Summary: Intelligent systems no longer simply execute instructions. They learn, recommend, and explain themselves, while authority and accountability remain with a named human being: the captain, the surgeon, the controller, the engineer. Speaking as both a human factors academic and an active airline captain and examiner, Dr. Ziakkas opens on a long-haul flight deck at 02:40, where a fatigued captain has ninety seconds to accept, reject, or interrogate a decision aid's recommendation before the decision makes itself. From that moment the keynote argues that integration is something a person experiences in real time and under pressure, and that the measure of an intelligent system is whether its human partner can remain in command of it.

The address offers three answers. Oversight is a behavior, and it is real only when the person has time, independent information, competence, and the authority to disagree without penalty. Drift becomes more dangerous when learning systems absorb everyday workarounds and reproduce them at speed and scale, with the appearance of a recommendation — a process the speaker and his co-authors call Algorithmic Normalization of Deviance. Competence in overseeing intelligent systems can be defined, trained in simulation, and protected across a career. Extending these lessons from aviation to road transport, healthcare, and industry, the keynote closes with three commitments for the community: design for the disagreement, audit what the system has learned to call normal, and invest in the human as deliberately as in the model.