Air Force, Space Force unveil tool for AI experimentation

The Air Force and Space Force launched a generative AI tool on Monday, encouraging airmen and guardians to experiment with using the technology for tasks like summarizing reports, IT assistance and coding.

The services wish to use the tool, which it’s calling the Non-classified Web Protocol Generative Pre-training Transformer, or NIPRGPT, to raised understand how AI could improve access to information and to gauge whether there’s demand throughout the force for the aptitude.

Alexis Bonnell, chief information officers and director of digital capabilities on the Air Force Research Laboratory, said the overarching goal is to make data more accessible and customizable throughout the services and to determine whether generative AI can facilitate that.

“Our goal is at all times to have the opportunity to say what technologies are relevant to our mission now and in the longer term,” Bonnell told reporters in a June 10 briefing. “NIPRGPT will really provide a spot where people can exercise that curiosity in a secure and appropriate manner, where they will understand what parts of their work can potentially be complemented.”

The Defense Department has been exploring how it’d use generative AI tools like ChatGPT to make each day tasks like finding files and answering questions more efficient. The Navy in 2023 rolled out a conversational AI program called “Amelia” that sailors could use to troubleshoot problems or provide tech support.

Collen Roller, senior computer scientist at AFRL, said within the briefing that his team on the lab has made a concerted effort lately to research how the Air Force and Space Force might use the technology for administrative tasks, but in addition for tactical operations.

“The world’s changing so rapidly and fast, now we have to have the opportunity to adapt to those recent things which might be coming out,” he said. “It’s super necessary from a [research and development] standpoint that we’re capable of adapt to whatever’s coming out in order that we are able to evaluate this stuff for our specific use cases.”

AFRL developed NIPRGPT using publicly available AU models and Bonnell noted that the service hasn’t committed to a specific approach or vendor because it builds on that baseline. As airmen and guardians begin using the system, AFRL will work with industrial partners to check and integrate their tools and determine whether or not they have utility for the services.

“We’re hoping that not only will this kick off the curiosity and experimentation that we are able to see in our users, but it should also, for those providers which have models, it should give us a option to actually test those,” she said. “We fully expect that some models are going to be great at some use cases and never so great at others.”

Together with helping firms experiment with different tools and models, the trouble can even help the Air Force and Space Force determine the perfect approach for getting theses capabilities, Bonnell noted. The suitable strategy, she said, will likely depend upon how the services use NIPRGPT and whether there’s sufficient demand.

“I expect the interest to be robust and hopefully to have the opportunity to drive lots of our learning in a really quick way,” Bonnell said. “This tool helps us understand what we would like the tip state to appear to be. And in order industrial tools come down and navigate our process or our system or security flow, then we’re all of the more smart once we buy them.”