Postmodernism is perhaps the most frequently misunderstood intellectual tradition in the contemporary academy. It is regularly caricatured as the claim that nothing is true, that all perspectives are equally valid, that reality is merely a social construction. None of these caricatures accurately represents what postmodern theorists actually argued.

What postmodern thinkers did argue, with considerable rigour and force, is that the grand narratives of modernity are not adequate to the complexity of human experience, that claims to universal truth require careful scrutiny, and that knowledge is always produced in specific historical, cultural, and political contexts that shape what can be known and by whom.

Jean-François Lyotard, in his 1979 work The Postmodern Condition, defined the postmodern as incredulity toward metanarratives, scepticism about the large overarching stories that modernity told about human progress, rational knowledge, and universal emancipation. Lyotard was not arguing that progress, knowledge, and emancipation do not exist. He was arguing that the grand stories told about them systematically excluded the experiences of those who had not benefited from them, and that this exclusion was not incidental but structural.

Jacques Derrida, the founder of deconstruction, developed a method of reading texts, including legal, philosophical, and scientific texts, that attends to what they exclude and contradict as much as to what they explicitly claim. Deconstruction reveals that texts depend on binary oppositions, reason/emotion, objective/subjective, natural/artificial, in which one term is privileged over the other, and shows how these hierarchies are constructed rather than natural.

These are not arcane philosophical games. For AI, they have immediate implications. The binary oppositions embedded in AI systems, relevant/irrelevant, creditworthy/uncreditworthy, high-risk/low-risk, true/false, are not natural distinctions. They are constructed, and their construction reflects particular values, interests, and power relations. Deconstructing them reveals what they exclude and whose interests they serve.

Postmodern theory is also not relativism. The rejection of absolute objectivity does not imply that all claims are equally well-supported or that evidence is irrelevant. It implies that the standards by which claims are evaluated are themselves constructed, and that these standards can be examined, questioned, and improved. For AI, this means examining not just whether a model is accurate, but accurate according to which definition, on whose data, evaluated by whose standards.

Reflection question: What grand narrative does your organisation tell about its AI systems? What does that narrative exclude or downplay?