LLMs fail on basic physical reasoning (e.g., whether to walk to a car wash 100m away) and exhibit no real understanding of physical intuition, instead performing retrieval-based responses fine-tuned on human feedback.