Not yet, you'd need to apply the same approach to agentive (decisions on how to act) problems as opposed to classification tasks; then you'd have the spark that lights the google skynet powder keg.