Not really. It means that the agent never mathematically proves that it will make a decision before it makes it. The agent makes decisions by comparing expected utilities on different actions, not by predicting what it will do.