How a computer system can learn

The process by which a computer can learn is demonstrated by asking it to solve increasingly more difficult versions of the Tower of Hanoi puzzle. Ultimately, the system may learn how to generalize in a particular problem domain Present uses of computers, valuable as they may be, are far from the ultimate in what might be accomplished. One of the reasons is that the solution of even well-defined problems, for which goals and rules are precisely known, can be extremely difficult to program (e.g., chess-playing programs). But intellectual capacities of machines might be extended by means of an adaptive system to handle increasingly complex and varied tasks.