Since the Air Force and Space Force first deployed the generative AI tool in June, more than 80,000 airmen and guardians have trialed the system, according to the Air Force Research Laboratory.
The lab told Defense News this week that early adopters come from a variety of professional sectors and are using the tool for a variety of tasks, from content creation to coding.
The services are using the system, called Unclassified Internet Protocol Generation Pre-Trained Transformer (NIPRGPT), to better understand how AI can improve information access and whether there is demand for that capability among military personnel.
“Information gained from this research initiative will help us understand demand and identify areas where AI can save airmen and guardsmen mission time,” the lab, known as AFRL for short, said in an email. “This will allow us to better prioritize capability updates and security considerations as we expand enterprise capabilities related to generative AI in the future.”
The military has been exploring how generative AI tools like ChatGPT can be used to make everyday tasks like searching files or answering questions more efficient. In 2023, the Navy rolled out a conversational AI program called “Amelia” that sailors can use to troubleshoot problems and get technical support. The Army began experimenting with the Ask Sage generative AI platform earlier this year and announced this week that it has integrated the system into its operations.
Alexis Bonnell, AFRL’s chief information officer and director of digital capabilities, told reporters in July that the high usage of NIPRGPT validates that there is demand for generative AI tools within the Air Force and Space Force workforce.
But what’s even more meaningful, she says, is how military personnel are using the platform: Rather than simply using NIPRGPT to ask questions to find documents, most users use a framework called RAG (Search Augmentation Generation) to customize the tool for their role.
“To me, this means people can be more proactive and entrepreneurial, and they don’t have to wait for dashboards to be developed or major upgrades to the system,” Bonnell says. “In some cases, they can just bring relevant knowledge to the table.”
The pilot is also helping each service begin to develop its capability acquisition strategy. AFRL developed NIPRGPT using publicly available AI models, but is building on that baseline and isn’t committed to a specific approach or vendor. As the pilot progresses, the lab will work with commercial vendors to test and integrate tools to determine whether they are useful.
Bonnell said AFRL will continue to evaluate usage trends to understand what capabilities Airmen and parents want and what value those tools have to them.
“I think the ability to look at actual usage, to inform actual buying, and to be in a different position at the commercial negotiation table of evidence-based insights is going to be really important,” she said. “So when you consider this rich range of commercial tools, it allows for much more evidence-based and informed buying than throwing in AI.”
Courtney Albon is a space and emerging technology reporter for C4ISRNET. She has covered the U.S. military since 2012, focusing on the Air Force and Space Force. She has reported on some of the Department of Defense’s most significant acquisition, budget and policy challenges.
