The US military is revamping its testing procedures to accommodate the rapid development and deployment of artificial intelligence (AI) and autonomous systems. According to retired Air Force Lt. Gen. John "Jack" Shanahan, the Pentagon's first AI chief, traditional testing methods won't cut it for AI systems that require continuous updates.
AI systems need to be updated regularly, sometimes in hours or days, which is a significant departure from traditional software development. This means that testing can no longer be a one-time event, but rather an ongoing process that occurs throughout the system's life cycle. Ensuring seamless integration with other autonomous systems and human operators is also crucial.
To address these challenges, the military is exploring new approaches. One such approach is decentralized testing, which allows unit-level testing and evaluation, rather than relying on centralized facilities. The military is also creating specialized training programs for personnel to develop expertise in AI testing and evaluation.
The Air Force is already implementing these changes, with initiatives like the VENOM program, which aims to develop autonomous combat flying technologies. The Navy is also working on integrating AI into its operations, with plans to develop "interchangeable" collaborative combat aircraft (CCA) drones with the Air Force.
As the military continues to adopt and integrate AI into its operations, it will be essential to develop and refine testing procedures to ensure the safe and effective deployment of these systems. The goal is to create a more agile and responsive testing framework that can keep pace with the rapid development of AI technologies.
