<p>This reply to Dr Alžběta Kuchtová’s “The Question of Diversity of Data in AI Development” supports her critique of the epistemic limits and colonial legacies embedded in contemporary AI datasets. Expanding on her insights, I argue that true data diversity must be plural, contextual and philosophically grounded. Drawing on Derrida’s notions of the archive, trace and iterability, I examine how AI’s data practices reproduce colonial asymmetries and how contextual interventions—such as participatory data design, dynamic monitoring, and rich metadata documentation—can reorient AI development toward inclusivity. Integrating Kuchtová’s framework with deconstructive theory highlights that ethical AI requires continual interpretive vigilance, institutional pluralism, and humility toward the limits of knowledge. This dialogue underscores the need to transform AI from a universalising monologue into an ongoing, globally situated conversation.</p>