This short paper summarises some of the work presented in our Keynote on how Generative AI and Reinforcement learning can enable robots to face unforeseen situations like mechanical damage and autonomously adapt during their missions. In particular, we introduce a family of Generative AI called Quality-Diversity algorithms that are well-known for generating thousands of diverse and high-performing solutions to an optimization task. This diversity of solutions provides robots with an extensive set of alternative options to face unexpected situations. We also present how Quality-Diversity can be paired with Deep Reinforcement Learning to learn more complex policies, or with Generative Dynamics Models to ensure fast, safe, and continual learning and collection of data. Finally, we present how these tools can address Challenge 3: Adaptive Learning for Evolving Drone Operations.