Tim and Chris discuss the concept of representing infinite objects in neural networks, debating the feasibility of mapping infinite states and the constraints required for computational usefulness. They delve into the challenges of achieving compositionality and systematicity in neural network models.
