Abstract
The paper presents a theoretical argumentation for AI governance through the lens of Luhmann’s social systems theory. After a historical review of AI, ethics, systems and sociological studies, Luhmann’s concept of expectation is proposed as an explanatory variable that has not being considered in AI research. Further exploration of the term through a thematic analysis highlights the increased risk of uncertainty provoked by a constant process of anticipation and advises the contingency of AI by its own kind.