The war in Ukraine reveals the growing challenges and opportunities for the North Atlantic Treaty Organization’s military forces. After surprising success early in the war, Ukraine is now running short of munitions, and the precision weapons Kyiv assumed would offer an advantage are now failing in the face of Russian electronic warfare and innovative tactics. To stay ahead in this conflict and future confrontations, NATO militaries and defense industries will need to both adapt and scale.
A Hudson panel at the NATO Public Forum with government and industry technology leaders discussed how NATO nations can get beyond “innovation theater” and quickly field relevant capabilities in the numbers needed for protracted wars.
Event Transcript
This transcription is automatically generated and edited lightly for accuracy. Please excuse any errors.
Bryan Clark:
Thank you, Shannon, for that introduction and for everybody for being here today. It’s a great honor to be able to participate in the NATO Public Forum of which it’s an institute, is one of the sponsors/organizers. With me today, I think we’ve got a terrific panel to talk about the role of emerging technologies in military operations and how we can bring them to bear to support the alliance’s efforts both in Europe and elsewhere. Dr. Bryan Wells is the chief scientist of NATO. Dr. Steve Walker is the chief technology officer for Lockheed Martin and Rand Waldron is the vice president at Oracle for cloud infrastructure.
Gentlemen, thank you very much for being here today, appreciate it. I’d like to start the conversation with one of the big lessons that comes out of the operations in Ukraine. We’re finding that we have to both be adaptable, but we also have to be able to scale military capabilities. We’re already seeing in Ukraine some of the capabilities that the West has given to Ukraine. GPS guided weapons are having difficulty being effective in a jamming environment that denies GPS.
At the same time, we have to be able to then scale up those solutions. We can’t just come up with an experiment that delivers a potential solution at a small scale. We’ve got to be able to deliver that at a relevant capacity to impact the war fight. How do NATO nations and the alliance achieve that combination of adaptability and scale that will be so important going forward not just in Ukraine, but in any future military operation? Dr. Wells, we’ll start with you.
Bryan Wells:
Bryan, thanks for that question. I’ll start by reminding ourselves of something we know very well, but it’s worth starting as a scene setter. This summit is celebrating the 75th anniversary of the NATO alliance. Let’s wind the clock back 75 years. In science and technology terms, the technical breakthroughs were predominantly the preserve of the defense sector. The defense owned the technical futures. If we wind forward to 2024, we now find that within government research investment by government in the civil research dwarfs that of government investment in defense.
When we compare government investment in research with commercial investment commercial investment is significantly greater in the vast majority of areas and the commercial decisions of commercial companies drive the future of technology across the piece. NATO recognizes this and it’s got a good history of adapting to changed circumstances and it’s changing in this way as well. This brings new choices for our policymakers. Acquisition processes are adapting to the new generation of commercial companies that we see.
I’ll give three examples. I can’t be exhaustive, but the trend I think is clear. The first is DIANA, which is not part of my organization, but I know there’s a great deal of interest in what DIANA is doing. This was launched in 2022 to work on early stage start-ups, but also mature defense companies. Both areas of the industrial sector bring innovation and it’s important to NATO to have different levers as it were to capture that innovation. DIANA runs by setting out competitive challenges to our industry.
Successful respondents receive grants and access to a network of accelerators and test centers. Ultimately, of course, their route to market can be assisted by NATO and by the allies. Three challenges have already been launched in 2022 and 44 companies were selected. A total of five new challenges were launched at the beginning of this month, 1st of July. The second area I would like to highlight again is not part of the NATO science and technology organization, but is an important part of our rapid adoption to market, and this is the NIF, the NATO Investment Fund.
I know that NIF is in the audience. Please bear with me for a very quick overview. This is a 15 billion venture capital fund. It’s the world’s first multi-sovereign investment fund and the first investments were made last month. The third area is in fact my own science and technology organization. It’s a network of 5,000 scientists across the globe drawn from NATO allies and like-minded partners. Looking at the young audience particularly, if you have a scientific background, do please feel free to look at the website, which gives you instructions on how to join. I hope some of you will express interest.
This is drawn from government laboratories, industry, and also academia. Let us not forget that academia remains an important part of the innovation network. These are just three areas of the NATO innovation ecosystem to bring new capabilities rapidly to market. The point I would make at strategic level is that there’s good links between all of these areas. Research from the science and technology organization could be matured by DIANA, funded if appropriate by the NATO Investment Fund. This means that the work of the whole of the NATO innovation system brings a greater value than the sum of its individual parts. Thank you.
Bryan Clark:
Thanks, Dr. Wells. Dr. Walker, that highlights one challenge of that might be then scaling the results of those. If you find something that seems to work, it solves a problem like GPS jamming in Ukraine, how do you make that turn into something that’s relevant capability and capacity for the kind of fight?
Steve Walker:
Right, I think, Bryan, your question was well stated because it’s adaptability. It’s innovation, but it’s also scalability. Scalability in my view, that’s all the ilities, reliability, safety, O&M, sustainment. Ukrainians have done a great job with what they had to work with and I had the amazing opportunity to go over in 2018 to Ukraine as the director of DARPA.
I was amazed at their venture and commercial capabilities then, small company focused. They’ve been able to show how in wartime they can take that, use that in the field, develop software, drones, et cetera, but I don’t think the scalability is quite there where it needs to be. Jim Taiclet, our CEO, came to Lockheed about five years ago. I did as well.
We put in place what we call 21st century security, which is really all about bringing those best defense platforms to the fight to deter our adversaries, but adding in the bleeding edge technology mostly in the digital world. Most of that is happening in commercial. Partnering up with commercial companies and then adding speed to that equation, that’s what we mean when we say 21st century security. We’re focused on commercial partnerships with big and small companies.
We have a venture fund, about $400 million venture fund, that focuses on bringing that small company innovation into the defense ecosystem. We feel like we understand the war fighter. We understand the military problem set and we can bring those ventures and small companies in. We can also work with the Microsofts, the Oracles of the world, the NVIDIAs of the world to take that bleeding edge tech, especially in AI and machine learning and other areas like that, work with them to solve problems.
In the case of NVIDIA, we’re working with them on fighting fires and what does that look like all the way from sensing, using AI and machine learning to sense to control the firefighting capabilities, multi-platform capabilities, to battle management and understanding what’s happening in the fire, in the forest at the time? That’s one example of how we’re doing that. Speed is the third part of that equation. We’re making huge investments within the company to make digital engineering and model-based systems engineering real and being able to use that and get faster at how we do the business.
All three of those pieces are going to be important to that scalability question and applying these innovative concepts. We do innovation too within Lockheed. The camera on the James Webb’s telescope is a Lockheed Martin camera producing those amazing pictures. Those three pieces are going to be critical to really creating a new defense industrial base, which is going to be absolutely essential for the wars we need to deter in the future from pure adversaries.
Bryan Clark:
Right, right, thanks, Dr. Walker. Mr. Waldron, the-
Rand Waldron:
The one non-doctor up here.
Bryan Clark:
Steve brought up the point about software and how digital technologies are really at the heart I think in a lot of this adaptability and scaling. How are you finding that software and what you do at Oracle is playing into this ability to adapt and then bring things to relevant capacity?
Rand Waldron:
It’s at the very heart of it. I’m not going to tell anyone new by telling them that bits move much faster than atoms. When we need to innovate, when we need to be at the very leading edge, the bleeding edge of capability, there is a lot you can do with metal and composites and atoms, but those take more time. Software takes time too, but you can build a base and then be able to iterate on it very, very quickly.
What ministries of defense, what governments have to look for is the way to make sure that you are buying not a single item, a fixed thing, but make sure that you’re buying an innovation pipeline, something that allows you to innovate and iterate very quickly. My Lockheed compatriot up here, like the F-35 is not just a physical thing. It’s a fighter plane that iterates software as fast as it iterates new armament. When you look at Ukraine you look at what they’re doing with drones and all of the other things that we see in the news every day.
So much of that iteration takes place not just on the physical item, but on the software that’s running on the item, on the software that’s on the backend on how they collect, how they analyze the data. All of our big companies, our small companies, our government agencies, we have those innovators inside who want to be able to push the envelope, who want to do that next innovation. When, God forbid, there’s a fight that’s going to drive innovation speed like nobody understands. What we need to make sure that we’re doing is providing the foundation so they can innovate that quickly.
What I would say is look at it not as making a single purchase, not as making a single change. This acquisition makes us now innovative. This is the proof we’re now innovative. It’s to create a technological underlayment and a cultural foundation that allows you to iterate continuously, that allows you to build that innovation pipeline because we don’t know what the next innovation is going to be. Make sure that you have today not the thing that’s fixed in time, but the thing that evolves as fast as you and your developers need to.
Steve Walker:
If I can just jump in real quick, Bryan, on that, the industry can set all that up and work together to set all that up. Our customers need to think about how to buy that.
Bryan Clark:
Right.
Steve Walker:
And how to be flexible in their procurement practices in order to do that but also think about new business models that are. . . Especially with software because it allows the customer to buy things in order to upgrade much faster if they tweak the way they buy it.
Bryan Clark:
Right, that’s right.
Rand Waldron:
If you need a new version of something, and that’s an entirely new twelve-month acquisition, you’re dead. If you need a new version of something and that’s just a new version, so you upgrade and you upgrade and you upgrade, that’s how you innovate fast.
Bryan Clark:
Right. So you have to invest in the innovation pipeline as opposed to buying the product.
Rand Waldron:
That’s right.
Bryan Clark:
Right, right. So Dr. Wells, when it comes to how NATO looks at this. . . I mean the three initiatives you talked about in a way, though they’re their own innovation pipelines. . . That’s NATO investing in an innovation pipeline, but you’re hoping that the products of that spin out and turn into something that can be delivered at scale. I think one example where a lot of this work has been happening, and both Steve and Rand brought it up, was in unmanned systems or drones and how those have become for more or less the weapon of choice, the sensor of choice, in a lot of ways, the main tactical tool of choice in Ukraine. How do you see the NATO countries evolve in their approach to unmanned systems usage and moving it from just being the dull, dirty, and dangerous to a tool that could be used in any application?
Bryan Wells:
Thanks, Bryan. A great question, and as you rightly say, the experience in Ukraine highlights some specific examples that we can generalize into the use of autonomous systems in warfare and defense and security, more general please. So I’ll start with some elements on how NATO uses autonomous systems, and then I’d like to spend just a little while looking forward, where do I think the main scientific developments are going to be in the future. So again, let’s wind the clock back just to five years and the NATO summit that was held in London at the end of 2019. Heads of state of government of the Allied countries agreed a roadmap on what they called EDT’s or that stands for Emerging and Disruptive Technologies. And this identified seven of them of which autonomy was one, and big data and artificial intelligence, which I’ll come back to, were two others.
Following the agreement of that action plan, NATO agreed what it calls an Autonomy Implementation Plan, and this sets out the NATO policy regarding the opportunities and challenges provided by autonomous systems. And it made very clear that NATO’s use will be founded on the norms and values of the NATO allies and our partner nations and also in accordance with international law. And I’ll add some color to that. So in terms of the opportunities, NATO is very clear that it exercises autonomous capabilities with the aim of driving the innovation and development of these capabilities at appropriate scale. And of course, interoperability amongst allies and partners is a very important element of this for NATO. And regarding the challenges, NATO is looking to develop measures to prevent and defend against novel threats from potential adversaries and strategic competitors. And NATO is very well aware of the ethical, legal, and moral aspects of the use of autonomous systems.
The Autonomy Implementation Plan noted the two other emerging and disruptive technologies, big data and artificial intelligence, are key enablers of autonomy. And NATO had already developed six principles of responsible use for artificial intelligence and made clear that these six principles of responsible use would also underpin the use of AI-enabled autonomy. Examples of these principles of responsible use are, for example, lawfulness, reliability of AI, where I think we’ve all had experiences, but also bias mitigation. Again, something that we’ve all had experiences of. Regarding what are known as lethal autonomous weapons systems, NATO has made clear that it will align itself with the relevant international frameworks on this. In this particular case, it’s the convention on certain conventional weapons and building trust in the use of autonomous systems is an important part of NATO’s Autonomy Action Plan.
Turning now to a little excursion into the science. Where do I think the key scientific developments or some of the key scientific developments will be over the next 20 years? In March of last year, the NATO Science and Technology Organization published its science and technology trends 2023 to 2043. And this looked at our best judgements on the development of disruptive technologies over the next 20 years and what the implications for NATO’s allies and partners could be and to the extent that we could share it, what we thought the implications for our potential adversaries and strategic competitors would be. It’s a public document, it’s available to anyone online, and this was deliberate because scientists know that scientific progress is made not least in part by honest and professional challenge of judgements.
I’ll just highlight three points here. The first is the importance of human-machine interaction or human-machine interface. We’re seeing that autonomous and intelligent systems are already exceeding the capabilities of humans and this trend will only increase. Decision speeds will reach levels that will require changes in the method of human-machine interaction if there is to remain a human in the loop as our norms and values require. The second is in computer programming techniques. Competition between Alliance nations and adversaries will generate increased evolutionary pressures on these computer programs as each side is in a competition to reach a decisive victory. And the third is really space-based capabilities. We’re already seeing improved space-based capabilities improving terrestrial autonomous capabilities, particularly in swarms in the future. And these three points really illustrate my final point, which is combinations of emerging and disruptive technologies are bringing new strategic effects in autonomous systems.
This might sound daunting, but it actually means for policymakers we have choices. We can choose the ways in which we achieve military ends by different means. And as I said earlier, NATO is very clear that it will use these technologies in ways that conform to the ethical, legal, and moral norms of the allies and partners. And this in turn impacts on the sort of image that NATO must bring to its public.
Bryan Clark:
Thank you. Thanks, Dr. Wells. So Steve, this raises a question of. . . You could see the developments on the battlefield such as in Ukraine progressing faster than NATO policy and rules can catch up.
Steve Walker:
Sure.
Bryan Clark:
Is there a way to align new unmanned system applications to use cases with the rules set that we may want to pursue as we get into a fast move-counter-move cycle with an opponent?
Steve Walker:
Sure.
Bryan Clark:
How do we do that?
Steve Walker:
I mean, there’s AI machine learning.
Bryan Clark:
Right.
Steve Walker:
There’s an ability to, especially on the expert system side of AI, to build in rules, right? "This will do this, it won’t do that." I personally think that as we look at more autonomous systems and at scale, the human is going to have a hard time being able to control a swarm of a thousand UAVs out there. So how do you build that swarm into the overall C2 system that you have? Because that’s going to be important if you’ve got manned systems in the mix, one thing we’re spending some investment in trying to figure out. I mentioned our work with commercial partners. We did a really successful experiment recently with IBM Red Hat, looking at building an edge compute capability for doing AI and machine learning right on the platform of a Class 2 UAV.
And we’re able to do that and we did that because we want that UAV to be able to go up with other UAVs and if something changes on the battlefield, to be able to adapt real time. The human may not have time to say that, "Our blue UAV just got shot down. I want you to go over here," maybe, but this stuff’s going to be at machine speed if we want to win. So a really successful experiment showing we could do the AI machine learning on the platform, on the Class 2 UAV, and adapt in real time when the situation changed. But as long as you have a rule set around that, "You can operate in this region, you can’t operate over here," I think that’s acceptable.
Bryan Clark:
Right. And Rand, the idea of AI being incorporated into warfare, I think there’s a lot of idea. . . Obviously there’s a lot of examples out there with large language models, but there’s other forms of AI and Steve just brought up expert systems as one example. Do you see that the military applications of AI are going to tend towards different models rather than large language models, small language models, and expert systems and techniques like that?
Rand Waldron:
Yeah, so I think there is a universe inside military applications and so I think there will absolutely be applications for large language models, there will be absolutely applications. . . There are today, just as Steve was talking about, for taking smaller models, taking them onto a small edge device, and then that small model at the very edge with a very specialized purpose can ensure that only relevant data has to go back over denied communications because it’s doing some of that winnowing, that filtering, dare I say, some of that deciding at the very front edge of the battlefield rather than having to bring it all the way back, but all the way back in big data center applications, absolutely. We’re dealing with uncountable amounts of data, different types, different streams. Are we going to want large language models that help us summarize and help us find relevant parts, help us think about these data sets in totally different ways? Yes, we are.
And so the answer is that throughout that entire chain, from literally where we sit here in Washington DC to a NATO soldier, whether they’re in Mali or Helsinki, that entire chain, you are going to want AI. You’re going to want AI helping that soldier, helping the officers that give him orders, and helping the intelligence analysts that point them in the right direction and making sure that, as you. . . I talked earlier about buying that innovation pipeline. Making sure that through that entire chain you are thinking about that, you are not buying a platform that doesn’t let you add a little bit of AI, doesn’t let you innovate, but you are buying a capability throughout that entire chain that lets you add that AI over time. That is the only path forward today.
Steve Walker:
And you’re going to want that pathway that Rand just talked about to be done under a partnership of defense and commercial.
Rand Waldron:
Yep. Yep.
Steve Walker:
Because you need a certain amount of scale and you need a certain amount of understanding the warfighter and what he or she cares about and the safety issues associated with all that. It’s going to have to be a defense-commercial partnership.
Rand Waldron:
I don’t think there’s anything anymore that is . . . Okay, that’s not correct. There are plenty of things. There’s going to be less and less that is purely commercial or purely military or government. Less and less, every single day. You’re going to have systems, you’re going to have chains of systems that have commercial elements or have very specific military elements. Just specifically to us, we build cloud that is the exact same as our commercial cloud. We put it across an air gap and we put it as an underlayment for intelligence activities and military systems. Is that a commercial system or is that a defense system? Yes, it is, and that is the approach that we’re going to have to have throughout the entire spectrum of capability.
Bryan Clark:
Right. Dr. Wells?
Bryan Wells:
Thank you. And from a NATO perspective, we recognize this as well that our data board has not only allies, the governments, but also representatives from industry. NATO’s own science and technology board has representatives from government scientists, but also representatives from industry as appropriate as well.
Bryan Clark:
So before Steve, you grab that water. So Dr. Wells brought up this idea that space was one of the three areas he highlighted as where a lot of future science will be made or a lot of new technologies will be incorporated. And I don’t know, it may not be clear to the average policy person or citizen, but space is really becoming central to military operations today. It is the linchpin, it’s the central way that we get information, identify targets, and then act on those targets. Where do you see the commercial role in space and how do you see being able to protect those space capabilities that are now becoming so central?
Steve Walker:
Yeah, so you’re right, even five years ago people were talking about space becoming a war fighting domain. It is. Space Force was created because of that. I happen to be on the Defense Science Board too, so I get to see what’s going on there and I have to say that, and we’re going to have a report coming out or it just came out on commercial use of space in space. The Space Force is still very much trying to figure that out and architect a layered approach to space that includes commercial. I’ve obviously, we’ve seen in Ukraine the Starlink capabilities being used. Commercial GEOINT has been really important. So there are commercial capabilities that can be brought to bear now.
I think the recommendation of DSB is going to be yes and more. We want it, but we’re going to have to think long and hard about where we rely on it for war fighting and where we don’t. And I would just say it’s not a done deal yet about what that architecture should look like. But I think the bottom line is it needs to be there. I had the opportunity to work with SpaceX in the early days. They’re an example of a very successful commercial space company. Not all of them are. And so do you have government backups? How do you do that? It still remains to be seen.
Bryan Clark:
Right, right. It suggests that Space Force and the government and other governments will probably be relying more on commercial space.
Steve Walker:
Absolutely.
Bryan Clark:
And it’ll take a greater share of the communications load and the sensing load going forward. It would seem like it’d be irresponsible not to take advantage of it probably.
Steve Walker:
No, it’s going to be there and it’s for lots of reasons. But if you watch what’s happening in the U.S. at least, a space development agency is also building a proliferated LEO constellation a bit slower than Elon Musk, but those satellites may have additional capability. So the answer’s usually always and, and for lots of good reasons.
Bryan Clark:
Right. Rand, to what degree does our communication and our ability to have this capability delivery pipeline that you talked about, how dependent is that going to be on space and long haul comms that we maybe take for granted today?
Rand Waldron:
We can’t take them for granted. In a near peer conflict, space is going to be contested and we will all have to evolve quickly on that. And we can see recent incidents in the Baltic around how even long haul fixed communications can be contested accidentally. What you have to be able to do is build a set of technical capabilities that are tolerant of losing that connectivity. Maybe not permanently tolerant, maybe intermittent communications, but you have to build a set of capabilities that are tolerant of living that. You’ve got to be able to work at the edge with intermittent communication or no communication to the core. And that core has to be able to operate without connectivity to other pieces.
So you may have connection to very specific military networks, but you’ve got to be able to operate, should some of those networks go, should some connectivity cross and out of those networks go, you’ve got to be able to continue to operate. That is something that I’m not sure gets enough thought and effort in some ways. U.S. aircraft used to have little bubbles even in the jet age so that you could take a star chart and navigate. I think a little bit of that skill’s been lost and we don’t always think about what happens when a space and telecoms become compromised, but we need to think that way. We need to build our technology that is resilient to that so the mission can continue.
Bryan Clark:
So Dr. Wells, this highlights the dependence that we have on the electromagnetic spectrum and on undersea cables for that matter that carry a lot of these communications. And I think Brian just mentioned the case about the Baltic Sea where you’ve seen challenges with cabling being damaged. How are the NATO allies going about protecting access to the spectrum and also protecting these undersea cables while also being able to afford civilian users the access and the capacity they need to be successful?
Bryan Wells:
Bryan, that’s a great follow-on question. And controlling the electromagnetic spectrum is clearly vital for any aspect of a military operation in any domain where information dominance is required. And I think a current theme of this panel discussion is the vital importance of information dominance and the electromagnetic spectrum is increasingly congested and also contested by our strategic competitors and potential allies. I’ll just raise two points. The first is that NATO does recognize the importance of this topic. It has an electromagnetic warfare policy and a strategy to govern NATO’s use. And we do conduct training and exercises to ensure that we remain at the top of this particularly important competition. The second point I’d like to raise is that the science is continually developing to give the allies and partners new tools to maintain its dominance in the electromagnetic spectrum. This is an area where for various reasons, I’ll just give one example, and this is known technically as cognitive radar.
And cognitive radar is going to be a game-changer in terms of electromagnetic spectrum dominance. It allows in simple terms, artificial intelligence to automate the changes in the waveform shape that is used to communicate. And this will optimize the use of the electromagnetic spectrum that is available to the allies and partners. And this will greatly outperform the current radar technologies. There’s a policy implication here and that is that exactly the same technology will be available to our strategic competitors and our potential adversaries. So we need to be ahead of the game in order to maintain NATO’s competitive advantage. We quite often hear discussions about whether NATO is ahead or behind our strategic competitors, and we can easily get ourselves into a circle of despondency here. Now, I’m not in any way complacent as NATO chief scientist on any of the emerging disruptive technologies, but the one point I’d like to land with colleagues is let us remember some of our particular advantages.
And the one advantage I’d like to emphasize here is the excellence of the academic base in the NATO allies. I’ll just give you one statistic. The Shanghai rankings of the world’s universities was last conducted in August of 2023, so less than a year ago. And in the Shanghai rankings of world universities, of the top 20 universities, the top 19 are on the soil of NATO allies, and the 20th is on the soil of Switzerland, which of course is a NATO partner. So this isn’t just about the sciences, it’s about wider academia as well. So it nurtures the excellence of our scientists. It nurtures the excellence of our policymakers, of our politicians, and of our decision makers. Thanks, Brian.
Steve Walker:
I was just going to say, Brian talked about the warfighting capability really well, but there’s this commercial versus defense use of the spectrum, and I think certainly I think the U.S. and within NATO could work on together is being out in front of that. And I know in this country we have a moonshot program on dynamic spectrum sharing, and I think that’s going to be really, really important as we move forward.
Bryan Clark:
Absolutely. Because if you want to be able to afford 5G companies or telecoms, the ability to use the spectrum, you’ll have to give that ability to share spectrum either automatically or like we are today mechanically.
Steve Walker:
Absolutely.
Bryan Clark:
And something that Dr. Wells brought up, cognitive radar is something that I was also going to ask you and Rand about to put you on the spot. So when you have these AI enabled capabilities, so cognitive radar can be enabled by machine learning, how do you go about testing and evaluating those to make sure that they meet the specifications that the government or the safety rules that the NATO policymakers might put in place? Because by definition they’re designed to change how they operate depending on the stimulus they get or the environment they’re placed in. How do you manage the test and evaluation of these new AI enabled or machine learning enabled capabilities? So Steve, you want to jump on that and Rand?
Steve Walker:
Well, especially in the spectrum area with AI machine learning, you got to take it into the field because EW and it’s all dependent on the environment, and so trying to do it in a closed little facility is only going to get you so far. DARPA had a program in this space several years ago and we had to take it in the field and we had to show it on multi-platforms working in the relevant. . .
Steve Walker:
Had to show it on multi-platforms working in the relevant domain and show that it can work. But that’s the only way to do it. And again, that’s why I think it’s so important to think about how we bring the best of tech into defense application. It’s going to require more than just a single company to do it.
Bryan Clark:
Right. Waldron, did you have any thoughts on that?
Rand Waldron:
Yeah, I would just put two things on there. One, and it goes right off of what Steve was saying, is you have to test in reality. Testing in theory is nonsense. You have to test in reality. And for some systems that means out in the field; for other systems, that means against your own data, against real data.
An AI model that works well on the internet and works well for some generalized thing, you bring it up against your own data, who knows? You have to test it in the real world against your real data.
And so bring the AI models to where the data is. Bring the AI capability, the GPUs, whatever your limiting factors are in doing that analysis. Bring it to the data--thing one.
Thing two: almost countervailing is there are some things that it is very hard to simulate, right? There are some things that just do not happen enough in the real world to create a large enough corpus of data. . . Thank God, right? You don’t want those things to be happening a lot in the real world.
And so you’ve got to create simple crimes, you’ve got to create synthetic data. And again, then bring your model to that synthetic data, bring your AI capability to that synthetic data and iterate against it.
But always know that in the end, the map is not the terrain, the synthetic data is not the real world. You have to still be able to take your model out and run it against your real data with real customers, with real interaction, or out in the field.
Bryan Clark:
Yep. Yep. Thanks, Rand. So we’re going to turn over to some questions from the audience. So audience members have questions. We have the microphones over here if you want to go jump on there and use that. So just go over there. Thank you. And I have a few questions from the online audience as well, but I’ll turn it over to start. Please state your name and affiliation and then your question.
Klaus Hommels:
Yes. My name is Klaus Hommels. I’m chairman of the NATO Innovation Fund. And so we are intrigued by the opportunity and the necessity, but foremost, we do need to make sure that we develop the ecosystem as a whole. And as an ecosystem as a whole, we also make sure that the corporation with the primes work.
So when historically in the publishing industry, TV industry, banking industry, that has always been sent in the gearbox, which is not ideal, but it’s not an important industry as defense.
So we would not be keen on seeing this sent being, again in the gearbox, in the corporation with the prime. So what is the expectation from the primes to seamlessly work with. . . and from NATO, seamlessly work with the new technology startups? Thank you.
Bryan Clark:
Thanks, Dr. Klaus. I’ll start with you, Steve, and then Rand can answer.
Bryan Wells:
Sure.
Bryan Clark:
Oh, but please, yes. If you. . .
Bryan Wells:
Regarding what NATO is doing in terms of making the handshake between the SMEs and the primes, yes, we recognize from national experience of innovation accelerators, the SMEs can have an idea, but in order to start to mature it at scale, you need the handshake with the primes.
DIANA is already getting a system of mentors in place, especially to make that sort of handshake and to assist the SMEs in the other changes to their business model that they need to do if they are to mature their technology. Thank you.
Steve Walker:
I think it’s. . . Hi, how are you? I think there’s a couple of things. As I mentioned, Lockheed has a venture fund, so we try and reach out and look at. . . And by the way, it’s not just US. We’re looking internationally as well. So working with other countries to see how we plug into their ecosystems as well. So that’s one way.
Some initial startups, the sand in the gearbox can come in when we’ve got a really interesting technology that we’re funding at a small company, and we got to convince our own business units to say, "Hey, this is really good stuff and it’s going to have an impact, but it might be five years from now, or six or seven."
So that’s partly my job, is to convince them that that’s worth the investment, not only of money, but of time, to pay attention to it. And as you know, I’m trying to do that with somebody right now. So that’s part of it.
But I also think it goes back to the customer side as well. So I think, and I’m saying this as a former customer, but customer encouraging those relationships between small and large, so far, what I see mostly is opportunities for small companies to get defense money.
But it would be more effective if they put solicitations out there for small and big companies to work together, commercial and defense, to work together and help remove the sand, help the transition happen within the industry, which helps them in the end. So I think there’s something to be looked at on the customer side as well there.
Rand Waldron:
I would just say Oracle embraces this space completely. From the perspective of a small company, it is really hard to interact with some of you as customers. I mean, I’ll just say it, it’s hard.
If you’re a small company, how do you operate across an air gap? How do you get the security accreditation to be able to operate on a classified network? How do you hire enough cleared people to provide a 24 by seven watch in case your thing stops working at Saturday at 2:00 AM?
What we try to do is provide that underlayment. "Hey, we have a large technology infrastructure that is accredited. It has the accreditation." "Hey, we have a 24 by seven watch staff that can at least do monitoring and call you if something goes wrong." "You don’t have to invest in your own sets of GPUs because we have an infrastructure that’s there."
The things that startups have take for granted in the open world, "Hey, I can just go to the cloud and spin up dozens of servers of different compute types and have AI services available to me, and I don’t have this big hurdle of buying equipment and getting security accreditation, all that."
Our objective is to provide that in the defense space as well. That same ease of use, that same rapid spin up, why isn’t it on a classified network? It should be. Why isn’t it on a restricted, not classified, but highly sensitive network? It should be, and it can be. And that’s the approach we take.
Bryan Clark:
Thanks, Rand. That’s very good point. So I think we have time for one question from the online audience. So I’ll pose this to you, Dr. Wells. What role can NATO play in deterring knowledge and technology transfer to adversaries which might occur in academic and research settings?
Bryan Wells:
Great question, and it’s something that is at the forefront of my mind. And what NATO is doing through the Science and Technology Board is having a series of workshops on national best practice of the allies and also partners. A number of allies: United States, UK, and a number of others have procedures in place, some with regulatory and legal force on the transfer of knowledge from allies to our strategic competitors.
In particular, NATO is a great convening forum as we’ve seen with the public forum, where allies that do have these regulations in practice can share in a suitably protected environment its own regulations, what works, what in confidence with close allies has not worked.
And what my team will be doing over the summer is producing a compendium of best practice that can be shared amongst the allies so that those nations that don’t have such regulations in place can start to learn of the experience of others and think about how best to place those within their own national context.
Different nations have different societal norms in terms of defense and its relationship with universities. It’s for individual allies to come to their own decision on how best to adopt guidelines and how best to move that forward.
I should say that I’m also in discussion with the chair of the DIANA Board of Directors who has exactly the same concerns about protecting the knowledge that is created by DIANA.
We’ve also been in discussion with the European Union as well, and I’m sure colleagues from European Union member states will know that the European Commission has now produced its own guidelines on this area. So we’re all moving in the same direction.
Bryan Clark:
And I assume that that also means it allows for maybe a comfort when academics are maybe less willing to share information with military officials. Having that kind of framework also maybe opens up those channels that-
Bryan Wells:
Of course.
Bryan Clark:
. . . wouldn’t be otherwise.
Bryan Wells:
Of course.
Bryan Clark:
Well, thank you very much, and we’ve run out of time. So before they start playing the music and making us leave, I just want to thank our guests. So Rand Waldron from Oracle, Dr. Steve Walker from Lockheed Martin, and Dr. Bryan Wells, the NATO chief scientist. So join me in thanking them. We’re going to depart so that Admiral Bauer and Heather Conley can come and wrap this up. Thank you very much.