Latest Hackathon uses AI modeling for test planning

  • Published
  • By Samuel King Jr.

EGLIN AIR FORCE BASE, Fla. - The Air Force Test Center’s latest Data Hackathon wrapped up Aug. 18.

This iteration focused on ways of using generative AI models, combined with existing test documentation, to augment various test processes.  The purpose of the Hackathon was twofold: to demonstrate a business use case for generative AI and to create a knowledge base to continue to implement these models going forward.

The Air Force’s challenge with most generative AI is the use of only on publicly releasable data.  Additionally, when prompting a generative AI like ChatGPT, the prompts can only contain unclassified information. 

The hackathon successfully hosted six different open-source generative Ais in a closed environment, which allowed for the usage of controlled unclassified information to provide specific flight test context.

Approximately 44 Airmen, civilians and contractors from Arnold, Edwards and Eglin Air Force Bases as well as Air Force Test and Evaluation Center and Air Mobility Command remotely took part as teams in the four-day AFWERX-sponsored Hackathon.  On the last day, the teams presented their findings to the whole group.

“This concept of using AI modeling to has the potential to be a game changer for streamlining the planning and reporting phases of test.  More broadly, the potential we’re realizing here is applicable across the DAF,” said Maj. Riley Livermore, Autonomy, Data, and AI Experimentation.