A framework that could improve the social intelligence of home assistants

Existing artificial intelligence agents and robots only help humans when they are explicitly instructed to do so. In other words, they do not intuitively determine how they could be of assistance at a given moment, but rather wait for humans to tell them what they need help with.