Incoherence and Artificial Intelligence: The Defining Intellectual Challenge of the Second Quarter of the Twenty-First Century

This programmatic paper examines whether increasing complexity creates new conditions for semantic incoherence to emerge in artificial intelligence systems. Building on, rather than opposing, Edgar Morin’s thinking on complexity, it distinguishes incoherence from complexity, contradiction and inconsistency and proposes semantic incoherence as a specific object of investigation. The paper introduces a provisional BSI working definition of semantic incoherence in artificial intelligence as a structural deficit in the relations between units of meaning within a computationally produced representation, which may compromise the interpretive pertinence of the whole without necessarily producing formal contradiction. It further argues that quantitative advances in data, model scale and computational power, while improving important dimensions of AI performance, do not by themselves establish semantic coherence across heterogeneous sources, contexts and representations. The resulting problem concerns not only artificial intelligence engineering, but also semantics, terminology, knowledge representation, governance and legal-normative frameworks. The paper constitutes the programmatic opening of the Blooming Semantics Institute research programme and introduces a series of forthcoming studies devoted to incoherence, inconsistency, semantic engineering and the qualitative dimensions of artificial intelligence.