There's some assumption here that there's a reason for the account blocking, and if only a human could be reached, it could be explained in detail. But that may

There's some assumption here that there's a reason for the account blocking, and if only a human could be reached, it could be explained in detail. But that may not be true. ML systems usually don't explain themselves by default; it's possible there's simply no further information available for a support agent to give. This is a fun talk about post-hoc explanations from ML systems. It's possible, but it takes real work: https://www.thestrangeloop.com/2017/just-so-stories-for-ai-explaining-black-box-predictions.html https://www.thestrangeloop.com/2017/just-so-stories-for-ai-e...