Reinforcement learning is a satisfying way to create something special and rewarding, often involving building and re-implementing from scratch. Generative tasks tend to be even more gratifying due to their properties.