While quantum computing technology is maturing more slowly than other innovations, its potential is vast. Rob and Jackie sat down with Murray Thom, vice president of product management at D-Wave, to discuss quantum computing applications already being put into place, and possible advancements in the future.

Mentioned

• Hodan Omaar. “The U.S. Approach to Quantum Policy,” (Center for Data Innovation, October 2023).
• Congress.gov. "Text - H.R.6227 - 115th Congress (2017-2018): National Quantum Initiative Act," December 21, 2018.

Related

• “Quantum Computing: The Case for a National Task Force,” (Center for Data Innovation, February 2022).
• Hodan Omaar. “What Should Congress Include in The Next National Quantum Initiative Act?,” (Center for Data Innovation, May 2023).

Auto-Transcript

Rob Atkinson: Welcome to Innovation Files. I'm Rob Atkinson, founder and president of the Information Technology and Innovation Foundation.

Jackie Whisman: And I'm Jackie Whisman. I head development at ITIF, which I'm proud to say is the world's top-ranked think tank for science and technology policy.

Rob Atkinson: This podcast is about the kinds of issues we cover at ITIF, from the broad economics of innovation to specific policy and regulatory questions about new technologies. So if you're into this stuff, be sure to subscribe and to rate us. It really does help.

Jackie Whisman: Our guest today is Murray Thom, who's vice president of product management at D-Wave, a Canadian quantum computing company that was the world's first company to sell computers to exploit quantum effects in their operation.

Murray was responsible for the development and delivery of the Leap quantum cloud service and the ocean open source tools. He's led teams engaged in customer projects related to algorithms, applications, and performance testing. He's even assembled a few early quantum computers by hand, and we're happy to have you. Thanks for doing this.

Murray Thom: Yeah, it's my pleasure. It's great to be here with you both.

Rob Atkinson: We've done a bunch of work on quantum. We actually had a great event with a quantum scientist from Australia who was named person of the year, and she was great. And she told me all this stuff, and it's kind of like in one ear and out the other, annealing and qubits and all that, and I kind of know what it is, but I doubt most people. Could you just give us an idiot's guide, two minutes, what's quantum computing, and how does it differ from conventional or traditional bit computing?

Murray Thom: Yeah. I mean, even the first part I can do in less than two minutes. So a quantum computer, at its essence, is a device, right, like a computer chip and its peripherals that allow us to interact with it that is taking advantage of quantum effects to accelerate calculations for us. So it's really that simple. A device to use quantum effects to accelerate our calculations. And quantum effects have a lot of weird phenomena that don't really make sense in our world.

So, for instance, if I had a loop of metal and I were to try to flow current through that loop, I could flow it counterclockwise, or I could flow it clockwise. And when I do that, I'll get magnetic fields, and those magnetic fields will interact with other things in our environment. In a quantum system, it can be both counterclockwise and clockwise at the same time, and we would normally think, "Well, those currents are going to cancel each other out. You won't get any magnetic fields. There won't be any interactions."

But quantum systems, that doesn't take place. In fact, both types of magnetic fields get represented, and both types of interactions carry on. If you think about states like being in multiple states as once as in it for a computer, that's information. That's how it's basically able to use those states in order to change the way calculations are done. And it turns out that being able to occupy those states, being able to put its base building blocks into those states allows it to find solutions for us more quickly.

Rob Atkinson: That's a great explanation. So let me ask maybe a stupid question. So in a traditional computer, you have a one or a zero. It's either on, off. In quantum, it seems like because it's not that. It's potentially infinite. Is that right?

Murray Thom: Well, it's... I mean, I have this conversation with developers a lot, and they're like, "So my memory is in multiple states at once," and they're not exactly clear how they're going to be able to write a program that leverages that. Effectively, I think probably the easiest way to think about it is if I describe a problem where I'm trying to assign grocery deliveries to a delivery fleet, and I'm trying to like... I'm going to say, "Okay, driver A is doing 1, 2, 3, and driver B is doing 4, 5, 6."

I may, as I'm evaluating how good of a schedule that is, want to swap one of the deliveries from one driver to another. And when I'm doing that, I have to actually change the state, look at it, be like, "Is this better? Yes. Okay." And then continue to explore from there. So the notion here is if I just define the task I want the drivers to do and the schedules that I want to be produced, if I can ask a computer, "Hey, explore all these possible configurations and the ways to arrange this problem, which one's most effective?"

And it starts exploring those spaces, and it starts sort of collapsing down and simplifying itself and finding a configuration that's less frustrated. Less frustrated for the computer, and less frustrated for the people actually driving the deliveries. That's all to the good. So it's actually a scenario where if you are posing the problem you're trying to solve, the computer will actually explore it for you. So it's, I think, a little less demanding for the developer.

Rob Atkinson: Interesting. Thank you.

Jackie Whisman: Access to data is not often a topic brought up enough in quantum policy conversations. But in order to use quantum computing for optimization problems in transportation, like you mentioned, you need mobility data or traffic data, and for health applications, you need health data. To what extent do you see access to data as an issue in applying quantum to public sector problems?

Murray Thom: It's very important. But let's say if there are folks listening to this from an AI machine learning perspective, their needs for data are much larger and much more extensive than what you might need for optimization. So certainly, if you want to optimize nurse schedules in a hospital, you need to know which nurses you have, what seniority they have, what their vacation preferences are. So that data's necessary. But it's not that you need sensors all over the hospital that are collecting massive gigabytes of data.

And so you need to make sure that you have the data that's relevant to the operations and the tasks that you're trying to optimize and make more effective or optimize in a more sort of responsive way. Having said that, to that model I was discussing earlier about expressing the problem you're trying to solve and having the computer explore the solution space to give you answers, you can do that in a very mathematical abstract way, and what that enables is that your end customer or your organization with the data can actually keep the data themselves.

So they're not actually transmitting data into the quantum necessarily. They're using the data to formulate an expression of, "This is the problem that I'm trying to solve." But they can basically keep the mapping between the mathematical representation to the actual data on their side, and that means what's actually getting sent to the quantum hybrid technology to create a solution is void, protected, and kind of abstracted and anonymized.

Rob Atkinson: It sounds a little bit like facial recognition technology. They don't need to have a record of your face. They just need to have a record of the dots and the mathematical representation of it, which sounds similar. So you alluded... one of the things I think that's interesting about quantum is a lot of people think it's, "Oh, we're developing this. In 10 years or 20 years, we might be able to do something with it."

Clearly that's not true as we've written, and clearly, D-Wave is you are selling machines, and you are doing applications. You mentioned a couple. You mentioned the delivery optimization. You also mentioned nursing schedule optimization. What are some of the other areas where people are actually using quantum as we speak to solve real problems?

Murray Thom: Yeah, that's a great question, and the important thing to keep in mind is that quantum computing is forming an entire industry. It's not about a device or individual technology or computer, but rather, like all the players moving this forward. And so what's really critically important for making sure that technology is getting driven in the right directions is that you have users applying it and telling you when it is and is not creating business value or mission value for them. So there have been a whole number of folks who have explored the applicability of quantum computing technology.

If we want to dive into more detail further in the conversation, I can explain how we're combining quantum and classical computers together to support these full-scale real-world application use cases. But just to give you a few examples. MasterCard is looking at business applications, including offer allocation for merchant offers being matched with cardholders and how to optimize that in a way where the merchants are seeing the business, the cardholders are seeing offers that they're interested in.

Everyone effectively is happy with the result. MasterCard's also looking at data analysis tasks like detecting fraudulent transactions in a group of other transactions. Davidson Technology is looking at a threat analysis application where they've explored, in one instance, 67 million scenarios and gotten an answer back from our quantum hybrid technology in 13 seconds. So really important from the perspective of bringing this to bear in terms of national security and defense applications.

Also, there's a small company, SavantX. They've created a software solution for the Port of Los Angeles that combines our quantum computers with classical computers. They were able to enable that port to operate and move 60% more cargo each day while reducing truck turnaround time to pick up those shipping containers by 12%. Another example that's very local to here is that the Pattison Food Group has been working with us to deploy applications specifically in e-commerce driver delivery, just like we were just talking about. And so when I order groceries from Save on Foods, the drivers who are bringing them to my house.

Their schedules are getting quantum optimized. That application has been in production for a year now since last October. So really, a wide variety of applications and a wide variety of industries. Because of that breadth you can sort of see, "Oh, classical computers are kind of a horizontal technology that you can use in multiple verticals to get an advantage." Quantum computing has that sort of similarity as well. So some real connection there and some real sort of driving business results that I think we can build on.

Rob Atkinson: I took statistics in my Ph.D. program, but I don't remember very much of it. But I do remember there were these problems that were so complicated, and I guess it's because there's so many different decision points. Just classical computing really, really would have a hard time with it, even though they seem sort of simple. To me, when you say optimizing a port, how hard can that be? But clearly it can be very, very hard because there's so many moving parts to put here then there, and all that. Is that sort of the deal?

Murray Thom: You're absolutely right. These are problems that, even with the extent to which classical computing has matured, right, 70 years of development, we've got unbelievably powerful computers available at our fingertips in our pockets. These problems have escaped being well solved, and the reason is because they're of a totally different class. They don't break down well into the small steps that classical computers do.

And so folks who are working in these application domains, they know that they're hard, but generally we kind of think like, "Well, the scale's pretty modest. They can't be that difficult." The other thing is that mathematicians who have studied them have actually formally proven these are amongst the hardest problems in the world to solve. And just some of the aspects with those [inaudible 00:10:37] is that when they're picking up containers. They're stacked. So just like Solitaire, you have to uncover the cards on the bottom in order to get to them. So the sequence with which you pick them off is important.

And then the other thing is they don't have an arbitrary amount of room for trucks to wait at the port to pick up material. They schedule the trucks to arrive in. Those trucks are moving through traffic, so they may or may not arrive on time. So you really want to be able to optimize moment by moment in order to be able to maximize that productivity and that effectiveness of the port. And that's what's great about the solutions SavantX built for them there.

Rob Atkinson: It seems to me one of the things that's cool about this now is that 10 [inaudible 00:11:14] years ago, maybe it would've been a lot harder to know in real time where the truck was, or in real-time which container... You might know it on paper, but in terms of being able to transmit it to a computer. So does that help the sort of IoT, Internet of Things world really combines together and makes that where a tool like this could really play a role?

Murray Thom: It's definitely... I mean, think about it as like a dual-edge sword here. So you have the availability of information so you can act on it. The challenging part is that there's so much information. It's making these computational tasks enormously complex. So it's the solution of bringing quantum, classical, hybrid solving technology together is kind of just arriving at the right time for many of these businesses that are struggling under the weight of all this operational information that they're having a hard time acting on.

Rob Atkinson: That's really interesting. More information it means we need more computing power, which means better solutions, maybe more information. So before we kind of jump into this... some of the policy questions and all I want to come back to, again, kind a basic question for the average or lay user.

You're developing both annealing and gate model quantum computing, and it's interesting because we don't... maybe we know, but certainly, there's a lot of different bets being placed around the world on different types of quantum systems. How are you thinking about that, and can you sort of explain what's the difference between those two kinds of computing?

Murray Thom: Absolutely. I think one of the difficulties in the quantum computing industry is that we're not talking about it enough. The goal is to use quantum effects to accelerate calculations. Not surprisingly, there's multiple ways to do that. And it turns out that the way that you... the model you choose to use is actually determines the applications it's going to be suitable for and the rate with which that technology is maturing. So we ignore that at our apparel. And in fact, that's a big part of the reason why we have huge gaps in terms of the US government policy for quantum computing and why we're really missing, I think, an opportunity to promote near-term applications.

And by near term, I mean applications that are built in 24 months or less because that Port of Los Angeles example from the very beginning to production runs was only 18 months. So the key thing I would bring to this is that there are two primary models, and D-Wave is building both of them, annealing and gate. I there was anything... If there was just one thing that people would take away, it's that annealing quantum computers are supporting real applications today, and gate quantum computers are seven-plus years away from supporting commercial and mission applications. The reasons for that are quite nuanced. Annealing took its inspiration from metallurgy, right, like thermally, annealing metals.

Turns out when you do that, if you do that slowly, the metals become soft. So it's great for making wire with high electrical conductivity that you're going to draw out. And the reason it's able to do that is because it has less internal frustration. So let's inspire ourselves to use quantum effects to anneal our problem configurations and find ways to schedule our drivers so that they're less frustrated, right, and that we're getting to better quality solutions with less driving distance and faster delivery times. Gate is taking its inspiration from the information sciences community.

If we take ourselves all the way back to our computer science roots, those classical computers are built on logic gates, and people say, "Oh, we have logic gates in computers. We'll make quantum logic gates. We've got bits in computers. We'll make quantum bits." It's very powerful model. It's going to be very relevant for molecular simulation, drug discovery, quantum chemistry, also three-dimensional diffusion equations, and partial differential equations. So really powerful modeling. It's not going to be good for optimization. And that gate method is basically using the quantum effects to allow us to store more information in the quantum computer.

The difficulty with that is that quantum states are famously delicate and quick to collapse, and when that happens, the information you put there is lost. So that's why that technology is maturing a little bit more slowly. Should that be part of the broader industry and part of the research program? Absolutely. But should we be myopic and just focusing on that and not realizing the fact that annealing quantum computers can bring us value today? That would be a mistake, an unforced error I don't think we can afford to make.

Jackie Whisman: Interesting. We just published a report called The US Approach to Quantum Policy. In it, we discussed that it's been five years since the first National Quantum Initiative Act, and policymakers are working to reauthorize it. What do you think has been successful in the past five years, and where might policymakers need to step in?

Murray Thom: Well, I think it's helpful to have the NQIA as a framework, and I think that it's helping to bring focus to what needs to be done in this industry because the last thing you want to have happen is that it's moderately complicated. You don't want people to miss that and miss those opportunities. But a lot of the work is remaining in the research phase in national laboratories, and a lot of times, it's focused on whoever's the loudest voice in the room. I'm sure that you have seen in other policy areas, you have really big players, and they're trying to squelch the voices of other innovative players who are trying to come up with advanced solutions because it just doesn't fit with what their approach is going to be.

So I think what needs to happen here is we need inclusivity. We need to recognize that there are different models of quantum computers being built and that those programs need to be developed specifically to recognize that and be open to that. Let players compete on the open stage in terms of actually bringing people value because I think that will bring a lot of benefit. And I also know that the ITIF report in 2021 really brought a focus to near-term applications, the kinds of things that can be built in 24 months or less, that I think is missing from the NQIA. So the NQIA needs to be reauthorized. It needs to be expanded. I think there needs to be support for the Quantum Sandbox Bill in Congress, which is focusing on these new term applications.

And I think if we do that, we can see a lot of value there. And then we need to provide support for the sort of fiscal year '24 NDA, so which is like a quantum pilot program for defense programs, a little bit like a Quantum Sandbox Bill for the Department of Defense. And we should also open our eyes to... What I want to make sure that nobody misses is that this can be brought to bear for challenging problems today. So even as we're trying to address wildfire emergency scenarios, we can start investigating how quantum technology can help us in those efforts. So yeah, let's see if that kind of expansive program that reauthorization and expansion can really get supported.

Jackie Whisman: Quantum computing is a difficult technology to innovate in. It's a difficult technology to explain to me. So it requires a skilled workforce, no doubt. How difficult is it for businesses like yours to find skilled workers and attract and retain the talent you need to innovate?

Murray Thom: Well, there's no doubt that talent development is critical. It might surprise people to find out that, in fact, physics is only one component of our workforce here. We've got electrical engineers, mechanical engineers, experts in machine learning in graph theory, and programmers and developers. So really, that needs to be broad-based in terms of expertise. And the difficulty is that some of the programs, particularly in the US, are focused on K to 12, but businesses don't have 20 years to wait for their quantum-ready workforce. We really need to be thinking about this in terms of talent upskilling and workforce training.

Certainly, at D-Wave, our... we are filling a particular niche in that training paradigm, which is one-week training courses for professionals who are trying to solve the problem right now. And I think there's lots of room in that ecosystem of talent generation to create programs, projects, and undergraduate programs, and also for folks who are close to graduating so that they can get skills, particularly those who are studying business optimization, which I think is a critical need. And to kind of make people aware of the fact that there's certainly a lot of focus on AI and ML tools, but they shouldn't be using it... those innovative tools and missing the fact that quantum computing is an important element of their toolbox.

Rob Atkinson: So would part of this be to integrate the sort of knowledge of what quantum can do into operations research degrees?

Murray Thom: Yeah, absolutely. I mean, that's a community that I would say the NQI is missing. The quantum information science community and physicists is missing and those are the users, right. As big as maybe a portion of the industry that quantum computing hardware developers are going to be the ecosystem and the industry of folks who are using quantum technology to get... to advantage is going to be much larger, and that's going to be that community of folks in operations research, but it's not just that kind of specialized knowledge.

One or two or a small team of those folks can actually support a lot of software developers who are not taking four-year computer engineering degrees but are able to write and are currently writing business software. So we would benefit a lot, I think, by recognizing that broadening out that support and talent training.

Rob Atkinson: This podcast will come out after my testimony next week. But I was asked by Senator Schumer and Senator Young to participate in one of their AI round tables that there was informal listening sessions on the question of jobs and skills. And one of the points I make in my testimony is, and this is not a knock on kids who take art because my daughter is... she loves art.

She's great at digital art. But something like 77% of high school kids take art, and 6% of American high school students take computer science, 9% take statistics. So pretty shocking really when you think about it. I think it's 1% take the APCS test, which is crazy, and then you go into engineering and physics, and it's probably even worse there. So there's just an awful lot we can do and striking that we don't seem to have the political will to really do anything other than talk about the problem.

Murray Thom: Yeah. Well, and the thing is that we have seen this [inaudible 00:21:17], this awakening for artificial intelligence, and that's not because suddenly 100,000 people learn statistics. It's because we were able to develop a powerful technology that became really easy to use.

Rob Atkinson: Right.

Murray Thom: The importance of that talent training and helping to encourage and drive folks into those technical fields, it's important to recognize the point that you're making, which is that they're not just naturally going to fall into that space, right. They need encouragement.

That program needs effort and intention and design. But the value of doing that is that those are the resources that can make accessibility much simpler for a broader group of folks. And then society as a whole is getting a larger benefit if we succeed at making those complex things simple.

Rob Atkinson: Yeah, absolutely. I spoke at an event at the National Academies this week on a release of a new report on critical technology assessment that was just done, and it looked at a number of technologies. And there was sort of this, I don't know, I thought kind of snarky remark about, "Well, Congress doesn't do anything." But I was kind of like, "Well, wait a minute. Hold on. The CHIPS and Science Act, that's not chump change."

And so I guess I'm saying I feel somewhat optimistic. I do think Congress is really, and the administration, everybody's looking at this issue. And as we talked earlier at the beginning, National Quantum Initiative Act is being reauthorized. So it's probably a good place to close on an optimistic note and say, hopefully, we'll do more, and we'll do better, and we'll keep driving this key technology along further.

Murray Thom: Let's support them in any way we can.

Rob Atkinson: Great. Murray, thank you so much for being here. This was really, really interesting, and I feel like maybe now I finally understand quantum computing, whereas before, when I listened to other people, I didn't quite all the way go in. So it was really good explanation.

Murray Thom: Thanks. It's a pleasure to be here, and we're all learning something new every day, right. So myself as well, even I'm learning things about quantum mechanics as we in quantum computing every day at D-Wave.

Rob Atkinson: Cool.

Jackie Whisman: And that’s it for this week. If you liked it, please be sure to rate us and subscribe. Feel free to email show ideas or questions to [email protected]. You can find the show notes and sign up for our weekly email newsletter on our website itif.org. And follow us on Twitter, Facebook, and LinkedIn @ITIFdc.

Rob Atkinson: We have more episodes and great guests lined up. We hope you'll continue to tune in.

Jackie Whisman: Talk to you soon.