Base concept is to place the system in a cognitive loop.

In very crude confines, a loop can also be defined as a "non returnable" outcome, such as a directive to calculate Pi to its' conclusion. Whereas a base loop would be to pose a question where the outcome triggers the initial question again. Example: "X = 6. If X > 5, (then X = 4 + 2)". Though an AI will need to be tricked, for simple variables as seen here are just a visualization tool, the logic, however, is the same.