UK officials use AI to decide on issues from benefits to marriage licences

> Is there any reason to believe in this case that decisions supported by AI, traditional or not, can be rationalised by humans?

That's a formidable question. There are many angles to the question, but in this context, it should be underlined that administrative due process is cast in stone in most Western countries. Therefore, the explainability, transparency, and liability issues of AI/ML are particularly severe in the public sector.