Executive Summary : The MS&T visual simulation community has traditionally used standard 3D graphics pipelines in an effort to provide realism, driven by manual content workflows and human-defined shading algorithms. In spite of decades of technological development, and with significant improvements along the way, real-time graphics still fall far short of fully reaching real-world fidelity. As AI rapidly matures, new methods are set to enable significantly higher levels of realism than has ever been possible using traditional graphics methods, both for visible and non-visible spectrum visual simulation applications. This talk will discuss the current state of the art for AI world foundation models and where the technology is likely moving in the next few years.