No matter what your ontological commitments happen to be, imagining what an LLM's umwelt might be like, and how it might approximate ours, is a useful (and fun!) exercise.