How to Build an Ethical Framework for Algrythmic Explainability?

Under what circumstances will we require an algrythm to provide an actual explanation, like the decision tre these are the factors from your specific data that led to this specific outcome? Toto, the extent that we you may know that's a traide off between explainability and things like precision and accuracy in things like that. If i can't do that and i want to have the less explainable or non explainable agarithms, when will i require interpret ability? Say, ok, in these situations, you must have interpretability, even if you don't have explainability, right? What are those circumstances? And what approaches and methods will i use for explainability