The decision-generating agent assigns a number to each problem (known as the "utility") that actions how much the agent prefers it. For every probable motion, it may work out the "expected utility": the utility of all attainable results in the action, weighted via the likelihood that the result will take place. It can then choose the action with th