MIT's latest breakthrough? Getting AIs to explain their decisions

MIT researchers have figured out a way for neural nets, which are typically black boxes, to reveal the rationale behind their decisions.