A more recent paper out of MIT pours cold water on that hyperbolic notion, drawing the conclusion that AI doesn't, in fact, hold any coherent values to speak of. The co-authors of the MIT study say ...