Swiss scientists have published research showing AI-informed robots can learn how to self correct and teach other robots how to behave. It raises questions of consciousness in artificial intelligence.