Kari Bingen, Deputy Under Secretary of Defense for Intelligence, said today that the Defense Department’s (DoD) rapid progress on Project Maven relies on use of agile development principles, and help from the private sector on DevOps software development practices that combine software development and IT operations to shorten development lifecycles.
Bingen spoke at the GEOINT Symposium on Wednesday, June 5, about progress on Project Maven, which is using machine learning and artificial intelligence to find better ways to analyze drone camera footage and other imaging sources.
“We had a great team that was persistent … It was a mindset and a culture,” Bingen said in describing the pace of the project. “The department is embracing agile development,” and bringing in DevOps specialists from the private sector to work on Project Maven, she said.
“We also have an operational mission imperative,” Bingen said, in talking about the need for the project to move along quickly. A central component of the project aims to free up DoD analysts from examining images on computer screens, and instead allow them to undertake higher-value decision-making tasks.
On the subject of AI technologies that fuel Project Maven and other DoD work, Bingen told GEOINT attendees that the department has to “have confidence” that AI tech developed by the private sector is robust and will produce the right results.
“We want to understand what’s going on under the hood to make sense that what is coming out of the black box is what should,” Bingen said. She also warned that AI security is a concern, saying, “Sometimes it’s not hard to spoof an AI algorithm, so we have to work on that.”