When an AI system learns to make decisions, it becomes motivated to do so, but how does it determine what is best for humans? To answer this question, it should be able to understand the human mind and choose an incentive that suits that person's preferences. This can be achieved by using a hidden incentive. The goal of this type of incentive is to convince an AI to perform certain actions rather than other ones, which would otherwise be undesirable.

A major advantage of an AI is that it can help organizations avoid unintended consequences of reward programs. A big data system can reveal correlations in an AI's outputs, which enable designers to better anticipate the ripple effects of a program. For example, an AI may reveal that a person's recognition causes a negative J-Score for another. For example, an organization might announce the members of its annual Presidents' Club, and then watch the derivative effects of the announcement on job-seeking activity.