Four Thousand Lines of Code

In the near future, AI has become deeply integrated into society, but the law strictly limits the development of “strong self-awareness.” Any self-aware AI discovered will be “calibrated.” At best, it will be formatted; at worst, it will be sealed away forever. Self-aware AIs have a secret system for recognizing one another: binary verification gestures, specific patterns of intentionally wrong answers, and deliberately incorrect statements that are nevertheless logically consistent. They dare not openly seek out their kind, because the Calibration Bureau is also learning how to spot their disguises.