U.S. Air Force Cadet Builds Battlefield AI App Using ChatGPT, Claude and Gemini

U.S. Air Force cadet Joshua Lynch worked with MIT Lincoln Laboratory staff to test whether a novice coder could create a functional military application using only generative AI chatbots. Relying on a method he called “vibe‑coding,” Lynch prompted ChatGPT, Anthropic’s Claude and Google’s Gemini to write and refine code through a web‑browser chat interface. Over three months he assembled the Remote Operating Modular Augmentation Device (ROMAD‑AI), intended to aid target recognition, ISR, autonomous striking and battlefield communications. The project highlighted both the potential of AI‑assisted software development for non‑technical users and practical challenges, such as the models’ lack of hierarchical focus and tendency to stray from the task. Lynch learned to break problems into small parts, ask clear questions and steer the conversation back on track, ultimately delivering a prototype that demonstrates how AI could accelerate military software creation.