Waking Up With AI

Katherine Forrest: Right, and we've talked in prior episodes about chips, semiconductor chips, and there are different kinds of chips, some of which are used with AI, some of which are used in just the computer for regular computational functions, but they can make the computations of a classical computer, which is what we're talking about first. That's what everybody's familiar with every single day, that's what makes the classical computers work really powerfully.

Anna Gressel: Yeah, and there's a concept called memory that's used in computing. I think we've all bought a laptop and heard about how much memory it has. And that's generally the concept of information that's stored in a computer and can be called upon for processing. So that's just like what we normally deal with with regular computers.

Katherine Forrest: Right, and how many times have you gone to the store to look at a computer and it has that little card with all that information on it and you have no idea what any of it means but what you do is you compare it to the card next to it and you see is it more or less, are the numbers bigger or smaller in terms of gigahertz and everything else. Anyway.

Anna Gressel: Well, I remember that the last time I did that was when I was in Abu Dhabi and my computer broke. So this is a very recent experience for me.

Katherine Forrest: Right, you're probably looking at it with all kinds of other languages as well. Now, with quantum computing, it's very different from this classical computing that we've just been talking about. Instead of someone who's a really good organizer and processor of information, sort of one step at a time, even if things are broken into parallel steps, it's really something that can do multiple things at the same time and instantaneously.

Anna Gressel: Yep. And last week we talked about superposition, which is when quantum particles can be in two states at once. And we talked about entanglement, which is like, by the way, Marvel's word, I feel like, for their whole quantum set of movies. But entanglement is basically the concept of two qubits [quantum bit] that can become linked together so that changing the state of one or understanding the state of one can immediately give you information about the other. Both of these concepts are both of these properties – superposition and entanglement – can help with making superfast computations. so effectively that results in massively parallel computations where all the possibilities for an answer that could take like millions or in my mind, bajillions, it's like my favorite word, of years to compute actually could be done instantaneously with quantum computing. And so when you actually put all that power into a computer, a quantum computer, it can do really extraordinary things.

Katherine Forrest: All right, and so that's different from the classical computer. And I want to just talk for a second about something that a classical computer can do that we just mentioned a moment ago, which is parallel processing. Because quantum computing is different from the kind of parallel processing that a classical computer can do. Classical computers have these chips that we talked about, as I said, in those prior episodes, and they can do these tasks at the same time, but they do it in parallel, so they take different steps and they break them down into parallel processes, but everything, every bit is still at all times either an o or a 1 until it's actually changed from that o to a 1 or from that 1 to an o. And so a quantum computer doesn't just split tasks across multiple processors or do things in sort of singular steps in parallel, it actually explores multiple possibilities at the same time. And that is done by what we talked about before, which is this superposition state. So the speed of a classical computer is limited by the number of processors that it has and something called clock speed, but a quantum computer can evaluate many possibilities instantaneously.

Anna Gressel: Yeah, one concept that can help explain this is a maze. With parallel or kind of traditional processing, a computer would explore the maze one path at a time. They'd figure out, you know, would it work, would it not work. We've all done that. But in quantum computing, all paths in the entire maze can be explored at the same time.

Katherine Forrest: So you'd never ever have one of those corn mazes again that you couldn't explore all at the same time. And so what this means is that if you have a really hard mathematical problem in the world of quantum computing, something like breaking a code or, you know, de-encrypting something, it can be done incredibly quickly because every possibility can be explored at the same time. And so one of the reasons that as we approach a time when quantum computing is becoming something real, there are a lot of cybersecurity experts who are becoming increasingly concerned about ensuring that our computer networks are really secured because the ability, again, of a quantum computer to find cyber vulnerabilities, to find those openings, those holes, in our networks will be incredibly powerful and can happen incredibly quickly and it would be a very scary thing if we didn't really lock things down.

So, many of the problems that are hard for classical computers and which underlie certain crypto systems that are out there today are actually really easy for quantum computers. Now NIST, which is the National Institute of Standards and Technology, has really foreseen this problem and as of last August has finalized three post-quantum encryption standards, those are encryption standards that are really oriented towards a time when quantum computing becomes commercially available.

Anna Gressel: Yeah, and those are global efforts. I think it's such a good example of global cooperation around something that is going to be a global challenge. But another really positive example of the power of quantum computing would be to try to simulate molecules for pharmaceutical drug use cases. That's already something that we can do with AI, but with quantum, all of those combinations, all of that computing could be run simultaneously very, very quickly. So that could really turbocharge drug discovery.

Katherine Forrest: Right, and there are computer scientists and people from virtually every corner of the tech world who are really fired up about the potential that quantum computing has because it could unleash the most powerful computers really imaginable.

Anna Gressel: Yeah, and this fits into our AI world since we know that AI is about having computational power, and the kind of massive computational power that quantum computing would bring would result in like pretty extraordinary advances in AI.

Katherine Forrest: Yeah, that's right. If we're able to actually make the advances in quantum computing that many people are hoping for and expect to happen at some point, then there are also going to be translated achievements, if you will, in the AI sphere.

Anna Gressel: Yeah, and I think one of the things that we don't always talk about with quantum computing personally, but I find really, really interesting is the idea that actually with quantum computing, we could use those computers to model the physical world. So there's also something about the fact that it can itself model molecules, which may really lead to an expansion of scientific discovery and material science. And so, I mean, Katherine, you know me, like scientific use cases get me fired up and this is one of them. But I want to just digress briefly into a somewhat geeky topic for a moment.

Katherine Forrest: Okay, and I just want to tell our audience, I'm worried. I'm worried because I know what's coming.

Anna Gressel: No, this is great! Katherine, you love both of the words I'm about to use, I'm just going to put them together. So you love the word quantum and you love the word algorithm. So we're going to make this painless, but we are going to talk about quantum algorithms. So quantum algorithms are designed to run on quantum computers. They're algorithms that take into consideration all of those massively parallel possibilities of the quantum world. The world where many things are possible at the same time or in like Marvel terminology, the multiverse. I'm just joking. But it's all of the powerful quantum possibilities we've been talking about so far in our two episodes.

Katherine Forrest: Well, you know I love quantum physics, but it does sound a little bit like science fiction.

Anna Gressel: No, I mean, I think, you know, it sounds like it, but quantum algorithms are one of the cornerstones of what makes quantum computing so interesting and exciting. So I think it's worth talking about it briefly. But there are huge numbers of experiments and labs all over the world that are really proving up the principles of quantum physics and the fact that they are real. And so we just want to understand what the algorithms are doing and what role they play in it. So I want to talk very briefly about one of the biggest issues that has to be solved in order for quantum computers to actually experience the next advance.

This is an important limitation of quantum systems, but it's also an important promise of them, and that is stability. So in order for a quantum computer to work, there has to be stability of these qubits. The qubits are really fragile. Remember, we've been talking about them. And their state can be changed by things like heat or electromagnetic interference and when something interferes with the stability of one of these qubits, it causes a process called decoherence.

Katherine Forrest: All right, now we're back to the world that I understand. And with decoherence, so when you've got a qubit that becomes unstable for one reason or another and you have this decoherence, that can cause an error in a calculation. And quantum computers need thousands or even millions of these qubits, and they need them to be held stable long enough for the calculations, these incredibly powerful calculations that we're talking about, to be performed. So getting to the point of stability for the qubits is the key to quantum computing.

Anna Gressel: Yep, and there are a bunch of elements to this stability. And so to really realize the advances in quantum computing, we have to actually create the right conditions to enable stability in our quantum computers. And one of those conditions is temperature. I mean, we all hear about how cold quantum computers are. Another one though…

Katherine Forrest: I don't think everybody is talking about how cold quantum computers are, Anna, I don't know what universe you're living in. But I'm talking about where my next, you know, what restaurant I'm going to next. Anyway, back to stability being impacted by cold and heat – by heat really. It has to be very, very cold to actually help with that stability.

Anna Gressel: I'm just going to; can I just make a call out to our audience? If we have any quantum physicists on the line who want to invite me to their quantum computers, I will come with my biggest puffy coat on and I will be super excited. I’m like manifesting this to the world that I want to be invited to go hang out with a quantum computer. But let's talk about this. One of the issues is temperature, we've talked about that. Another one though is, you know, these quantum algorithms that actually enable error correction in this process because they deal with this issue of decoherence. These algorithms are called quantum error correcting codes. That's kind of like our big vocab word of the day is quantum error correcting codes. The algorithms help to correct errors that emerge from all of the noise in these quantum systems.

Katherine Forrest: All right, so okay, we've made it, through quantum algorithms, and they're actually incredibly important to assisting with this stability that we're talking about, which is one of the keys to quantum computing. And now we've actually got, with our two episodes, superposition, entanglement, we’ve talked about the difference between classical computers, are the ones that we all use every day, and quantum computers. And we've talked about the issue with some advances for quantum computing, which are things like stability, and then your favorite, the quantum algorithms that help with those quantum error correcting codes. And so now we've given everybody the background, and we're ready to talk in episodes down the road. We're going to give people probably a little break from this. We'll talk about some of the really incredible strides that are being made in the quantum computing area right now and that are also making some real advances in AI, at least theoretically possible. But that's all we've got time for today, Anna, I'm Katherine Forrest.

Anna Gressel: I'm Anna Gressel. Like and subscribe to the podcast if you're enjoying it.

Katherine Forrest: And invite Anna to your quantum computer if there's anybody out there who's got one in their garage.