Tim and Thomas discuss the challenges of building generative models in machine learning, emphasizing the importance of minimizing surprise in model selection. They delve into the subjectivity and structure learning involved, highlighting the balance between prior knowledge and data-driven model building.
