This paper addresses the increasing number of AI paradoxes that appear in the academic literature. It appears that research in this area is still in its infancy, with a great deal of conceptual confusion still prevailing. The objective of this paper is to collect the AI paradoxes that appear in the literature, systematically assign them to the phases of the AI life cycle, and finally assess whether they are actually paradoxes or rather other forms of contradictory decision situations. The findings of this study provide a foundation for the precise regulation of terminology, which will facilitate further research in this field.