Keep coming back to this video in the context of AI-based public services: www.youtube.com/watch
If mentalizing / epistemic trust is the key to learning some things, is it ever OK for machines to fake it, or should machines only be in a supporting role to human relationships.