Artificial intelligence (AI) and machine learning (ML) are transforming the U.S. defense sector, but adversaries are aiming to outpace with their own technologies.

Lockheed Martin is working to stay a step ahead with its 21st Century Security vision, say Rod Makoske, chief engineer and senior vice president of Engineering & Technology, and Steve Walker, Ph.D., vice president and chief technology officer.

1. First things first: How does AI/ML fit into Lockheed Martin's 21st Century Security® vision, and why is it so critical for defense?

Walker: The global security environment is increasingly complex. Near-peer rivals are developing capabilities to disrupt or deny the ability of U.S. and allied militaries to prevent or intervene in global conflicts. Success depends on dominance in all domains: air, land, sea, space and cyber.

• We need to connect systems and platforms across multiple domains at the same time. That's going to require new ways of doing business and new technologies.
• AI/ML are key for accelerating militaries' ability to sense, make sense and act faster than an adversary. There is deterrent value when potential adversaries know U.S. and allied militaries have this advantage.

When we talk about 21st Century Security, we're talking about digital technology. Lockheed Martin has always been a leader in building hardware and platforms, but several years ago we committed to further developing this crucial digital side of things.

• Specific technologies include resilient communications, Intelligence Surveillance and Reconnaissance, persistent surveillance and AI/ML.

In addition to the capabilities that AI/ML provide to customer militaries, we are focused on how AI/ML can transform how we design, test, build and sustain programs.

The Lockheed Martin AI Center (LAIC) is making AI a real engineering discipline at Lockheed Martin — because in order to use it, you have to understand it, and you have to apply it while maintaining sound engineering practices.

Makoske: Anyone can do what they think is AI, but you must have a trusted environment and the test capability to make it explainable.

We've invested hundreds of millions of dollars in this infrastructure and the test capability to be able to integrate the whole warfighting environment. Then all the data that we have is used to train those AI technologies.

We use AI for a broad range of capabilities like autonomy or command and control.

• As an example, using AI to analyze radar data from Houthi missile attacks since last October, the U.S. Navy and Lockheed Martin have made software changes to the Aegis Combat System which results in significant improvement in enabling destroyers to shoot down UAVs.
• These Aegis In-Theater and Crew Performed Updates, which rapidly field small changes, have increased capability and helped counter drones.

Because of the depth and breadth of our systems expertise at Lockheed Martin, we are uniquely positioned to provide these capabilities and how they interact.

• We not only provide these capabilities for the AI tech we produce, we also provide the mission integration environment that enables our customers to integrate AI solutions from other providers.
• 21st Century Security is about making connections across everything and making more out of what you have — not just faster, better platforms. AI makes these connections happen.

2. The strategy: Can you talk about what Lockheed Martin's key strength is for AI/ML?

Makoske: We're able to scale AI to meet any mission. That means training AI algorithms and having the data to understand what's happening and how to tune things.

• We've coupled that with investments in machine-learning operations (MLOps) infrastructure and warfighting simulation infrastructure developed over decades. We're integrating both of those environments together.

Let's say there's a simple AI technology that's autopiloting a Class 1 or Class 2 uncrewed aerial vehicle (UAV) — that's hard enough on its own — but then take a team of those.

• They're interacting with a crewed platform that's interacting with a ship, and that ship could be interacting with spacecraft, cyber or other AI-enabled technologies our militaries need in order to perform their missions.

The more complex it gets, the more you scale AI solutions. Having the environment to do trusted development and test becomes critical. I think that's a key discriminator in what we have to offer.

3. An example: What are a couple of AI/ML projects you're most excited about at Lockheed Martin? What kind of impact could they have?

Makoske: We're using AI across our internal workforce and our infrastructure including projects that leverage the latest advancements in AI. One example is a large language model we call Lockheed Martin Text Navigator that's hosted entirely on our secure Lockheed Martin AI Factory.

• Currently, more than 34,000 Lockheed employees are using it regularly on various tasks, increasing productivity.
• We're also asking ourselves: how do we do things twice as fast and half the cost? Right now, the long pole in a lot of that is testing and validation.

We're using AI to add in our simulations and add to our analysis. We're able to rapidly go through test data — something that previously may have taken weeks to do from a flight test. I'm talking about a model-based system that is transforming how we develop, test, manufacture and sustain programs.

• Now overnight, you have the answers — you know what your anomalies were and what to look for.

We have this set up in modular packages that we use across our business. We have a product we call ARISE which is a family of integrated toolkits used to build system-level weapon simulation — or "digital twin" — tools.

It has enabled us to reduce product development time, decrease cost and detect design problems before live flight tests.

Walker: I'm most excited about the human-machine interaction that's going to come along with this GenAI capability that we've all seen in the last year.

We're currently researching how to use a large language model in the planning phases of a battle to help decision-makers understand what's going on.

An operations center is very complex — lots of menus, typing and trying to figure out what's going on the battlefield. It takes time.

• These large language models will help interpret that for the decision-maker, which is why they have to be really accurate.

We're looking at how to input our own data into those models to make them less susceptible to "hallucinations," which are inaccuracies or falsehoods that could inhibit our ability to successfully use AI.

• As we do that, we can make models useful so a commander can understand what's going on and act quickly.
• Large language models enabled by GenAI are transforming what we do and how we do it at Lockheed Martin.

4. An expert take: Can you talk about why the ethical use of AI is so important and how Lockheed Martin is leading in this area?

Walker: The Department of Defense came out with five principles of AI ethics early on, and we were proud to be one of the first companies to adopt it:

1. Responsible
2. Equitable
3. Traceable
4. Reliable
5. Governable

Ethics is critical in our approach to AI. We are also partnering with government and industry to help shape the future of AI Ethics strategy and have developed AI ethics training for our workforce.

Former Secretary of Defense Ash Carter said, "technology is unstoppable, but it is shapeable."

• That's our view — this technology is going to advance, and we need to lead as a country, but we can shape it as we go. The ethical piece of AI is really important.

5. Looking ahead: What impact do you think the technology will have on the defense sector over the next five to 10 years?

Walker: It's going to change everything we do. I think it's going to help us make decisions faster. I don't envision — at least in our country — machines making decisions on their own.

• I think AI will be essential to help military members make decisions faster as we move forward. That's an advantage we need as a country. Almost everything will speed up.

Makoske: Unfortunately, our adversaries are doing their own work on AI. The threats will evolve at an exponentially increasing pace, and we're going to have to be able to keep up.

• Think about the impacts AI will have on cyber and the speed with which you can be attacked. I really think if you wanted to sum it up in one word across the board, it's going to be "speed."

Learn more about AI at Lockheed Martin.