Secular Cycles - Peter Turchin
Del denne Masterclass
Dette lærer du om i denne Masterclass
Velkommen til Lørn dot Tech. En læringsdugnad om teknologi og samfunn med Silvija Seres og venner.
Silvija Seres: Hello and welcome to Lørn Masterclass. Our topic today is Quantum Computing and our guest today is Dr. Mark Mattingly Scott. I’m Silvija Seres and I’m going to introduce Mark in just a second. I’m going to say two words about the series itself. So, Mark, you are now a professor, but in a conversational set of interviews, lectures for lectures at 30 minutes each, where the first one is going to be about introduction to the subject of quantum computing. The second one is going to be about your favorite examples in the topic of quantum computing. The third one will be the tools of the trade. If somebody is considering getting an early move in applying quantum computing, where would they go to start both learning but also doing? And the fourth lecture will be an attempt to do a workshop. Okay, so let’s think, how could I apply this? About 30 minutes each and very informal and chatty. So with that, welcome, Mark.
Dr Mark Mattingley-Scott: Thank you, Silvija. Good to be here.
Silvija: So you and I have chatted about the subject before, and every time I try to understand quantum computing and I have a background in theoretical computer science and maths, you know, my brain gets fried and you are the only man who managed to get me at least closer to the subject. So I thought I’d like to share this with the listeners of Lørn. So thank you so much.
Dr Mark: Well, if it’s any consolation, my brains are seemingly in a state of permanent deep frying. I can hear the fat sizzling in the background. So you’re not alone?
Silvija: No, it’s a consolation to hear that. So tell us briefly about yourself. Who are you? Why do you care about quantum computing?
Dr Mark: So I’m a British guy. The name is Scottish. Well, there’s some Scot in there, but I was born in England, so I better say I’m British, otherwise somebody will get offended. I live in Germany, in Heidelberg, and I’ve been living here for 32 years. I moved over in 1989 from a job in electronics research in London. I’ve done my PhD in Durham in northeast England. I’d been working in electronics research in London, and I basically got an offer. I got asked to come over to Germany to work for IBM as well, I call it a midwife to the World Wide Web. So the World Wide Web had just been thought of as highly scalable hypertext systems, and IBM was interested in this and how to make this work with their products and their platforms and recruited me and a bunch of other people. And we were the midwives to the web at IBM, 32 years at IBM or just actually just short of 32 years at IBM, basically always working on innovative technologies, developing innovative technologies. I left IBM on the 30th of August this year. So, yes, exactly as this recording happens 13 days ago or 14 days ago, and I’m now managing director of a company called Quantum Brilliance, their German subsidiary and a general manager of the Australian Mother Organisation for Europe, Middle East and Africa.
Silvija: And quantum brilliance sounds like it’s focusing on quantum computing.
Dr Mark: It is. And there’s a giveaway in the name. So we’re I think we’re the only, yeah, more or less the only company using or implementing qubits, things called qubits, which work at room temperature using diamonds, hence the word brilliant. So everybody loves diamonds. We’re using them to make quantum computers.
Silvija: So at least we’ve started with images that people can start having in their minds, probably completely wrong images. So you’ll have to tell us about the size of these diamonds and the role of these diamonds. But before that, can you just open the word qubit for us? What is a qubit?
Dr Mark: Yes, I can. Or I can try. And I’ll try not to fry my own brain. So quantum computing is something truly new. And this is from the mouth of somebody who’s been saying this is truly new for lots of things over the last 30 years. But this time, that statement is grounded in fundamental physics. So we’re now able to use a principle of fundamental physics, which was basically exposed about 100 and 1020 years ago. So up to that point, physics had been everything had been ticking along nicely and and and everything was consistent. And then things happened, which led to some contradictions and, it was, among others, Albert Einstein who came up with an explanation for why this was what might be happening, which then led to the development of a field of physics called quantum mechanics or quantum physics. Out of that, in the actually starting from the early 1960s, from the year of my birth, 1961 through to the early 1980s, a series of developments where the possibility of using small quantum mechanical systems to compute, to perform computations became a dream and then a possibility and then reality.
Silvija: So what you’re saying, Mark, is that we are approaching a position where quantum computing might become a practical reality.
Dr Mark: I think we’re actually there. So we have these things called qubits. And a qubit is basically the simplest quantum mechanical system you can imagine. There are a number of ways to actually do that in the practice. What it comes down to is you’re either using the so called magnetic or electromagnetic spin of an electron or electrons. When they circulate around the nucleus, they either rotate in one direction or in the other direction. You can trick those electrons into having spin, which is, as long as you don’t look at them, in both directions at the same time. This is where your brain starts to fry. Or you can use the nucleus, what’s called nucleus spin. Or you can use both.
Silvija: But you basically make these things spin. Or you exploit this uncertainty or…
Dr Mark: Exactly. Exactly. So the best analogy I have is if you want the answer to a problem and you can formulate that problem as a one of a number of possible outcomes.
Silvija: Then the cat is alive or the cat is dead.
Dr Mark: Yeah, but imagine something much more complex with not just two alternatives, but two to the nth, thousands or millions of alternatives. Then if you take a truly random number generator and repeat whatever it is you’re trying to do enough times, sooner or later it will come up with the right answer. What a quantum computer does is it accelerates that. So it removes the sooner or later and makes that sooner. So what it does is it says for a fixed number of small fixed number of repetitions of what we call a quantum computing program or a quantum computing experiment for a small number of experiments repetitions, it will produce a statistically significant answer. So it kind of gets around or cheats pure chance uses quantum mechanics to speak. Yes, I could see.
Silvija: That’s so. To a person who’s studying algorithms and complexity theory, which is my kind of thing, where I was trying to solve this clue to the power of end problem by thinking about very smart algorithms. You’re actually making me completely irrelevant because you’re saying, don’t worry about that. We’ll just have a computer that can calculate all those possibilities at once. And we don’t have to worry about computational time anymore.
Dr Mark: Yes. So you remember non deterministic Turing machines from your computational theoretical computer.
Silvija: Maybe you can teach our audience a little bit of the basics of computer science as well.
Dr Mark: Well, here’s where you correct me if I mess this up. So basically a guy called Alan Turing, who was a genius, worked for the British coding or the Department of Funny Works in the war decoding or helping to decode the German cipher. Came up with this idea. It’s a very theoretical idea of what’s called a Turing machine, which is a kind of abstract, finite state machine. And he determined that there were computations, computations you can compute on this abstract Turing machine, essentially what it looks like…
Silvija: Like, a model of a computer?
Dr Mark: Yes, yes and no. He was actually linking it to a physical piece of hardware. So the idea was you have a tape with symbols. On the tape, you have a red head which looks at the symbol. And then based on what it sees, it can move the tape forward or backwards and it can rewrite the symbol. And he proved that there’s a very large class of problems which you can solve with this Turing machine. These are Turing computable problems. And he proved that that the class of problems that could be solved in what’s called polynomial time. So a time proportional to the size of the problem. The whole field of computational complexity theory really took off in the same time as at the same time as the birth of qubits. So the early sixties, when it became clear that there were other kinds of problems which you could use a Turing machine for to help you with, in particular a class of problems where you could use what’s called the non deterministic Turing machine. And that’s one where you basically randomise it and repeat, repeat and rinse many, many times. But at each decision point, you add a random factor. It doesn’t really get you anything because you have to repeat. You have to repeat that particular program a number of times, which is combinatorial, basically proportional to the power or size of the problem. Then quantum qubits came along and quantum computers came along, and we can basically remove that particular constraint. We can find a solution to problems in a similar way to a non deterministic Turing machine, but much faster.
Silvija: So… So…
Dr Mark: And how does that work?
Silvija: No, wait, wait. We’re not there yet. But so you can now solve many of the problems that we thought were either unsolvable in any kind of human time, which is the basis of things like, let’s say, cryptography, where you create these security solutions that require computers to compute so much that you couldn’t, you know, you have to know the mechanisms, otherwise you can’t. And now you can break that with quantum computing. So there are so many problems, also things that artificial intelligence does or things like materials, technology, protein folding, health stuff, other stuff that we will talk about later, which suddenly moves from not just the realm of science fiction, but basically the realm of magic into practicality. I mean.
Dr Mark: Yeah, there is. There is. Yeah, it’s very strange why that particular randomness finding patterns within true randomness and that quantum quantum mechanics and qubits give you that. Do I have an explanation as to why? Does anybody have an explanation as to why? There are many, many interpretations about why, like the many universes argument, quantum determinism. I think if you take 11 physicists, then you’ll get 12 explanations. Probably the most promising for your listeners is something called quantum Darwinism, which is probably the only explanation which leads to experiments that can actually be performed. And you can look at the results and check whether it’s false or not. But in essence, the jury is out on why quantum mechanical systems and quantum computers in qubits, actually, fundamentally, why they provide this speed up. But they do. And you’re right, there is a large group of problems which under certain constraints, which we can talk about later, provably accelerate the solution to large groups of problems and probably accelerate means either exponentially accelerate. That’s where it gets really interesting, where fundamental principles of how society and commerce work change. And then there’s another group of problems where you get at least quadratic speedup. And that alone is, you know, huge. And we can talk about that. But essentially…
Silvija: I just think that, you know, this is where I think I would say that 99.99% of the population or probably a much larger percentage actually of 100, hasn’t come to the point where they’ve started considering this because they have more than enough just grappling with AI, and this will move that needle to a completely new level. So again, trying to just simplify it and I want to go back to Moore’s Law and chips, and I’m trying to find a way to understand this a little bit more tangibly. It’s really difficult to think about this whole Schrödinger cat problem. And is it alive? Is it that we don’t know until we open the box, etc.? If we think about chips, computer chips and Gordon Moore and say 1970 and he notices that these four generations of chips seem to be doubling in their efficiency per square centimeter or dollar used to make them every one and a half years. And then he prescribes this for another ten years. People laugh at him. He still sticks to his plan. He created Intel, I think.
Dr Mark: And I think it was initially he was from Fairchild but yeah.
Silvija: Yeah, Fairchild and and then people have been kind of expecting this Moore’s Law to eventually dry out and they keep saying, well, you can’t do much more. So what I want you to help me understand is the evolution of these chips. So I’m thinking, you know, we started with the big mainframes and then some years later, we figured out how to make these new kinds of chips that were very well suited to desktop computers. And then a Whole New Revolution follows on the back of that. Then we figure out how to make these new chips that are suitable for mobile phones, and a whole new revolution follows on the back of that. Then we figured out how to use these Nvidia chips that are very good for graphics, very parallel, also good for AI. A new revolution follows on top of that. And now with this, you seem to be moving the limits again. So this is a new revolution based on these qubits that you’re creating.
Dr Mark: Well, it’s not just moving the limits. It’s fundamentally changing what the word limit means. So, just to draw back the analogy with Gordon Moore, under the assumption that you’re building two dimensional structures, so on the surface of silicon. Then Gordon Moore noticed that the size of the what’s called the die, the die is the actual size you have a wafer of silicon and you make multiple dies. He realised that the size of the wafers and the size of the dies was increasing constantly. The other factor is what’s called a feature size. So the actual working part of a transistor is dependent on processes, which is dependent on things like lithography, precision, manufacturing, mechanical precision. He noticed that those feature sizes were also getting smaller linearly. If you combine those two effects, that leads to a doubling in the number of transistors per unit area. That’s Moore’s Law. When you project that forward from the 1960s, late 1960s, up to today, what has happened in the interim is that you very rapidly reach a point where the communication overhead between different parts of a chip dominates the performance. That led to particular architectures, computer architectures. So you mentioned, for example, the Nvidia chips. So GPUs graphics processing units, these are basically large pipelines, matrix linear algebra machines. They’re very efficient at performing linear algebra on matrices. The whole architecture is optimised for that. They’re not very good at doing other things. We now have neuromorphic chips where you simulate neurons on chips. There are other things you can optimise for in the physical architecture on a chip. This is all still two dimensional. There was some work done on making it three dimensional. And there’s potential there for some building layers. And the very latest processor chips and memory chips are kind of two and one half dimensional. Okay. Now, taking all that, you get these scaling factors. The fundamental limit, though, remains an atom. You can’t make a chip where the gate is smaller than an atom. Actually, you can’t make a chip where the gate is smaller than at least a handful of atoms. Because then, you know, what are you actually switching? What are you switching? And with the current state of the art in processor architectures at the future sizes, we’re talking about a few nanometers. Those transistors are statistical anyway. They have errors, they aren’t precise, and you have to compensate for that in the architecture. So there’s lots of error correcting memories, all this sort of stuff. Along comes quantum computing and quantum computing says, with a single qubit, I can represent not just one of two states, but I can represent all possible combinations of one and two states. As long as I don’t look at it. With two qubits, I can represent four states, and with three qubits eight. So the number of potential states you can represent at the same time is exponential. That means when you get to a couple of hundred qubits, you’ve got more potential states than there are atoms in the entire earth. It very rapidly gets to unimaginable numbers.
Silvija: So sorry, we need to backtrack just so this is a picture we can have in our mind. But just help me understand how big is the qubit and potentially can I have a computer with a couple of hundred qubits or is it more expensive than all the resources on this planet? Or how does it build?
Dr Mark: Well, can you have one right now? Difficult. Will we be able to get one in a few years? Certainly. The way you make those qubits is you have to create a quantum mechanically quiet environment, whatever that means. And as I mentioned before, there are two types. There’s two ways to implement qubits. One is with electron spin, and with electron spin, you need to create an environment with very low electromagnetic influence, which means it has to be cold. So no thermal energy. So you’re at Milli Kelvin or a few Milli Kelvin. You have to have a very high vacuum. So no molecules bouncing around, hitting your qubits, upsetting their value. And they have to have no magnetic fields. So you need to put it in a special kind of metal container. So those are electronic qubits. You can also use photons to make qubits, photonic systems. They have their own set of challenges because it’s difficult to get photons to interact with each other and one of the things you want qubits to do is to interact with each other and not with the environment. And then there’s a third one. There are a couple of other ways, but the third interesting way is to actually not use electromagnetic quantum mechanics, but to use nuclear quantum mechanics, because then you don’t really care about the environment. So you can run them at room temperature. You can put them in a diamond, for example. So in five years, will you be able to buy a quantum computer? Will there be a, I don’t know, a GP like a GPU card for your gaming PC with a quantum accelerator on it? Probably. That’s certainly something my company is aiming for in five years. What will it cost? No idea. Will it be affordable? Will it be usable? Almost certainly. So I see a future in five years. These things will be in PCs or potentially in PCs in ten, 15 years. Who knows? I mean. It could be very small.
Silvija: So we talked a little bit about the origins of this, and we’ll have to go back to that again. But I just want to go back to the word revolutionary, because people are now listening and they didn’t know they cared about quantum computing at all. And I’d like them to care. And I need you to help me formulate that once more. So you said revolutionary is almost too weak a phrase. Why?
Dr Mark: Because I think in comparison to the technological revolutions which have happened since the sixties, which have all been kind of directly or indirectly driven by the exponential rise in the power of computation. So the Internet, World Wide Web, e- commerce, the motor driving that has been Moore’s Law and the power of computation. If I take what Moore’s Law gives us and I give it a, you know, a rocket booster, in the sense of it’s a car driving around a Formula one racing track and suddenly I put wings on it and say, Oh, you can fly. That’s the kind of revolution it is. It opens up completely new areas of exploitation. So, in terms of what, and I don’t want to leap forward too much, in terms of what, what could you do? What might you be able to do? Solving problems which are completely unthinkable today, that they are so science fiction we can’t even imagine that we people… You would go what? Impossible. There are problems out there where we are pretty sure that a quantum computer will be able to answer those kinds of questions, in a way that will fundamentally change our economies, the way business works, the way we live. So, saying it’s a revolution is not really enough. It’s fundamentally disruptive.
Silvija: So. You know, two things I want to try to underline here. One is that in a world that’s already running on computing power, where everything is controlled by data and computing, getting this evolutionary leapfrog computing animal will be a game changer for those who know how to use it. And then the question is, will we be good at asking the right questions, giving it the right tasks? So we’ll get back to examples of where this can be useful in the next lecture. But in the meantime, Mark, can you tell us a little bit about the origins, you know, and we talked about how this thing came about from quantum physics. I’m still shaky on where we started making it practical? And then if you could relate it a little bit to the cloud and security problems.
Dr Mark: Sure. So I mentioned the early sixties and the seventies as a formative phase for quantum computation. So back in the sixties, a guy called Landauer proved that if you compute things, if you do a calculation on bits, it actually increases the entropy of the universe by a really, really, really small amount. But it does happen. And, that was kind of… There was a warning signal. Something interesting here. Is there a kind of computation which wouldn’t increase the entropy of the universe? And then the early seventies guy called Charlie Bennett proved that you could take one of your Turing programs, any Turing program, Turing function, and rewrite it so it’s reversible. In other words, you run the program and then you run it backwards and all you’re left with is the original tape and an answer. If you combine those two things, you’ve got a model of computation that can be run on a quantum mechanical system, because what it says is you do some stuff. And then you do it, reverse it, and then you look and all you see is the answer. And otherwise the universe doesn’t care. You’ve not created or destroyed any entropy. Everything’s as it was. You’ve just got the answer to a question. Then in 1983, to the exact date, there was a conference on the physics of computation, and among others, Richard Feynman, who the physicists will know his name immediately, the father of quantum electrodynamics, more or less, he said, we could actually use this to compute problems in physics, which are hard. Hard means forget it, we can’t do it. We can’t do anything useful. And in particular, the problem he identified, which would be efficiently computable using such qubits, such quantum mechanical computers, was simulating what’s called entanglement, which is where two quantum mechanical systems become entangled forshanked in German. And he showed that is a problem which is horrendous to compute on a classical computer, still is, but is actually quite easy to compute on a quantum computer. And with that, the gate was up and the horses are running to see, oh, wait a minute, what else could we do? And then we get into the area of cryptography and the leap that happened there, or the significant event that happened there, was again in the early nineties. I think Peter Shor, who was a professor at MIT, showed a way of mapping a classical algebra problem into a quantum mechanical formulation where it could be solved exponentially faster. And exponentially faster means the difference between billions or tens of billions of CPU years on the fastest computer we have today, to a few minutes or hours on a quantum computer with a couple of thousand of qubits. That’s, I mean, that’s a huge difference. And Shor’s algorithm, as it’s called, is one of the reasons that the National Institute of Science and Technology is reformulating all the key distribution protocols. They’re almost finished with that. And it’s why a lot of people want to talk about the impact of quantum computing on security. It’s Shaw’s algorithm.
Silvija: Anyways, we’ll get back to the topic of security in our next lecture. But in a way, if I’m trying to simplify again, what you’re saying is, and we haven’t even mentioned Claude Shannon or information theory in this series so maybe we have to set up the next one with that. Good… But..
Dr Mark: Don’t get me started on Claude Shannon. Yeah.
Silvija: Yeah. No, I’d like to get you started on Claude Shannon, but let’s let’s save that. So what you’re saying is that until now, we’ve been twisting our brains to formulate problems, find important problems for us humans, and then formulate them in a way that makes them practically computable.
Dr Mark: Exactly. Perfectly said.
Silvija: And we can redefine the way that we view our computing powers by actually just throwing all these real problems at the computer and being able to actually come up with the solutions.
Dr Mark: So there’s a theory called co-evolution, which you encounter in cultural anthropology. Cultural anthropology. If you look at technology and media anthropology. There is the idea that humans and human societies develop and evolve and change. And hand in hand with those changes are, of course, humans imperative to develop tools and improve tools and use tools for new things. We’re very good at that. And with every technological innovation comes a change in organisational innovation. And you’ve seen this over human history. It’s a very robust principle with classical computation. We’ve seen this. And you’re right. The problems we perhaps think of as well, the ultimate thing we can solve with a computer are in effect. If you look at it the other way round, they’re just the things that we can solve. Using today’s computers that moves on. It gets faster and faster and more and more of it. But essentially, we can’t, it’s very difficult to think about what would happen if I had this limitation, if I completely threw it out the window. Quantum computing is throwing the limitations, not all of them, but a lot of them, throwing those limitations out of the window. So in terms of what we can do with it, it’s a fundamental change in a way that the invention of the silicon chip and the transistor wasn’t. It goes much further.
Silvija: So final question here, Mark, before we end this first lecture. I was reading some bits from Guns, Germs and Steel the other day, and one of my kids asked me, “So, Mummy, if the Chinese invented gunpowder, how come they didn’t invent guns or how come they were not invented there?” And this cultural side of innovation is an amazingly fascinating thing related to what you talked about here. You know, we invent technologies that fit with our culture and without going too far into this determinism discussion. You know. Quantum computing is this immensely powerful game changer. And human nature. You know. What are your thoughts on that? Should it be somehow very carefully handled as a, you know, very dangerous medicine or…
Dr Mark: Well, yeah, you’re touching on all sorts of sleeping dogs and skeletons. They’re potential skeletons. So it’s not even potentially, it is a strategic technology, that’s for sure. If you look at where quantum physics happens? Where did most of the physicists live and work? A significant proportion of them came from the German speaking world Germany, Austria, Switzerland. Those three countries are dotted with people who formulated quantum physics. And so there was a heritage in Central Europe. And there is still if you look at the university, the academic environment around quantum physics, quantum mechanics, quantum computation, quantum chemistry, I mean, Germany is you know, it’s just incredible the density of research happening in Germany. Historically, we’ve always seen that the Anglo Saxon world, in particular, the Anglo Saxon world to the west of us across the Atlantic Ocean, is very good at exploiting technologies there. Certainly they’re certainly putting their foot on the gas pedal for quantum technologies. Is there an opportunity for Europe to lead? Yes, definitely. Is it too late? No, but we’ve got a lot of catching up to do because the US is very much driving forward. And you mentioned gunpowder in China. China is also investing multiple billions in quantum technologies in general and quantum computing specifically. And it will be a society forming technology. I don’t think there’s any doubt about that. So the short answer is it’s a game changer both at the economic, business and societal level. You didn’t ask this, but it was implicit. How is Europe positioned? Definitely could do better. But… There is a lot of potential to overtake the other two because the game is by no means over yet. And that’s one of the reasons my company has moved to Germany as we want to be… We see Germany as key. But other European countries, almost all the other European countries, have a definite value to add in that equation.
Silvija: And I think we have to stop talking about this as blue sky research and move it down to something that’s in our hands. Definitely. Thank you so much for this first lecture. We’ll meet again in a couple of minutes for the next one.
Du har nå lyttet til en podkast fra Lørn dot Tech. En læringsdugnad om teknologi og samfunn. Nå kan du også få en læringsserifikat for å lytte til denne podkasten på vårt online universitet Lørn dot tech.
Velkommen til Lørn dot Tech. En læringsdugnad om teknologi og samfunn med Silvija Seres og venner.
Silvija Seres: So welcome back to our Masterclass discussion with Dr. Mark Mattingly. Scott, who is trying to teach us through a friendly conversation about quantum computing. This second conversation is going to be about some of his favorite examples in the area. And again, I’m not quite sure how concrete one can be with technology this new, Mark. So what examples can we talk about?
Dr. Mark Mattingley-Scott: We can talk about three kinds of fundamental types of problems where quantum computers are going to change how we compute things. And the first group of examples are really things that came out of or were derived from or resulted from Richard Feynman’s demonstration that quantum computers or qubits would accelerate the simulation of entanglement in the early eighties that basically led to the insight that simulating material, simulating atoms and molecules and things was going to be accelerated, all using quantum computers. And essentially what you do is you take any kind of molecule or thing/ material and you map the quantum mechanical description of that system to qubits in a reasonably scalable way at the atom level. You can map energy, energy levels, essentially energy levels to qubits and then through an iterative process, solve those energy levels. Once you’ve solved those energy levels you can look at, you can use that to infer information about, for example, what a molecule looks like in three dimensions or what its binding energy is, how efficiently it will interact with or react with other molecules. What that means in practice is for things like chemical production, in the area of, well, basically everything but… Things like energy, energy chemistry, batteries, efficient solar cells, those kinds of things. That’s all about getting very, very efficient materials and using quantum computers to do that will speed those things, speed those development processes up. Also in the area of pharmaceuticals. So one of the things you’re concerned about when you’re doing pharmaceutical research, the first step is, have I got an active substance which does something and can I show it doing something in mice or rats or whatever? And then the challenge becomes, can I find a way to deliver that in a way in humans which doesn’t kill them? And then can I develop that so that its dosage and means of administration are tenable, accelerating that process because you, for example, can simulate what a molecule does will have enormous impact. If you look at the costs of developing a pharmaceutical single drug, it’s in the billions. If you can accelerate that, it will have an economic impact, of course, for the pharmaceutical companies and for patients. But it will also mean we get a step nearer to bespoke and individualized therapy regimes. So we may see some significant speed up in those processes in the mid term. So that’s materials, pharmaceuticals, battery design, materials design, surface design. The next area really comes out of that.
Silvija: Sorry. Sorry, Mark.
Dr. Mark: Yes.
Silvija: You mentioned pharma usage. You mentioned other kinds of construction materials or, you mentioned, batteries. And it led me to think about energy solutions. So Norway is fundamentally an energy and natural resource through advanced technology. This could be applicable also to the new energy solutions for say, whether it’s hydrogen or.. What we are really struggling with when we build for example, offshore wind farms is creating them robust but flexible enough and experimenting with the right kinds of materials and the chemistry of it might get a real push through something like Quantum.
Dr. Mark: Could well be, actually, if you look at that entire energy creation, distribution, storage and consumption, that entire value chain. The materials part of it does come in. It’s an important factor. So, for example, if you’re using wind energy to directly generate hydrogen, it’s kind of a no brainer. If you’re putting wind generators in the middle of the North Sea and using them to generate hydrogen. It’s fairly obvious there’s water there. And why not? Is that efficient? I don’t know at the moment, probably not. But could it be efficient? Is there a lot of chemistry involved in making it efficient? Yes. Are there more efficient ways to store hydrogen than just as a gas or as a liquefied gas? Maybe, there is work going on on that. Can quantum computing be used to accelerate that and prove that? It’s probably I wouldn’t know, you know, without knowing more about the details of the problem, is that going to be relevant in what time scale? But those are the kind of problems where quantum computing is at the forefront, where it’s about things behind that energy generation problem you talked about. It is a second class of issues, if you like, which is around logistics optimization. What’s the most efficient way to do things, to combine things? And that comes back to the second class of problems where we look at… Basically derived from the work that Peter Shaw did with his factorization algorithm. So coming back to Peter Shaw, what he did was he took a problem which is computationally indeterminate factoring in number and translated it into another form. So if you take two numbers and you multiply them, it’s very fast and efficient on a classical computer. If you take a number and try and determine what the factors are, it’s horribly inefficient. It doesn’t make a difference if the number you’re talking about is 15 because everybody knows that 15 is five times three. If the number you’re talking about is 1000 bits long, it’s computable. It would take billions of years and a half hour size computer, Peter Shaw said, hmm, finding those factors is kind of like finding the fundamental frequencies in a musical chord. So if I sit at a piano and I play a chord then that chord actually consists of distinct notes. Assume the code consists of two notes. If there’s a way to translate the factorization problem into determining what notes those are. Then this is what Peter Shaw did. Essentially. He said, I can translate that factoring problem into something in the quantum domain and then, this is where your brain is going to get fried here, and I then use something called the Quantum Fourier transform to identify what the factors are, and then I can translate it back and out pop the two factors. And it’s incredibly efficient. That led, that work of Peter Shor, led to two things. First of all, everybody suddenly got concerned about cryptography, rightfully so. And the second second thing is people started to look at are there other areas where we can use those principles to accelerate what are called classes of linear algebra problems? So factoring a number is a simple linear algebra problem. Inverting a matrix is another linear algebra problem. Can we use a quantum computer or can we use qubits to invert a matrix fast, rapidly. Turns out you can. And also, I think it was in the late nineties, three researchers whose names I always forget, but their initials are H, H and L, and the algorithm is called the HHL algorithm proved or showed a way to exponentially speed up matrix inversion. And since then, we’ve had a whole class of algorithm research working on using qubits to solve those kinds of problems. There is a but with these problems. To do that, you need error corrected qubits. You need qubits which behave perfectly. We don’t have them yet. And this leads to the third class of problems, which are essentially algorithms which don’t in general provide this exponential speed up. They provide quadratic speed up and they use the real qubits of today. There are those and others. You find those algorithms around things like optimization, Monte Carlo analysis. So where are you looking? Lots of different alternatives to a problem. Can you speed that up? Can you take your Turing non deterministic Turing machine and speed it up a little bit? The answer seems to be yes, you can speed it up. Quadratic. Quadratic means squared for the listeners squared doesn’t sound like a lot, but if you’re looking at big enough problems, it’s the difference between doable and impossible. So those are the three classes of essentially the three classes of problems. And if you look at the first one.
Silvija: Sorry if I just go back, there was the first class just summarizing. We need to stop a little bit and give people time to breathe. So the first one was materials technology simulating. Real things in an incredibly, much faster way. The second one was computational problems like cybersecurity. Cryptography, things that require things like inverting a matrix and factorization.
Dr. Mark: I also think that that class of problems also includes a lot of machine learning and AI because that’s about inverting matrices.
Silvija: So that’s what I wanted to get back with you on. So Machine Learning A.I., you mentioned Monte Carlo simulations, for example.
Dr. Mark: They were the third category.
Silvija: So a third category.
Dr. Mark: Things you can do with real qubits, real qubits are subject to errors there. They lose their state after a certain amount of time.
Silvija: And Monte Carlo simulation is important. Where do they pop up? What sort of real world problems do we see those in just for our listeners?
Dr. Mark: Well, the classic example is from the finance industry. Monte Carlo simulation is used to do things like options, pricing arbitrage, currency arbitrage, hedging, risk analysis in the finance industry. What you essentially do is you take your problem and throw random decisions at it and look at what comes out. And then you repeat and rinse, repeat and rinse, repeat and rinse a number of times. And Monte Carlo Simulation says that given enough repetitions, you will get a statistically robust answer. What the quantum computer does is it accelerates getting to a statistically robust answer, quadratically. And that can be for if you’re managing a hedge fund, that can be a huge difference in time.
Silvija: So given the fact these real qubits already exist.
Dr. Mark: Yes.
Silvija: And I don’t know if you can buy them, but they exist. So whoever gets to use them efficiently, given the potential of earning money or whether it’s hedge funds or controlling markets or doing anything where type game semantics comes into play. I can imagine the income. Differential the income potential from somebody being able to do this is just so huge.
Dr. Mark: Yeah, in the finance industry, typically these are a performance advantage either in time or accuracy of per 1000 is enough to dominate a market. So if you can shave a fraction of a percent off the time to execute or the accuracy of the result, then you’ve just eradicated your competition. So the, you know, the sensitivity is very, very high, just really, really small improvements. And using a quantum computer to achieve those kinds of even a small improvement is of such immense potential value that the willingness to invest in the technology, to understand their technology, to start getting ready to use it is very, very high. So most banks, insurance companies, most financial institutions, the central banks, they’re all very, very interested in quantum computing for that reason.
Silvija: But also because in the hands of rogue people or your enemies, this could be a system disruptive thing.
Dr. Mark: Yeah, but I don’t think that’s specific to quantum computing. That’s the same with every technology. Every technology that can be used for benign purposes can be misused. Just look at human history. It’s an absolute you know, it’s a kind of constant and human, human nature. Then the question then becomes… It’s not optional. You can’t ignore technology. For two reasons. First of all, it will give you an advantage. And secondly, if you don’t take it seriously, you will be at a disadvantage. Disadvantages that may be just commercially or it may be socially. It may be so disruptive that the potential is there to change so much you better be in front of that technology, but understand how it works, what the implications are. You’d better be ready for it.
Silvija: What you’re saying. I’m trying to simplify. I’m using my Lego language now. So you’re saying that you have a choice to say this is too complex for me and I’m waiting. If you’re running a central bank or are in charge of some very systemic cybersecurity stuff in defense or… But then what you’re risking is that whoever moves first will define how this will play out.
Dr. Mark: Yes. As I say, that’s not specific to quantum computing. That’s every technology that’s come along, every technology from gunpowder to going back to, you know, the gunpowder example or the invention of the Longbow and the British archers Agincourt. Apologies to the French viewers. Every technological advance, the potential is there. And if you ignore it, then at your peril. And the central banks, as I say, are by and large, they’re all very interested in quantum computing, which is actually quite encouraging. They’re taking it seriously. The banks certainly are. And, I think if you look at the less salubrious countries on this planet, they’re also looking at quantum computing and how to use it and almost certainly how to misuse it. So it’s just a technology like every other. It cannot be ignored.
Silvija: Yeah. I want to ask you about two possible areas in addition to the ones that you’ve kind of given us three types of problems. But it’s still abstract to many listeners who are not computer people. But I heard you speak about applications in robotics and autonomous cars and satellites and what you call disconnected things. And then I also want to ask you about possible applications in computer centers.
Dr. Mark: Okay. So where quantum computing is coming from right now is from a legacy of computer centers. So the electromagnetic qubits based on electron spin and photonics, they all need to run at very low temperatures, which means you need cryogenics, very low temperatures, which in general means you’ve got to have a kind of like an inverted steam engine technology. You need to be in a computer center. You need infrastructure. And typically that also means that you’re then accessing that machine via some kind of cloud service. And most of the players on the quantum scene are providing quantum computers in qubits through that model.
Silvija: So, so sorry. So this is too many concepts and too complex. So when you say cryogenics, my first association is Peter Thiel and where he’s going to go when he dies, basically being frozen. But this is not what we’re talking about. We’re talking about a very, very cold environment where you can have your computing center and that’s where these qubits work in a relatively non disturbed fashion. Or do I have the wrong image?
Dr. Mark: Kind of, you’re getting there. So actually the physical location of the computer center is not important. What is important is that within that computer center, you have a computer, quantum computer, and within that quantum computer, there is a whole bunch of electronics and interface and IO and networking, which is standard stuff. But there is also what’s called a cryosat. But typically this is, I don’t know, a few tens of centimeters or a meter across and maybe a meter or a meter and a half high. And inside that cryosat, the cryosat is normally made of a special kind of metal, which basically eliminates magnetic fields inside the cryosat. It’s under an ultra high vacuum, so within a volume of maybe one and a half, two cubic meters, you probably only got a handful of nitrogen and oxygen molecules bouncing around. And it’s at a very, very low temperature. It’s around, it depends on the technology, from a few milli kelvin to less than a kelvin. Kelvin is the measure of absolute temperature, so one Kelvin is one degree above absolute zero. In order to get that environment and especially in order to get the temperature down there, you need some advanced cooling using mixtures of different helium isotopes in a diffusion dilution refrigerator. I won’t bore you with the details, but it’s like your fridge at home. But it’s using two different kinds of helium to basically get the temperature. It takes a few hours to get the temperature at the business end of the quantum computer down to a few million Kelvin. That obviously works in a computer center.
Silvija: I need the picture in my head. So there is this super cooling thing unit, your cryosat and it is surrounding this thing that has all the qubits and that’s the computation, or it’s inside?
Dr. Mark: It’s surrounding it.
Silvija: Right? So it’s in a fridge, in a super…
Dr. Mark: In a super cold fridge. A super cold fridge. In order to interact with the qubits, you need some electronics and that’s a challenge in itself is the electronics need to work at the same temperatures. It’s not trivial, but essentially it’s more or less doable. So you have these super, super cool superconducting qubits. You have some electronics to interface with them and then those signals go outside the cryostat. And then you’ve got sort of traditional electronics around it which allow you to interact with the qubits and do things with them. And then normally what you do with that is those electronics and that interface, you then embed that in a cloud service so people can access the quantum computer via the cloud. And that’s what Honeywell and IBM and Alpine Quantum and almost all quantum computer manufacturers are doing if you are using qubits based on nuclear spin. So you’re able to run them at room temperature, particularly if you’re using nitrogen and nitrogen vacancies to do that. As quantum brilliance is, then your qubits work at room temperature. The electronics become significant, the control electronics become significantly simpler. And that’s a technology where a quantum computer fits at the moment into a 19 inch rack unit. So a 19 inch rack is a standard format for computer centers like this wide. So it’s about this high. And the qubits are in there and control electronics. And you just slide it in and you can connect it up to whatever, the cloud, if you want, or a local machine or. Something. So those room temperature quantum computers are ,I believe, the future of quantum computing, because if you remove that need for superconductivity, for cryogenic systems, and additionally have simple electronics to control it, then it’s all miniaturized, which means you can start to really focus on making things small, portable, and potentially start to put them in places which you would never, ever be able to do with a cryogenic quantum computer. Unless you ensure that it’s always connected and always has access to the cloud, and there are lots of things where that’s not possible.
Silvija: So again, trying to kind of find the bits of my brain that aren’t fried yet. So we talked about this very, very strong fridge freezer, and that’s the cryogenic based quantum, which works really well. But it’s kind of not so practical because creating this fridge has its size limits. The way I think I answered you. But you say that it’s possible to use this other kind of quantum thinking – something nitrogen, something which… I’ve forgotten what you said about nitrogen vacancies, and I won’t go there. And that means that you can make a computer that both fits in the computer centers of today with these racks, but it’s possible to miniaturize even further eventually, which makes it extremely useful for all kinds of built in. And that’s where we are getting to things like cars and robots and the Internet of Things..
Dr. Mark: And aerospace… So if you don’t have any connectivity. If you’re literally disconnected from any kind of network or if you’re in an environment where connectivity is a challenge. So things like mining your underground or heavy machinery where there’s a lot of electromagnetic interference or a steel plant where, you know, any kind of any kind of heavy industry, anything that’s robotic, disconnected, anything that’s in the air or in space or deep underground or under the sea, shipping, all these things where you have limited or no connectivity. You’d like to have a quantum computing technology, a quantum computer which can still work, still be used, and where you’re not concerned about providing a quantum mechanically silent environment, which typically almost always has to be done by employing cryogenics. So is cryogenics miniaturized? Maybe. Maybe. Is it easy to produce a few milli Kelvin? No. You’re always going to need helium. Helium three. Helium four. Diffusion pumps. You’re always going to need ultra high vacuum. Can you put that in your phone, in your 3-D goggles? No. Could you put a diamond in the 3-D goggles? Absolutely, not today but in the timescales we’re talking about that’s realistic. But the key question here, Silvija, is what can you do then that is useful? With that quantum computer in that disconnected device. And then we come back to these three classes of problems, material simulation, anything to do with linear algebra and things where you can use variational methods.
Silvija Seres: Sorry, I have to talk in pictures. So let’s say you make this into not a 19 inch thing, but into a small pill that I could or chip that I could build into my body. So it could be my personal little laboratory that could do all kinds of simulations and analysis for me. Or I could build it into my car, which would make it many times as fast a learner and many times as safe. Or, you know, where would we need this immense amount of computation in things that are being built in? I can see a rocket on Mars, but…
Dr. Mark: So we’re in the realms of science fiction now if it’s in a pill and it’s in your body, no idea if that’s going to be possible. And if then when. I think it’s more realistic to look at where you can use this acceleration potential in devices which are not permanently connected and typically in situations where you have the requirement for or you would have the requirement for very large computations if you had the possibility. So, yes, it’s things in robotics like anything to do with sensing their environment. The way that’s done today with Graf, with GPUs, with special environmental sensors, with LIDAR, the chips doing that. Typically GPUs typically consume quite a lot of power and do a good job. But, to train them, to train the models for those kinds of applications still takes a very long time. Might it be possible to drastically, drastically accelerate learning and machine learning using quantum computers? Maybe you hear me saying maybe and using the subjunctive a lot because we don’t know yet. That’s why one of the key things is to start experimenting with and playing with these technologies in order to understand exactly where the benefit is, both in terms of performance and in terms of where’s the first economic impact, where can we expect the first situations where quantum computers actually show utility? So that’s why everybody in the quantum computing business is talking to customers to understand where the benefit will come from.
Silvija: And I guess the customers get their eyes crossed out or cross, you know, when you start talking to them. But, you know, we’re going to move on to our tools lecture in a few minutes. But I said the last question already. But still, I have to ask you one more thing, Mark. And you were there from the start of practical Internet, and you’ve seen how and I remember those days as well. I was at Alta Vista at some point around then, and I remember that, you know, people were looking at this thing and saying, we see it’s going to be hugely potent and disruptive. It’s just we don’t know how to make money on it or what to do with it. And it sounds to me almost like you’re in a similar situation now with Quantum. You know, we can see that it’s going to be disruptive like anything ever seen. It’s just we don’t know what to do with it.
Dr. Mark: Yeah. So there are definitely parallels and it took from 1989 with HTML and http. And I think back then it was pretty clear this is hugely disruptive. What does that disruption look like? No idea. And it was in 1996 that the first kind of milestone in how it was going to disrupt happened. So we had the very first electronic commerce platforms ever, so end to end electronic commerce. And then it became a lot clearer about where’s this particular rabbit going to run? And a lot of what’s happened since then, also the urge of monetization of search machines, monetization of social networking, where we are, the product, we just don’t realize it. All those things were then at the point where we had e-commerce foreseeable more or less. Whereas what’s the parallel with quantum computing? We’re beginning to see where that might come, where there might be first impact. Again, it’s much more embedded in the guts of reality. So material simulation, basically anything that can be optimized anywhere. Logistics. Shopping. Transportation. Anything of that nature. If you’re in that business, you need to get your eyes on quantum computing right now because it may well fundamentally change your business and you need to be aware of that. So we’re at that kind of inflection point where it’s not yet totally clear, but it’s clear which industries will be impacted and who needs to look. Keep an eye on it.
Silvija: Mark, we’re going to finish our second lecture now and in a couple of minutes, meet for the third one where we’re going to talk about, well, how do we get started then? Okay so far.
Dr. Mark: Thank you.
Du har nå lyttet til en podkast fra Lørn dot Tech. En læringsdugnad om teknologi og samfunn. Nå kan du også få en læringsserifikat for å lytte til denne podkasten på vårt online universitet Lørn dot tech..
Velkommen til Lørn dot Tech. En læringsdugnad om teknologi og samfunn med Silvija Seres og venner.
Silvija Seres: Hey and welcome to the third lecture on quantum computing in Lørn Masterclass with Dr. Mark Mattingly-Scott. So this is the lecture where we try to make it applied. So let’s say I’m a manager, I’m a leader, I’m a board member. I’m an investor in one of the industries you’ve mentioned. And I decide, okay, I want to figure out what to do with this quantum computing. The only thing I’ve figured out so far is that I should quickly figure out how to invest in quantum brilliance. But I missed out on Google at the IPO and I remember sitting in a corridor and laughing, saying that $150 a share, you know, it’s just madness. And I wish I knew then what I know now. So perhaps I can avoid it with Quantum. But you mentioned the three areas of applications, and I’ll try to summarize it in my kind of simple language. So the first one was this simulation application in the real world. So you said anything material, chemicals, pharma, oil, energy, life, science, agriculture where you need to figure out how things would develop if you push them this way or that way, you could have a great advantage using Quantum. Then the second group was you call them linear algebra problems and these are solvable based on these ideal qubits which are still kind of difficult to make. And this is the machine learning A.I. sort of group of problems which are in every industry, really. And then the third one was this area of real qubits, which I don’t even remember exactly the difference between the two. But these have a bigger tolerance for failure or statistical imperfections. If I understood you correctly, and this is where finance is already scratching their head on how to gain an advantage on the others. Have I summarized it more or less?
Dr. Mark Mattingley-Scott: More or less? Yeah. Pretty good. You don’t need my help.
Silvija: Yeah, well, you can’t imagine how shaky a foundation this is based on. Okay, so if Silvija is going to bluff one step further on quantum computing, how does she get started? Where do I go and start scratching some more?
Dr. Mark: So I guess there are two or three things you need to do. First of all, you need to read because you need to lay the foundation of learning what qubits are and how to actually use them. There is actually a large but still finite number of scientific papers and books on quantum computing out there. I’ve totally lost track, and I was keeping track of every book on quantum computing and quantum mechanics. My bookshelves, you can see them behind me. This isn’t the quantum mechanical section that’s over there, but literally hundreds of textbooks on and around this subject.
Silvija: With all due respect dear Mark you’re a nerd, and I’m not sure how long we normals would last in one of those hundreds of books. I mean, there is still not a Netflix documentary, I guess, on quantum computing. But you have a TEDx.
Dr. Mark: Wait, wait, wait. I have a TEDx talk where I attempt to explain how qubits and quantum computers work. And for viewers and listeners, if you just Google my name and Ted Stuttgart, you should pick that tech talk up where I use an analogy between a spinning ball and a spinning coin to explain how qubits work there. There is a lot of information out there. In addition to my talk, literally hundreds of YouTube videos explaining from the very basics to very advanced hardware related concepts. IBM is at the forefront in that educational push. But there are others as well. I mean, it’s then more a case of how do you curate and recommend the best videos and the best source of information. It also depends on your language. So in German there are some really good German language explanations of how it all works. There are introductory papers, white papers, blogging, blogging activities. One person who’s really, really good on the blogging side is Scott Aaronson. And for you, Silvija, as somebody who understands computational complexity, Scott Aronson blog would be something you should definitely take a look at.
Dr. Mark: So I think in terms of raw information. Where do you position yourself on that wave of information collapsing over you in terms of YouTube and other sources? There’s a lot of information out there. If you actually then want to take the next step and start to use qubits at the moment as a just generally an interested person, the only effective way to do that is through IBM, through their quantum experience, cloud offering. That may change or that’s almost certain to change in the future, because I can’t imagine that IBM’s competitors will sit around and twiddle their thumbs. But that’s also, you know, that’s a way to actually use real quantum computers. There are also a number of quantum computer simulators out there, which are also accessible either via the cloud, or you can download them from GitHub and at least for a couple of qubits, run them on your PC. So there are a lot of opportunities to actually interact with real and simulated quantum computers in order to get to the point.
Silvija: How do we do that? So we need to. Is there a programming language? Is there. Sorry.
Dr. Mark: So yes, typically you’ll be accessing quantum computers today via a cloud service. And a cloud service is the technical term to a service endpoint. What that means is there is a TCP IP address, so an Internet address with a port. And as long as you send an HTTP request to that port with your API call encoded in the right way, there’s a little a server on that port and it will listen and say, Oh, okay, I’ve got to do something now and it’ll decode what you’re sending it and then it’ll do that, wait for the result and send it back to you. So a service invocation by the cloud and there’s all sorts of APIs to enable you to to use that using particular programming languages, typically Python but also other languages, Q Sharp from Microsoft, Rocket from Amazon. There are many different languages. And essentially the language is determined by the developers and the people you want to have using the quantum computer. For the quantum computer itself, it doesn’t care what language you’re using to program it. It’s going to be doing operations at the hardware level, much the same as if you’re using a computer. You know, there’s a processor here in my Apple Watch. It’s programmed using I forget the name. It’s a variation of C which Apple developed. My laptop is also working, my other laptops running windows. They’re all using different primary programming environments. The hardware doesn’t actually care. And it’s the same with quantum computers. The hardware doesn’t actually care. So you’ve got your qubits, you’re accessing them. You’ve learned about them.
Dr. Mark: You’ve watched some YouTube videos. You’ve read, hopefully read one or two books, and you’ve probably reached the point where your brains are fried. Sooner or later, everybody gets there. And then you’ve also found a way to actually start playing with a real quantum computer, and now you’re faced with a challenge. And the challenge is, how do I get my head around how to usefully program a quantum computer? And all I can do here is relate a personal anecdote which may help people. So this is me personally, something I did very, very, very many years ago. I did a PhD in information theory in 1985. And I developed a method to generate what are called spread spectrum code sets. You don’t need to know what those are. They are nowadays used in 5G and I developed an algorithm. And then about two years ago, when I really started to get my teeth into the programming side of quantum computing, I went through an exercise of trying to rewrite my PhD algorithm, a PhD thesis using a quantum computer. And by doing that, I learned a lot about programming quantum algorithms and what the difference is. And I think one of the core takeaways from me was that and this comes back to what we talked about in the last session, the way we solve problems today is in a very strong sense, determined by the limitations of technology we’re using to solve those problems. So the way we write computer programs is designed to be efficient, to run on a computer when you move to a quantum computer.
Dr. Mark: Some things become much, much more efficient and other things become less efficient. So you basically need to throw out all the assumptions about what order things happen out of the window. And that was a key insight for me, was that actually in the classical algorithm, I did things in the order A, B, C, because that was the most efficient way to do it. When I translated my algorithm into the quantum realm. I actually ended up with CBA because that was more efficient. If you look at Peter Shor and his factoring algorithm, he went through that same process. He determined that actually there was a more efficient way to do things, the more efficient order to do things which got you to the result faster. So in terms of how do you move from just being able to use a quantum computer, to understanding what a qubit is actually. Creating your own unique thing with a quantum computer. There is a big step in that. My own personal life. The history and challenge. With that, the probably the most useful thing I did was read the history of and details of how people like Peter Shor, Lev Grover, all the other people who developed some of the fundamental algorithms, we now know about how they actually did that, what their thought processes were. And Peter Shor… I think there’s a video up on YouTube where he explains exactly how he got to his algorithm. That has to be compulsory viewing for everybody. If you want to get to the point where you’re actually using or actually able to use quantum computers to write new algorithms.
Silvija: I think it’s incredibly interesting and the bigger point than just quantum computing. And it is how, you know, when you have a hammer, everything looks like a nail. And so we have the classical computers today and we are solving, as you say, all problems around us with the understood restrictions and constraints of those hammers that we have. And it is a very difficult retraining of your mind to understand the new kind of a shotgun that you have suddenly and what it will do and I just think the people that understand the real advantage that this will give you. It’s a bigger strategic challenge than a technical challenge is, I guess what I’m trying to say.
Dr. Mark: No, I think the people who understand that fundamental principle that problem solving today is constrained by the exact details of the technology that’s available today and not the other way around. Which is a mistake a lot of people make. We think, oh, we can solve problems. So let’s just take the latest, greatest computer and we can solve them. If you get to the pointy end of business or the pointy end of science and technology, you very quickly realize that actually it’s the other way round. We’re constrained by technology and not enabled by it. If you then take it, bring our technology into play, which is fundamentally different and fundamentally capable of dramatically accelerating certain things, then the real challenge and the people who are really going to benefit both in a societal and business context are the ones who suspend that, who are conscious of that the way things work now and say, okay, I need to reframe in the sense of what are the new limitations? Not what are the new capabilities, but what are the new limitations? How can I reframe? How can I express problems? How can I implement solutions using this new technology and therefore achieve drastic speed up in certain classes of problems? And part of those are in terms of the fundamental technology itself. And part of that is in terms of how do I access it, which is why I come back to: What happens if you have a quantum computer which is autonomous or you don’t need connectivity, what does that mean? What kind of problems will you be able to solve? To be honest, I have absolutely no idea. I can’t tell you today I have an inkling of some problems, but what’s the full scope going to be? No idea. Ask me in 1989 to 1996. That’s seven years. Ask me in seven years, that’s when we will be. That’s the difference between HTML and E-commerce. With seven years it’ll be, I think with Quantum it’s going to be a bit quicker.
Silvija: Noted 2028.
Dr. Mark: Yes.
Silvija: I have to digress. Talking with you is so fun. And we always end up, you know, in a couple of universes away. And by the way, one of the images in my head when you started explaining this whole ideal cubit, etc., was Douglas Adams Hitchhiker’s Guide to the Galaxy and the great answer of 42. But always the question.
Dr. Mark: Yes, exactly. Exactly.
Silvija: So I want to get a little bit philosophical here. It is a matter of innovation. So just a couple of days ago, I was in a board meeting where somebody was actually exasperated with this obsession with innovation. And, you know, they’re saying we are paying too much heat and too much respect to innovation. And to me, that sentence contrasted with what you are telling me today is just, you know, I can’t even parse it. It’s like if we don’t care about innovation and we close our eyes to the opportunities you’re talking about, then we are just in the sunset mode. And then I want to ask you, you know, in current times, I see this obsession with us putting in young people, new heads, as guarantists of innovation in many jobs. So now, by my calculations, you’re about 60. I’m about 50. And both of us are obsessed with understanding something. So you’re 61.
Dr. Mark: No, I was born in 61. I’m rapidly… The date… 13th of September. I am rapidly approaching six zero. It’s not many weeks away.
Silvija: We’ll celebrate. So I’m just thinking. First of all, youth is no guarantist of innovation. Age is no guarantist of experience. So how do we make sure that people, independent of their age have enough combination of experience, curiosity, and and just this hunger to figure out what’s next to actually be able to use something as disruptive as this. What are your thoughts on innovation and maybe society?
Dr. Mark: So the analogy I’m going to use is with biology, and in biology stagnation is death. Biological systems which don’t adapt, which don’t change, die off. Because the environment, the universe is always changing and requirements and the environments are always demanding. Adoption and innovation in a technological sense for me, is just a particular way of expressing that same principle. Technology has changed. The capabilities of technologies change. Ignoring those is just not an option. You will stagnate, you will regress, and you will become more or less. I wouldn’t say superfluous. That’s too strong a word. Or maybe it’s not, but your ability to embrace, capture and cope with the resulting change in society, in the business world and the economy will then be severely limited. And if you’re not in front of innovation, then you’re probably going to be consigned to the history books. That’s the risk. I can understand sometimes, sort of, why people are tired of innovation because it is exhausting. But that’s in a sense. And if I look at my own life, starting a family, having kids, having a career, juggling all the balls, and all that involves. It’s not easy.
Dr. Mark: You have to get up every morning and dust your shoulders off and say, right, ready to fight? That’s in a technological sense, in a societal sense, acknowledging that imperative. I think it’s very important. What I’m not saying is the survival of the strongest and the fittest and everyone else is irrelevant. Absolutely not. Because I think on a personal level, one of the key measures of a civilized society is its ability and commitment to embracing the needs of those who are disadvantaged or are unable to cope for whatever reason. The reason is irrelevant. So it’s how do we help and support the disadvantaged in whatever form that’s a measure of a civilized society. So I think the challenge then is how do you do both, how do you embrace innovation and how do you stay civilized while doing that. That’s something that, and I know you’re going to smile at this, I think that’s something that in Europe, we can, with a few exceptions, demonstrably do, which maybe other societies, other places on the planet are not quite as good at doing. We’re very good at that.
Silvija: I have to throw in two books here. So we always end up talking about books. And I’m listening at the moment to a book by Fareed Zakaria. Ten Lessons for a Post-Pandemic World. And he’s brilliant. It’s really, really, really nice. And he has this observation about lots of you know, I’m never promised culturally to myself, I’m never going to become a pure vegetarian. But the way that he, for example, argues for vegetarianism almost made me so. But he talks about it. Modern humans are actually a man made invention. And I think the difference that I see to your biological argument, where you have time to adopt, you have generations to adopt, and it’s the speed of change of evolution and nature that that fits with this kind of adaptive Darwinism. It is on exponential timing now in modern human culture. And it’s how do we get there fast enough without overheating? Absolutely everything. And basically getting the whole world out of the spin. And before this book, I was reading a book by Scott Galloway. It’s called The Algebra of Happiness, and I was fascinated by it.
Dr. Mark: I have it! It’s over there.
Silvija: So I love him. I love his writing. And yet I’m thinking he’s a perfect adaptation to society that I think is wrong. It is so competitive and so out of what you’re just saying now where you know, yes, the best ones can win and you have to compete to be those few best ones. But, you know, we can’t have a society where everybody is best and we need to stabilize it somehow. And so I’m getting back to your last argument now.
Dr. Mark: So, just as an aside and maybe an interesting source of information, the idea of secular cycles and if I won’t distract by going and looking, but I have a pile of books here and the book is in there somewhere, the idea of long term historical processes which are invariant and have shown to be been shown to be invariant across human history. Secular cycles, it’s called, is also very relevant in this context.
Silvija: What does it say?
Dr. Mark: It says, well, don’t quote me because. Approximately what it says is that there are supra generational processes which human societies go through super generational means over many tens of years.
Silvija: And this is not a different cycle?
Dr. Mark: These cycles typically take, I think, around 200 years. They can be pretty clearly demonstrated going back at least a millennia. And they basically revolve around the idea of elites and surplus. And what’s the opposite of surplus. I’ve forgotten. Deficits! Deficits and surpluses and the establishment of the elite. So an elite group is the establishment of an elite and then the collapse of an elite. And this particular cycle. So, you know, of population growth. These cycles repeat themselves throughout human history. And there’s a very, very interesting guy who wrote this and I’ve forgotten his name. Apologies to him, he’s written a book about modern America. And it’s really, really, actually quite frightening that America is on the cusp of going down. So the rise and fall of the British Empire, the rise and fall of the Roman Empire, these are all examples of secular cycles. They’re not the only ones. So that’s one point. I think Darwinian survival of the fittest taken to extreme is, of course, very antisocial. You come from and live in a society which is quite social.
Dr. Mark: I’ve come from a country which was somehow in between, and I live in a country which is also quite social, where the idea of not letting the underprivileged or the disadvantaged or those who have absolutely no reason are unable to fend for themselves, for cope to cope with for themselves. You do not let those people drop into a horrible existence or you at least try to to avoid that. I think that’s… There is a phrase that my father used to say to me. He’d say, Always, always remember when you see somebody on the street who’s begging or you see somebody who’s got nothing. There’s a phrase in English called There, but for the grace of God, go I. And I always have that in my mind that it’s taking care of people who are disadvantaged. And an internal ethical standpoint is key, both for the individual and for the society you live in. And there are unfortunately, there are lots of places on Earth where that doesn’t seem to play a role. But not here.
Silvija: Mark, we again strayed a little bit from practical tools and tactics for applying quantum computing in our work. But I think it’s really interesting to have a touch of social reflection when you reach a technology as disruptive as the one we are talking about now. But what I get from our conversation related to tools here is that we need to read, we need to perhaps watch YouTube videos and these your TEDx talk will look up and follow Scott Aaronson. And then those of us who have programs might try to play with these simulators and, most importantly, try to rethink the way that we slice our current problems and solutions in new ways.
Dr. Mark: Yes. Very succinctly put, you did what you did what I’m incapable of doing, which is explaining it clearly.
Silvija: No, you are. You’re extremely interesting and I don’t think I dare touch this topic with many other people. So with that, we’re going to end our third lecture and we’re going to meet for a very brief fourth lecture where we’re going to try to apply some of this. I’m not sure that’s practically possible, but we can do a thought experiment. Thank you.
Du har nå lyttet til en podkast fra Lørn dot Tech. En læringsdugnad om teknologi og samfunn. Nå kan du også få en læringsserifikat for å lytte til denne podkasten på vårt online universitet Lørn dot tech..
Velkommen til Lørn dot Tech. En læringsdugnad om teknologi og samfunn med Silvija Seres og venner.
Silvija Seres: Welcome back to the fourth lecture on Lørn Masterclass in quantum computing with Dr. Mark Mattingley-Scott and I got the name order right. Right.
Dr. Mark Mattingley-Scott: So doing pretty well. I mean, yeah.
Silvija: Yeah, I’ll explain why I’m struggling but off the record, so Mark, this fourth lecture we’re going to do very briefly and partly because I’m sure that we exhausted our audience, partly because the fourth lecture is a workshop and it’s usually okay. So we talked about A.I. So what would that mean to me in my company, let’s say lørn. And with quantum computing, we are not there yet, but we can do this thought experiment. You said call me back in 2028. So imagine we are in 2028 and I’m calling my pal Mark and asking him: So I heard, you know a lot about Quantum and I’m wondering if I should, you know, do something with Quantum in my company. So I’m not doing pharma, I’m not doing materials, I’m not doing oil or batteries. I have a service company and we do Educational platforms for businesses. And where could it play a role for me? What would you tell me?
Dr. Mark: So we take the grid of where quantum computing is going to be useful. And we lay it over your particular business process, your value chain, and that applies to you, but it’s going to apply to everybody. So first question, is there anywhere in your business and your particular value add where materials and properties of materials are being investigated or used? Probably not. Okay. So the second one is.
Silvija: Pure service company.
Dr. Mark: Okay. So the second one is, is there anywhere where you’re using machine learning or where you are using deep learning or statistics or looking at patterns? I would guess yes.
Silvija: Very much so. And that probably means yes, for just about every company. But let’s say in my case, we are gathering data. Not yet, but we are building a platform. And on that platform we are going to be gathering data about people’s learning preferences and then we are going to be trying to use that data to optimize the effect of learning. So personalized learning paths, clustered groups of students with similar interests for social learning, tagging of content for better structured, personalized universities, etc.. So something along those lines.
Dr. Mark: Okay. So then the question is, I would ask is if you are able to do any of those in real time. So typically things that where today you may have a significant turnaround time. Where would it make sense or where would it be of benefit to do something in real time? In real time means as somebody is typing in on the keyboard, you’re anticipating or predicting what that might mean both for them as an individual and the group.
Silvija: So, again, I’m going to allow myself, you know, some not poetic freedom, but like science fiction freedom. And you have to tell me if I’m completely off. But one of my big dreams with this thing we’re doing is we’re now gathering a lot of content and we are tagging it, you know, for a topic, maybe for the perspective, whether it’s a scientist or a founder or a politician who’s talking. But I went to Oxford as a teacher and I love the Oxford model where you have a mentor. It’s this kind of Socratic style teaching. And so the real dream would be that any student on our platform would have a conversation with a platform which would serve it the right bits of knowledge from this extensive library that we are building. And for that, the system would have to be able to understand the different bits of every conversation and be able to pull out whatever is relevant. So, you know, it would be the sum of all the knowledge that we managed to record now and served in just the right way. So. Something like that.
Dr. Mark: So I think there are two, two particular facets to that. The first one is the eternal dream, if you like, of truly universal AI. In other words, systems which can reason and think like humans do. I have my own very, very clear thoughts on that, which I won’t go into right now. But suffice it to say, I don’t think we’re going to be at the point where machines can truly reason, even within eight years. The other one is a case of scope and capability to churn data and to do pattern recognition on data, to label data. And that’s certainly something where quantum computing almost certainly will have a significant impact. In other words, as knowledge is being created to incorporate that knowledge actively, we see the beginnings of that in attempts to assign or apply the scientific method to business processes. I think that’s going to accelerate in terms of just where’s the data, who’s creating data, who is creating new knowledge, and how is that new knowledge being assimilated, structured, analyzed, categorized. Quantum computers will play a huge role in that. So the bottom line is, at least on that side of things, quantum computing will impact your particular business. The third, third filter or sieve we need to apply is around things like Monte Carlo. In other words, things using randomness and uncertainty to predict to find patterns both in data and in structures. So the question there is, and in a learning environment, that’s almost certainly also the case. Certainly around the idea of what’s the raw information, what’s the information you’re disseminating. But maybe also in the area of. What’s new? What am I missing? If I’m operating a learning platform, I mean, there’s stuff out there happening in the area I’m teaching about. But one thing I can guarantee, if you look historically, at the historical record, there is for sure stuff out there happening which you’re not aware of. But will, in retrospect, proved to have been of key value and doing scenario planning and trying things out and looking at, well, what am I missing? What might I be missing? Those kinds of problems, those kinds of problems and solutions to those problems, using things like Monte Carlo, exploring randomness in a very scalable way that will almost certainly also be further than it is today or possible. And that’s almost certainly something you’d be very, very interested in.
Silvija: I’m thinking. So I’m not interested. My primary problem is not maximizing revenue or business models, but it’s maximizing learning and knowledge models. So one of the real hard problems I keep meeting is that, first of all, these psychological tools that try to guide you to what you should learn more about. But they’re very abstract in a way. So I would love to have a really good recommendation engine for people to become more, let’s say, cross subject. Get out of your echo chamber or your silo. I guess things that would be really good recommendation engines based on your learning data, optimizing your learning effectiveness, things like that could be then personalized efficiently.
Dr. Mark: Yeah. So I think that the state of the technology today using classical computing, classical networking, classical data, databases and knowledge management, knowledge representation and machine learning, where we are today in the next 6 to 7 years will also make huge, huge leaps in performance. What I think quantum computing will bring will be the capability to examine forward looking scenarios to do what if, what if this happens? What if that happens? What would that mean? And I’m actually as I sit here I’m looking at the book on my bookshelf on Schleicher, it’s in German, but it’s Scenario Management. So this was a science of economics invented many years ago, I think World War Two, it originally originated around scenario planning. So what might happen? What could happen? What was almost certainly not going to happen with the objective of determining what are called key indicators. So what are the key predictors that something’s likely to happen or something is not likely to happen? And what are the outliers? What are the extreme cases? How do I spot them early? There’s a whole science behind that. It’s been very theoretical, I think things like that might become more mainstream with the advent of quantum computing, because you can start to do what if analysis. You can start to look at sensitivity analysis. You know, where are the things that when I change them, they have a massive impact on the outcome. So in a learning environment, that’s going to have a huge impact.
Silvija: Both about what you need to learn, but also perhaps how job demands are going to change our collective knowledge and where it has to evolve, etc..
Dr. Mark: And also very specifically, what’s the best way to learn? So again, a personal example. When I came to Germany 32 years ago, I spoke German. My German consisted of Slabia bitter and again, it’s a personal anecdote, but the fastest way to learn a language is, as I say, on the pillow. And my girlfriend at the time, she spoke many, many languages, five or six languages. And she insisted that the fastest way to learn a language was to learn like a baby. In other words, you just repeat the sounds you don’t already know. You don’t worry about vocabulary, you don’t worry about syntax, you don’t worry about any of that. You just make sure, and as an English person, you make sure you can pronounce a, ø, æ because they don’t exist in English. Now, looking back 32 years, that was exactly the right thing to do. Did I know that? No, I was you know, I was headed for Lingua and all the language schools, other people that may not work. Being able to detect, adapt and use that in the most effective way to learn not just languages but other things. That’s a very, very interesting capability. And if you have the flexibility and a system that can detect and adapt on the fly, that would be, I think, be very, very valuable. Other than my appalling anecdote about how I learned German.
Silvija: But adaptive systems that are truly adaptive to your kind of learning psychology. I think that’s one area where we’re trying to experiment because the systems that exist today are incredibly one dimensional in the way that they think about pedagogy. Here’s a video and then there are three quizzes, multiple choices. Go ahead and do that for all the topics. There is a professor. He does a monologue. We’re done. You know, rather than finding case based discussions for somebody, it’s a short video for somebody. It’s a story that they listen to while they hike. And for somebody, it might actually be a reading text or lots of images or… So now with digital tools, this box is open and there are many tools for super easy production or reproduction of the same piece of learning in many different formats. But we’re not doing that. You know, schools are still putting out videos of lectures and calling it digitized education.
Dr. Mark: So please don’t do this to me. I’m in Germany. This is digital. Digital LA Vista for the education system. Okay. I hope that with the elections coming up, we’re going to have a government who takes digitalization seriously. I think, in Norway and in the Nordics in general, because you have such huge distances, you are much further than Germany, certainly in the area of, when I was active in the area of telemedicine, you’re much, much further ahead. But I understand what’s possible and what’s actually done are two very different things.
Silvija: Mark. I think that workshop wise, I’ve exhausted my quantum computing chamber. I think I’m going to basically read up on some of the books that you recommended, and they might not be the qubit books, but things like secular cycles. I’m going to head for now and I’m going to look up your YouTube. No, sorry, your TEDx talk, wherever I find it. And I think we’ll move from there because I think looking for good examples of somebody already thinking about applying this is going to be the easiest way forward, at least for me. You’ve opened up a subject that I was really scared of but knew that is really important and I’m hoping it had the same effect on our listeners. Thank you so very much.
Dr. Mark: I hope so, too, and thank you very, very much for having me, Silvija.
Silvija: Pleasure talking to you, Mark. And thank you all for listening.
Du har nå lyttet til en podkast fra Lørn dot Tech. En læringsdugnad om teknologi og samfunn. Nå kan du også få en læringsserifikat for å lytte til denne podkasten på vårt online universitet Lørn dot tech.
Du må være Medlem for å dokumentere din læring med quizes og motta læringsbevis. Prøv et
Allerede Medlem? Logg inn her
Du må være Medlem for å dokumentere din læring med quizes og motta læringsbevis.
Allerede Medlem? Logg inn her