“one of the essential insights gained by Turing as he observed his “children” was that if machines were ever to advance toward true intelligence, they would have to be fallible: they would need to be capable not only of error and of deviation from their original programming but also of random and even nonsensical behavior. Turing believed that such randomness would play an important part in intelligent machinery, because it allows for novel and unpredictable responses, creating a large variety of possibilities among which a search program could then find the appropriate action for each particular circumstance.”