Last month, OpenAI released a user study that showed nearly 15% of work-related conversations on ChatGPT pertained to “decision-making and problem-solving.” Now, it’s emerging that at least one high-ranking member of the US Army is using LLMs for exactly that purpose.
According to reports, Maj. Gen. William “Hank” Taylor said at the Association of the United States Army conference in Washington, D.C., “Chat and I have gotten really close lately,” using an annoyingly endearing nickname to refer to an unspecified AI chatbot. “AI is something that, as a commander, has been very, very interesting to me.”
Army-focused news site DefenseScoop reports that Taylor told a group of reporters that he and the Eighth Army he commands out of South Korea are “routinely” using AI to modernize their predictive analysis for logistical planning and operational aims. Taylor said that this is helpful for administrative tasks like “just writing weekly reports and those types of things,” but also helps inform their overall direction.
“One of the things I’ve been working on lately personally with my soldiers is decision-making—individual decision-making,” Taylor said. “And how we [make decisions] in our individual lives, when we make decisions, that’s important. So, it’s something I’ve been asking and trying to build models to help us all with. Specifically, [regarding] how do I make decisions, personal decisions, right—that impact not only me but my organization and overall readiness?”
This is still a far cry from the “Terminator” vision of automated AI weapons systems taking lethal decisions out of human hands. Still, the use of LLMs for military decision-making may give pause to anyone familiar with these models’ well-known tendency to fabricate bogus citations and obsequiously flatter users.
In May, the Army launched its organizational LLM workspace—which is built on the commercial Ask Sage platform—to streamline simple text-based tasks like press releases and personnel write-ups. For other military “back office” administrative tasks, early tests have shown that generative AI may not always be the most efficient use of military budgeting.
Click here to preview your posts with PRO themes ››
“Often, we find that folks are using the technology to answer something that we can just do in a spreadsheet with a math problem, and we’re paying a lot more money to do it,” Army Chief Information Officer Leonel Garciga told DefenseScoop in August. “Is the juice worth the squeeze? Or is there another way to get to the same problem that may not be as technologically interesting, but it’s more operationally practical?”
In 2023, the US State Department enumerated best practices for the military use of AI, which focused on the ethical and responsible deployment of AI tools in the human chain of command. The report emphasized that humans must retain control over “decisions on the use of nuclear weapons” and must retain the ability to “cut off or deactivate deployed systems that demonstrate unintended behavior.”
Since then, the Army has signaled interest in using AI tech in the field for everything from automated targeting systems on drones to “improving situational awareness” via OpenAI’s partnership with military contractor Anduril. In January 2024, OpenAI removed a ban on “military and warfare uses” from ChatGPT use policies while still prohibiting customers from “develop[ing] or us[ing] weapons” via LLMs.
