable things that we've understood about the difference between quantum and classical systems is what I made reference to earlier when I spoke of the quantum church-turing thesis, that we think that in general you can't efficiently simulate a quantum system using any classical system. I think that's one of the most interesting things we've ever said about how quantum is different from classical.

And so there's a strong incentive to try to validate that in the laboratory, in a real experiment, to the extent that we can. And so I suggested that as a challenge for the field to do experiments, which indicated that we could do something with the quantum device, which surpasses what we could do with our most powerful classical computers. And I called that quantum supremacy. The concept was not new at the time that I proposed it, of course, but it's nice to have to frame the challenge in kind of a concrete way. And, you know, various experimental groups and companies took on that challenge and with some fanfare, the Google group announced in late 2019 that they had achieved

what they asserted was a demonstration of quantum supremacy. They built a very interesting piece of hardware in the Google AI quantum lab based on superconducting technology. Not that different from the technology you use in your detectors. The magic coming from the wonderful Josephin Junction, which makes things sufficiently nonlinear that the quantum physics becomes interesting. And using that technology, they built a device with 53 qubits, and they were able to perform a sequence of entangling gates,

which cause neighboring cubits in a two-dimensional array to interact with one another, up to 20 layers of such two cupid gates where in each of those time steps all of the cubits were participating in the gates that were applied, and then they measured all the cubits. And it's a rather noisy device far from perfect, so when they measure, most of the time they get junk,

but maybe a few times out of a thousand they get a useful result and then they can run that computation millions of times in just a few minutes in order to boost the signal to noise and get some statistically useful information from the experiment. And then you could imagine

trying to simulate what that device is doing on a classical computer and it's possible, but it's sufficiently challenging that that classical computation would take far longer than the few minutes that it took Google's device to do the same task. So I like to call it Quantum David versus Classical Goliath because the classical supercomputer covers a couple of basketball courts, and it's using megawatts of power. And Google's device is just, you know, a little chip in a dilution refrigerator, more of a desktop-scale experiment. And yet that little chip is doing something that that big, powerful supercomputer, well, I wouldn't quite say can't do, but it would take far longer for the supercomputer to duplicate the task.

and so that was more or less what I envisioned when speaking of quantum supremacy. Now that's not to say that it's of any practical importance that that demonstration was achieved. The particular task that the Google device carried out is not something that we particularly care about for any reason other than demonstrating the quantum supremacy. but I still consider it to be kind of a milestone in technology that we've sort of entered the era where quantum devices potentially are capable of doing things that we couldn't otherwise do,

even with our most powerful classical computers. Yeah, so one of my guests are folks in the chat room, I want to point out, we're talking with legendary physicist, the Richard Feynman Professor of Physics at the California Institute of Technology, which is a small technical college in Pasadena, California, where I've spent some of my time. This is Beaver's Week into the Impossible podcast.

We had Frank Wilczek on Monday. We had Michael Saylor, MIT alum, on Wednesday, or Tuesday and Wednesday, respectively. And now we've got Beaver from the West Coast, none other than my friend and really a mentor in many ways that he looks at physics the way that I would like to someday see it. I call John kind of the Chuck Yeager. I'm a pilot, John.

I don't know if you remember that. But all pilots emulate the late great Chuck Yeager. You know, we're talking to air traffic control. And we're saying, we're going to go to the Northwest sector. And we try to emulate. Well, physicists, we try to emulate you, John. You don't have to respond to that. But I am one of-

I will say the right stuff is one of my favorite books. And the part about Yeager is the best part. Sorry to interrupt you during your enjoyment of the Into the Impossible podcast with my friend. my hero, my mentor, John Preskill. In his first interview of his kind, I hope you're enjoying this unique opportunity to learn from a great physics titan,

one who is deeply connected to physics past and its future, Richard Feynman, up to Murray Gelman, up to Lenny Suskind, Sabina Hassanfelder, and beyond. So please do me a tiny favor. Please like this or subscribe to the podcast, first and foremost, if you're listening to this. And if you're on iTunes, please leave a review, please leave a quick rating, which it could be take you one second to leave a one star to five star review.

You'll hear reviews that I will then read to you like this one by Benderbilt. Dr. Keating is a fantastic interviewer, and he interviews many fascinating guests. You do not hear from elsewhere. 11 out of 10. Wow, 11 out of 10. That really makes up for my mother giving it the 0 out of 10. Just kidding, Mom. Anyway, please do subscribe and leave a short.

quick review. I read everyone. I'll read yours on the air next time on the Into the Impossible podcast. Thank you. Now, back to our regularly scheduled programming with Professor John Preskill. So going back to questions from the audience, one is asked by one of my most loyal guests. His screen name is six Bob Holmes. And he wants me to ask you about the physical versus logical number of qubits for a useful machine. I guess this is what is the minimum number of physical versus logical number of cubits for a useful machine, a machine that can do things that classical computers would be less efficient. Yeah, well, let me explain what I think is being asked here in the distinction between physical and logical qubit. As I mentioned, and for example, this Google

vice, the hardware, although impressive, is still quite noisy. And that's a serious limitation on the scale of the computations that you can do. Because if you try to perform a computation with hardware that's so imperfect, with a large number of qubits or with many operations with many gates, there will be no signal to noise. The imperfections will completely wipe out any useful result. And so if we're going to use quantum computers to solve really hard problems,

we have to somehow solve that problem to reduce the noise. And potentially, someday we'll be able to do that by making much, much, much better quantum hardware than we have now. I think someday we will do that. but there's another approach which is more of a software approach with what we call quantum error correction and in quantum error correction we use many physical cubits to represent a single so-called logical cubit where the logical cubit is much more reliable can be manipulated much more accurately than the single cubit and just to provide some context quantum computing is really hard. Yes.

That's why we've only gotten as far as we've had so far. And there are several things that make it hard. But perhaps the most formidable problem is what we call decoherence, which is an important principle in physics, that, of course, classical physics is what we use to describe what we experience in everyday life, and it works extremely well. And yet, underneath it all, we think the world is quantum mechanical. Quantum mechanics holds sway at the scale of atoms,

but at the scale of humans, it doesn't seem to be that important. And the reason for that is this phenomenon of decoherence, which makes quantum systems behave classically, to be well described by classical physics, which is fine because our lives would be much stranger if we were quantum mechanical at a human scale, although I'm sure that would be fun. But for a quantum computer, it's really bad news

because decoherence can make a quantum computer behave like a classical one, and all that magic that makes the quantum computer more powerful would be lost. And this decoherence occurs because we can't stop the quantum system, or it's very hard to stop the quantum system, from interacting with the outside world.

And this is really a manifestation of the uncertainty principle in a way that if you acquire information about a quantum system, then you have to damage the system. And even if we're not acquiring that information ourselves, if it's leaking to the outside world, to the environment, then the quantum system will be damaged, and that's going to make our computer fail. So we have to find a way of overcoming that phenomenon of decoherence,

and that's what quantum error correction is designed to do. And the principle is that we can put many qubits together in such a way that when the environment interacts with just a few cubits at a time, the information is extremely well-concealed. So it doesn't easily leak to the outside. world. The trouble is that to make that work effectively, we need many physical cubits to realize one very well-protected logical cubit. And I think that's what the question is about. Well, it's about two things, really. Because I think the question is how many logical cubits do we need

to do something interesting? And it was at least implicit that that's much different from the number of physical qubits that we need. And it's hard to be very precise about the first question. But once we have, you know, of order 100 cubits or

a few 100 cubits, and we can do many highly accurate gates, then that takes us well beyond what we can simulate with classic computers and we can envision applications to chemistry for example, if our gates are sufficiently accurate,

or assimilating materials that could be done with just a few hundred of these very well-protected logical cubits. But then there's the question, how many physical cubits do we need for one well-protected logical cubit? And that's also a hard question to answer precisely because it depends on several things. One is how accurate is our hardware.

If the hardware is so pathetically unreliable that the quantum computer is making mistakes all the time, then quantum error correction isn't going to work at all. It's not going to give us any improvement unless the hardware is good enough. And then the better the hardware gets, the less overhead costs we have to pay to make a very reliable logical cube. but you know it's likely unless cubit technology gets much much better than it is now which i think eventually will happen um that uh you know we will need uh potentially hundreds of thousands of physical cubits for one really well protected logic cubit so that's pretty discouraging yeah from the point of view of the current state of the technology where you know we maybe will have devices

is with the border 100 physical qubits in the near term, but we have to make a very big leap to maybe millions of physical qubits to get into that regime where we can perform computations that are reliable and far surpass what we can do with our classical computers for applications people might care about. Interesting. So I have a question coming in from Ernesto Eduardo Dolbar Ganes, and he is asking, is there any implicit Moore's law equivalent for quantum computing? Well, I don't know.

Now, people have tried to imitate Gordon Moore and make a plot of how quantum computing has been improving over the last 10 or 20 years to spot such trends. we speak sometimes a bit jokingly, perhaps, about Shelkoff's law after Rob Shelkoff. You probably know Rob. The name. I don't know. Yeah. He was a Caltech student who worked on superconducting devices sometime back.

But, you know, Rob has said that we get something like an order of magnitude improvement in the superconducting quantum devices every grad student generation over however many years that is, maybe three or four. And that was true for a while, but the metric that he and others used was what we call the coherence time, how long a superconducting qubit can sit around before it gets hit by an error.

And that did improve very impressively for a while. It might be slowing down now already. But that's not the only thing we care about anyway, because we really need accurate quantum gates. And the coherence time is about storage errors. You know, if we put a qubit in quantum memory, and we come back an hour later, is the qubit still there intact?

But even if that's true, it doesn't. necessarily mean that we can do entangling quantum gates on pairs of cubits with very high accuracy. In fact, the competing technology of trapped ions, the cubits are just wonderful. The qubits are very well isolated, single atoms, and you can prepare a cubit state in an atom, and it will survive for many hours if you just leave it alone. but that's not enough for quantum computing. For quantum computing, we need to do these accurate two-cubid gates. And the state of the art is that there's an error rate per gate

under the best conditions now of about one in a thousand. And, of course, we'd like to see that continue to improve. But it's hard. It's really hard. It's all hard. building a quantum computer is really challenging. That's my next question. That comes from someone named Brian Keating, which is that most of the applications that we have in our laboratory require,

I always joke people come to San Diego for the weather, but in the case of our quantum computer, you know, capable laboratory that uses these Justin Junctions and squid amplifiers, we are the coldest part of, you know, San Diego, to my knowledge, although some of my physics colleagues can go a little bit cooler, but we get down to six millicelven, six one thousandths of a degree above absolute zero. These are, you know, uncomprehensible temperatures. And we're finally getting some macroscopic, you know, cooling powers of order microwatts,

you know, and abilities to remove heats of joules, you know, kind of a day timescale. But, you know, what are kind of the limits on practical quantum computers? I know you're a theoretician, but you have a lot of contact with the experimental community. and obviously you advise many of these people that are attempting to overcome current limits. Is it going to take developments in room temperature super connectivity to have practical quantum computers that we could actually use? Or is it, you know, are we basically just going to have a couple of quantum computers around the world and they'll do everything, you know, for the entire planet's computing needs?

First of all, it's important to that there are a lot of, different approaches to the technology which are advancing in parallel. Because we really don't know at present which technology has the best long-term prospects for scalability to large devices that can solve heart problems. And they all have their characteristic advantages and disadvantages. Some are cryogenic, like your detectors. and operate at 10 to 20 mil Kelvin in a dilution refrigerator. That's true of the superconducting devices.

It's true of some of the spin-cubit technologies where the qubit resides in a spin of a single electron. It's not true of all the technologies, or at least not to the same extent. So in the case of ion traps, they can operate at room temperature. It turns out that they're quieter, they're less noisy, if they run at liquid helium temperature in about 4 Kelvin. And at least some of the ion trappers are going to go in that direction to get more reliable hardware. Another example is what people call NV centers, which is, defect in material. It could be in diamond, for example. And in that case, you can get long coherence times even at room temperature. And that might be important for some sensing

technologies, for example. But your broader question is, what do we really need to advance the hardware? And there's no single answer to that. In some cases, materials advances are going to be important. That can help with the superconducting technologies and with the spin qubits. Having more accurate methods for control, particularly ones which are scalable to large systems. That's also important. Designing better qubits. qubits, in the case of a superconducting device, there's a lot of freedom in how we fashion a qubit out of the

ingredients, putting together joseous and junctions, for example, in an electrical circuit, there are a lot of ways to do that, and some of those are more resistant to noise than others. And so that's another important way to advance the technology. and the extreme case of achieving big progress through improved materials is what we call topological quantum computing, which is an idea for realizing materials that have a kind of intrinsic resistance to the errors that afflict quantum hardware. And that in the long run could be a breakthrough technology, but like everything else, it's very hard to achieve. Yeah, I'm actually getting a question. I was about to ask you for 20 pounds of British sterling from a friend of ours, Stefan Roche in the UK. And he's asking, does John think that

quantum materials such as topological insulators or other manifestations of entanglement in Maharana, maharana fermions, or in many body physics, can offer more possibilities for information processing. From what you just said, it would seem the answer is yes, but primarily for error correction. Is that correct? Well, that's the advantage that topological materials potentially have. It's a wonderful idea. I mean, for a theorist, it's an irresistibly beautiful idea. It came from the fertile imagination of Alexei Kataia, my Caltech colleague.

Actually, I'll tell you something funny about that, which maybe says something about me, and also about Alexa. So, Alexi's idea was to take advantage of something called an enania. which is a concept and a word, you know, which Frank Wilczak is responsible for. It means particles that have interactions that we call topological. What does that word topological mean? It means, well, you know, it's topologists talk about properties, which remain unchanged when you smoothly deform an object. So you can't smoothly change a sphere into a donut.

And it's very natural to think about topology in the setting of quantum computing because we would like the way we do are quantum gates, the fundamental quantum operations, to remain invariant when we deform the hardware by introducing the noise. and Kataev

proposed this in 1997 and what's funny is that over the preceding eight or so years I had been very interested in

anions and in particular what we call non-ebellion aneons which are the ones which are promising from the point of view

of quantum information processing and then I got very interested in quantum error correction, how we can make quantum computers robust in the presence of noise. And somehow, I didn't appreciate that these two ideas are extremely closely related. That, in fact, anyons can provide an approach to doing quantum error correction, which has

some characteristic advantages. Kitaya did realize that. and when I first heard about the idea during this first visit to Caltech in 1997, I was immediately excited because I was perfectly prepared to appreciate the idea and how brilliant it was, but I hadn't thought of it myself. And so that tells you something about me, and it tells you something about Kataia. But I guess it also illustrates the principle, which is often applicable, that things that seem obvious in retrospect are not necessarily obvious at the time of their discovery.

And so that was one of the things I missed. And, you know, it was just, I think, a manifestation of my own limitations. Yes, well, we all have blind spots in our education. But yours are very much more narrow than most of us mortals. So getting a couple more questions, but I do want to remind folks that we are raising funds for an extremely worthy cause, close to my heart, close to John's heart. And that's the Foothill Unity Center, which is located in Pasadena. Let me see if I can get that up on the screen here. There we go. Foothill Unity Center. There we go. So very worthy charity with many great beneficiaries and doing just a tremendous amount of good in my former hometown.

When I was at that technical college known as Caltech, please get involved. Donate in the super chat and I will double it. We've already gotten a tremendous number of donations. Thank you so much, but we can always use more so they can do more. And it's a very, very important project, rather, for 2021. So help us do good. My audience is the best in the known multiverse. We're going to turn to cosmology in a second.

I do want to ask folks to please subscribe to this channel, exercise your face, regularly. Real doctors will tell you to do that. Push the subscribe button, push the notification bell, hit the thumbs up button, and that will make folks happy. If you're listening to this on iTunes, please leave a review and a rating because that helps us get great guests like John. You know, he looks at my ratings and reviews, and he decided he would come on finally because of the erudition, the perspicacity of this amazing fan base. And you could tell John from the questions we're getting, this is no ordinary YouTube unboxing channel. We're going to unbox a quantum computer next time you come on the show. I want to turn towards cosmology. We've had on

folks like Frank Wilczek, we've had on Roger Penrose, we've had on Leonard Malad Now, and also his friend Deepak Chopra. Actually had a conversation live, the four of us, Deepak Chopra, myself, Frank Wilczek, and Leonard Malad Now, and we talked a lot about theories of everything, about the essence of singularities, et cetera. The first thing, before we, we take a deep dive into cosmology, I want to first start with a bet that you made with Stephen Hawking, which according to Leonard's book on his friendship with Stephen Hawking, he says a bunch of things, but he remarks specifically about you that nowadays your whole third quarter course at Caltech for graduate students in general relativity when you teach it is

just dedicated to Hawking radiation. So is that true or is that false? it was once true. I haven't taught that course for a while. In fact, I first taught it, gee, I don't know, I don't think Leonard was around at the time, but I guess he heard about it. I first taught it in the early 90s. And actually, I was encouraged to do that by Kip Thorne, who thought it would be a natural way of capping our three-term general relativity sequence to devote to the third term after they've already learned the principles of general relativity and about properties of black holes to the quantum field theory on curved space time, which is relevant for understanding hawking radiation. Now this was, as is often the case, very educational for me, because although I had some

working knowledge of the topic, you know, when you teach a class, you just get a much deeper understanding. And so I worked really hard on that class. And I did teach it a couple more times. And the goal was to start with no prior knowledge of quantum field theory, but to the students were in the third term of a general relativity sequence. So they all knew something about general relativity. Although for this course, you didn't need to know that much about that either,

except some appreciation for what a black hole is, and to, starting from nothing, more or less, arrive at an explanation for why black holes emit walking radiation. So that was a lot of fun, and it deepened my understanding of that subject, and it also got me more deeply involved in my own research in thinking about the quantum behavior of black holes. as has often been the case for me, and I'm sure others, the store of knowledge I collected in the process of teaching that material of something I've often drawn on in my own research in the following years.

So I actually have a bet from theory.caltech.edu, and that is whereas Stephen Hawking and Kepthorne firmly believe that quantum information, that information swallowed by. a black hole is forever hidden from the outside universe and can never be retrieved, revealed, even as the black hole evaporates and completely disappears. And whereas John Preskill firmly believes that a mechanism for the information to be released by the evaporating black hole must and will be found in the correct theory of quantum gravity, therefore Preskill offers and Hawking accepts a wager that when an initial pure quantum state undergoes

gravitational collapse to form a black hole, the final state at the end of the black hole evaporation will be a pure quantum state. And the winners of the encyclopedia of the winner's choice. And this is witnessed Kip Thorne, John Preskill, Stephen Hawking, 6th of February 1997. So now you won that bet, but I want to push back just gently and always with respect. And I brought this up to Roger Penrose, even himself. And I want to bring it up with you. I always hear the following. We need a theory of quantum gravity because at the singularity, the singularity, of a black hole's core, the laws of general relativity must break down, and the singularity must therefore be a reflection of the fact that some other physics is needed. That's one

instantiation of the notion of a singularity. The other one being the initial singularity that was believed by Hawking and Penrose to be the initiation point of our current observable universe. I point out the following, you know, the Godunkin experiment. What if you get a letter from the old one, as Einstein used to call? Here come my finger puppets, my voodoo dolls. Here we go. I got one of you, but it's on order. It's got too many pins in it.

But let's say the old one, as Einstein called God, tells you, well, the Big Bang might not have happened as Sir Roger, newly minted Nobel laureate, who left his Nobel Prize in my office foolishly. But it may not have happened. It may be conformal cyclic cosmology all the way down. Furthermore, we have no ability to access the singularity of a black hole and no way to get the information out, even if we could access it. So what if there are no singularities? Would you still say we need a theory of quantum gravity?

I would say we, whether you needed, I couldn't say, Brian. But we want to understand what happens. inside a black hole. It may seem rather exotic, well it is rather exotic, let's face it, to consider experiments in which we enter a black hole and see what happens to us. And I think part of the point you're making is that if we stay outside the black hole, we won't have, as these said, not in any obvious way, direct access to the outcome of an experiment that occurred inside. But you know, for all we know, we're inside a black hole now, Brian. It might be a really, really big one, and we're tumbling towards the singularity, and that would, you know,

focus our attention more on wanting to understand what happens in the late stages of gravitational collapse. Whether you want to call it quantum gravity or not, that's just words. You just want to understand how it works. Now, as far as early universe cosmology, she is concerned, I'm a little surprised hearing this from you because what I hope you will succeed in doing someday is finding evidence for primordial gravitational radiation from very early in the history of the universe. And that will, of course, be a real milestone in physics. and we hope it will illuminate our understanding

of very early universe cosmology and that in itself is really quantum gravity in a way because we're talking about gravitational waves that were produced by quantum fluctuations in the very early universe at least that's our

current understanding of the explanation of the signal that you've been looking for that's quantum gravity and we hope that'll give us some insight into the so to speak initial singularity. Hello everybody. We hope you've enjoyed part one of Brian Keating's extended conversation with John Preskill.

Part two is available now. Please subscribe, comment, share, and review. Watch on YouTube, listen on iTunes, Spotify, Google Player, Stitcher. We appreciate hearing from you and are always open to your suggestions for future episodes. For more information, and to sign up for Professor Keating's mailing list, go to Brian Keating.com. Follow Professor Keating on Medium and Twitter at Dr. Brian Keating,

DR. Brian Keating. For more information on the Clark Center, go to imagination.ucsd.edu. Into the Impossible is a production of the Arthur C. Clarke-Clark Center for Human Imagination at the University of California, San Diego, in the Division of Physical Sciences. Eric Vary, Director, Ryan Keating, co-director, produced by Ryan Keating and Stuart Balco.