Insufficient data volume and quality are particularly pressing challenges in the adoption of modern subsymbolic AI. To alleviate these challenges, AI simulation recommends developing virtual training environments in which AI agents can be safely and efficiently developed. Digital twins open new avenues in AI simulation, as these high-fidelity virtual replicas of physical systems are equipped with state-of-the-art simulators and the ability to further interact with the physical system for additional data collection. In this paper, we analyze trends and techniques in digital twin-enabled AI simulation; derive a reference framework to situate digital twins and AI with respect to each other; and identify challenges and research opportunities for prospective researchers.