The US Army is developing its own AI-powered chatbot for soldiers to integrate artificial intelligence into the military chain of command and decisions.
In an interview with Alex Miller, the Army’s Chief Technology Officer wiredUnveiled a prototype of a system called Victor, which combines a Reddit-like forum with a chatbot called “Victorbot”.
Victorbot can be helpful to soldiers in many ways. For example, using chatbots soldiers can manage to configure electromagnetic warfare systems for a particular mission.
The model has been trained on information from ongoing conflicts, particularly the Ukraine conflict.
When a user submits a technical query, VictorBot generates a response citing specific posts and discussions from fellow service members.
“Electromagnetic warfare is a difficult topic. The chatbot can generate a response and cite all the lessons learned from (different) units,” Miller said.
Project Victor is currently under development within the Combined Arms Command (CAC). According to Lieutenant Colonel John Nielsen, who is monitoring this mission, soldiers make similar mistakes during different missions and Victor will fill that gap.
The ultimate purpose of Victor is multimodality, allowing soldiers to upload video and images for analysis. “Victor would be one of the only sources with access to official Army information,” Nielsen said.
According to Lauren Kahn, senior research analyst at Georgetown’s Center for Security and Emerging Technology and former Pentagon adviser, Victor has immense potential to automate “back-office” military administrative tasks.
The recent project is in line with the Pentagon’s broader goals regarding the integration of AI into the military. Last year, the Defense Department launched GenAI.mil to encourage AI adoption among the workforce.
When it comes to data, Miller said more than 500 repositories of data have been fed into the system, with the goal of building the same accuracy as commercial chatbots.
Can AI integration into the military be safe?
Some experts warned that the integration of AI models into military-based missions could create a variety of problems.
Paul Scharre, executive chairman of the Center for a New American Security and a former US Army Ranger, cautioned against AI sycophancy tendencies because they can be highly problematic in terms of intelligence analysis.
Furthermore, agentic AI adds a new layer of security risks, complicating the chain of command in military operations.
The Pentagon’s ambition to integrate and adopt AI on the battlefield is not new. Previously, the department had used cloud-like AI tools in recent operations, including the capture of Venezuela’s president and a strike against Iran. However, this is the first time that the Pentagon is building an AI tool on its own.
