Postmodern theory gives us essential tools for interrogating the claims AI systems make about truth, objectivity, and knowledge:

Postmodernism is not relativism. Incredulity toward metanarratives, deconstruction of binary oppositions, and attention to power-knowledge are analytical tools, not denials of truth or evidence.

Foucault’s power-knowledge relation shows that AI expertise is simultaneously a form of power, defining what counts as a problem, what counts as evidence, and who has authority to speak.

Discourse shapes what is thinkable. The dominant discourses of AI, techno-solutionism, safety concern about superintelligence, ethics as technical fix, determine what questions are asked and what remain unasked.

AI systems construct social reality. Classification systems produce the categories they appear to describe. Looping effects mean that algorithmic classifications change the people and phenomena they classify.

Explainability has limits. Transparency is necessary but insufficient. The question of justice cannot be resolved by the question of explanation. Contestability, the power to challenge not just how a decision was made but whether it was legitimate, is also required.

Big data is a Foucauldian system of power-knowledge that defines legible subjects, enforces norms, and extends disciplinary power into new domains.

AI ethics frameworks embed assumptions about the individual, technical solutions, and universal values that deserve deconstructive scrutiny.

Postmodern humility, situated ethics, epistemic modesty, plural and provisional design, genuine contestability, offers a constructive response to the limits postmodern theory reveals.