The acceleration of artificial intelligence (AI) technologies has reached the tactical edge, where Department of Defense (DoD) leaders are utilizing generative AI (GenAI) and other AI applications to act on data generated where the mission is happening.
In a recent interview, Daniel Carroll, field CTO at Dell Technologies, discussed AI use cases in areas including signals intelligence and robotics – and how industry is working with military leaders to foster innovation at the edge.
MeriTalk: The 2023 DoD AI adoption strategy urged military leaders to “innovate at the edge,” and the Pentagon’s top AI official recently called for leveraging data in AI close to the tactical edge. In what ways are field commanders carrying out these directives and embracing innovation?
Carroll: To set the context, DoD has been using AI for decades. That experience informs today’s military leaders as they gain a greater understanding of where data is created, who has access to it, and how to categorize it – all of which is essential for moving forward with AI at the edge. The adoption strategy for the DoD faces the challenge of improving data quality, and AI solutions are helping with that, working through large data sets to help categorize and tag data to drive the next step into GenAI and higher-performing AI capabilities. Since the strategy came out, AI technologies have advanced so rapidly, especially GenAI. DoD can access data through more tools than ever, and it is up to us in industry to try to help in that process. That is just what we are doing.
MeriTalk: AI edge deployments allow data to be acted on near its creation point, so organizations can leverage AI where they need it. What are some future military applications for AI applied to data generated at the edge?
Carroll: A lot of people, when they think about AI and DoD, immediately jump to mission systems and weaponry. But the DoD is also the largest logistics, fleet management, and the largest IT organization in the world. All of those elements aren’t just being executed in garrison units and bases – they are being executed at the tactical edge, where the mission happens. So AI is having a huge influence today on how commanders look at their capabilities to improve responsiveness and better execute the mission. This applies to everything from logistics – tracking your food, medical and other supplies, and supply routes – to leveraging sensors on vehicles for fleet management and predictive maintenance. The growing importance of AI has led to numerous AI use cases for the military, with new ones coming at breathtaking speed. Among those use cases are signals intelligence, cybersecurity, robotics, autonomous vehicles, and digital twins. In the end, we expect that every device and piece of equipment will have some sort of processing and analytics capability at the edge.
MeriTalk: What challenges has the DoD experienced in moving AI to where the mission is happening, and how is industry helping?
Carroll: A lot of the challenges facing GenAI and AI in general are around power and cooling. Some Federal AI initiatives have even slowed down because data center power and cooling requirements were underestimated. This is especially an issue at the edge, where you don’t have an overabundance of power, and it isn’t always cool. The DoD is driving requirements back to industry to miniaturize equipment and reduce power consumption. Industry is also helping military leaders access faster and bigger computing infrastructure and train GenAI models. Once the algorithms are trained, they can be deployed to inferencing, which is another edge use case.
MeriTalk: A new partnership between Dell and NVIDIA, the Dell AI Factory, has been in the news recently. Does the AI factory have DoD applications at the edge?
Carroll: In the area of edge solutions, yes, the Dell AI Factory will definitely have an impact at the DoD. It is the industry’s first end-to-end AI enterprise solution that can help organizations explore the types of workloads and test capabilities to help them on their AI journey. There are requirements to provide this type of capability in data sensitive environments, and Dell is having discussions with various DoD organizations about working with the AI Factory in a closed security environment.
MeriTalk: Your role at Dell Technologies focuses on cybersecurity, and having confidence in security is a key factor in accelerating adoption of AI at the edge. What steps should defense organizations take to ensure that AI at the edge is secure?
Carroll: I am very passionate about cybersecurity, and because cyber is part of every DoD requirement, it should be a key part of every AI and edge solution. AI is driving investments into new infrastructure due to the compute limitation of existing environments, so this is a prime time to be implementing cybersecurity and zero trust best practices, because it’s all net new.
MeriTalk: Given the breathtaking progress of AI and edge technologies in recent years, let’s project ahead. Where do you see AI at the tactical edge capabilities being in five and even 10 years from now?
Carroll: Looking ahead, some AI systems like real-time language translation that takes into account local dialects and accents are on the cusp of taking off. Heads up Displays (HUDs) exist in limited use today in the DoD and will expand as technology and AI mature to provide more real time assessments of the environment troops work in to achieve the mission. Robotics has also accelerated tremendously – everybody has seen these little robot dogs that carry workloads. They’re only going to get smarter and be able to free more people to focus on the wider mission space. In the future, I think we will see great expansion of agents and copilots – basically, everybody will have their own robot to help with physical tasks, along with their own co-pilot to help them talk to their data in real time. I know that sounds like quite the vision. We’ll see if it becomes reality.