The US Army Developing a Combat Chatbot

The US Army is working on AI models based on data from actual missions, aiming to launch a chatbot designed specifically for soldiers.

“We have accumulated numerous lessons from missions like the Ukraine-Russia War and Operation Epic Fury,” states Alex Miller, the Army’s chief technology officer, in a conversation with WIRED. “There is a vast amount of knowledge at our disposal.”

Miller presented WIRED with a prototype named Victor, which merges a Reddit-style forum with a chatbot called VictorBot to assist troops in accessing pertinent information, such as optimal configurations for electromagnetic warfare systems tailored to specific missions. When a soldier inquires about setting up their equipment, VictorBot generates a response and directs them to relevant posts and comments from fellow service members. “Electromagnetic warfare is incredibly complex,” Miller explains. Victor, he continues, “can formulate a response and reference all the lessons learned from [various] units.”

In the last two years, the Pentagon has intensified its initiatives to integrate AI into military systems, yet Victor stands out as a unique example of the military developing its own AI. This project illustrates the US military’s eagerness to master the intricacies of AI and suggests that this technology could significantly alter daily operations for many troops.

Miller indicates that the Army is collaborating with a third-party vendor responsible for running and refining the AI models behind Victor. He withheld the name of the specific company as the contract is still pending announcement. Over 500 repositories of data have been integrated into the system, and Victor aims to minimize errors similarly to commercial chatbots by citing reliable sources.

The push to implement AI into military systems gained momentum after the launch of ChatGPT in 2022. Recently, Anthropic’s technology reportedly played a crucial role in operational planning in Iran through a system driven by Palantir.

As these systems have become more sophisticated, however, conflicts have arisen over the appropriate applications of AI. Earlier this year, Anthropic clashed with the Pentagon, asserting that its technology should not be utilized for autonomous weapons or the surveillance of American citizens.

Same Mistakes

Victor is being developed within the Combined Arms Command (CAC). Lieutenant Colonel Jon Nielsen, who supervises the CAC’s efforts on Victor, notes that it is common for different brigades to repeat the same mistakes across various missions. The aim with Victor, he adds, is to ultimately create a multimodal system where soldiers can input images or videos and receive insights. “Victor will be one of the few sources with access to official Army information,” Nielsen remarks.

Lauren Kahn, a senior research analyst at Georgetown’s Center for Security and Emerging Technology and a former policy adviser for the Pentagon, states that project Victor underscores AI’s potential to automate numerous mundane back-office tasks within the Department of Defense. Late last year, the department launched GenAI.mil, an initiative intended to encourage greater AI adoption among DOD employees.

If Victor succeeds, however, Kahn believes the Army may eventually partner with a prominent AI company to enhance the system’s capabilities. “The leading labs undoubtedly possess a comparative advantage” in terms of creating and deploying cutting-edge AI, she adds.

Intel Failures

AI could present new challenges for militaries, warns Paul Scharre, executive vice president of the Center for New American Security and a former US Army Ranger. Scharre points out that AI models tend to be overly agreeable, which could be particularly troubling. “I could foresee scenarios where that would be especially concerning in the realm of intelligence analysis,” he explains.

Scharre further notes that as AI evolves from chatbots to agents capable of utilizing software and computer networks, the complexities of adoption may increase. “Agentic AI introduces an entirely new array of security challenges,” he observes.