Top defense and technology companies are creating new battlefield simulations generated by powerful artificial intelligence models in hopes of stopping war from breaking out around the globe.
Lockheed Martin is assembling an “AI Fight Club.” Google is working toward creating digital twins of various battlefields that could be used, for example, to deter an invasion of Taiwan. These are just two of many AI projects in the pipeline from developers chasing large defense contracts in Washington.
New defense-tech partnerships took center stage at a major AI conference in downtown Washington this week, where national security officials and technology companies looked for new ways to adopt AI into military systems.
Lockheed is creating an AI Fight Club to create simulations for AI companies to test their models in simulations meeting the Pentagon’s standards for defense applications.
The fight club is designed to create a venue to prove that AI solutions work as advertised in the glossy marketing brochures floating around the Special Competitive Studies Project’s AI+ Expo, according to Lockheed’s John Clark.
Mr. Clark told the expo’s attendees the fight club would vet everything from small algorithms to weapons system-level concepts.
“With AI Fight Club, we’re going to be inviting anybody who wants to pit their AI up against the government environment with our native test and evaluation setup to determine how well their systems are performing,” Mr. Clark said on Tuesday. “A lot of these small companies that have great ideas and great AI but they don’t have the capital to support the full test environment and the full ecosystem that underpins it, they can now connect in and tie into our AI Fight Club environment and explore.”
Lockheed Martin told The Washington Times it is at work developing its initial fight club arena and plans to host its first battle in the final quarter of 2025.
Google is working with Lockheed on using generative AI for national security and is similarly interested in creating virtual battlefield scenarios.
Google’s Josh Marcuse, a former Pentagon official, said his company’s models are getting more sophisticated for making predictions and the Big Tech company is incorporating new physics and weather models.
He told attendees at the expo that Google is working to have “at least the ability to create a digital twin of the battlefield.”
“My hope is that the technology we are building will lead to greater peace and greater stability and less conflict because we will be able to consult those digital twins and be able to say, ‘We can predict with a high degree of accuracy that if you invade Taiwan, you will lose, and I have the receipts,’” Mr. Marcuse said at the expo. “‘And I hope that that is how we, as the free world, prevent great power competition from spilling over in these negative ways and I think the models we’re building hold a key to being able to unlock that potential in the future.”
Lofty aspirations of using AI to deter any Chinese invasion of Taiwan aside, the potential for new business has attracted many AI companies to rush to collaborate with national security officials and contractors working on hard problems.
AI company Anthropic and software giant Palantir have partnered with Amazon to deliver a family of powerful models to U.S. intelligence and defense agencies. The partnership, first disclosed last year, is far from the last of its kind.
OpenAI revealed at the expo this week that it is working toward big projects with both the U.S. intelligence community and the Department of Defense. The maker of the popular chatbot ChatGPT has already begun working with U.S. National Labs and with defense tech company Anduril.
OpenAI’s Sasha Baker, another former Pentagon official, said the company’s work with Anduril is intended to create a user-friendly interface for the military to deploy.
She said the recent delivery of the company’s model weights, or the highly sought-after parameters of its powerful models, to the labs’ classified setting felt like a scene from a “Mission Impossible” movie.
Ms. Baker told attendees it was not an easy choice for the company to hand over its “special sauce” to the U.S. government for the first time this week.
“We had a lot of internal conversations about should we do this, and is it safe to do this,” she said. “And as the sort of resident national security person inside the company, I raised my hand to say, ‘These are the guys that literally, they secure the nuclear weapons. I think if we’re going to do this anywhere, this is the place to start.’”
• Ryan Lovelace can be reached at rlovelace@washingtontimes.com.