Although artificial intelligence (AI) continues to make rapid advances in the automation of many analysis and operations tasks, the development and training of AI models typically require detailed AI engineering or research expertise. Once a model is created, it can be difficult to interpret or predict its behavior, leading to reduced trust or a reluctance to adopt AI in the first place. This webinar will discuss methods that adopt explainable, do-it-yourself AI (DIYAI) principles that address these problems. DIYAI enables end-users to rapidly build, test, deploy, and adapt customized AI solutions at the pace and scale of future conflicts, without any expertise in deep learning or programming, by employing interactive AI techniques. In DIYAI, users can interact with an AI model during training or refinement to customize the model to their needs and specific problems. One such method will be described, in which interactive query refinement (IQR) enables users to rapidly create deep-learning analytics to find salient items within large data archives while simultaneously creating training sets for improved classification accuracy.