I think most everybody hates this idea, including me.

However, I think the technology could be great for personal assistant apps, reminders/scheduling software, news feeds, and games/entertainment (especially). Anywhere a talking head could fit in a UI... as long as it's clear it's artificially generated. That is key.