This seems like a false dichotomy. There is nothing preventing a human staff from declining to provide an explanation, nor is there any reason a machine learning algorithm couldn't summarize its reasons.

The issue isn't AI vs human, it's transparent vs opaque.