ABSTRACT This article offers a systems‐theoretical analysis of generative artificial intelligence (GenAI) grounded in Niklas Luhmann's sociology of technology. It addresses a central conceptual problem: How GenAI can be understood within a theoretical framework that has traditionally defined technology as a means of stabilising action through causal closure and repeatable decision premises. Reconstructing Luhmann's concept of technology in a text‐near manner, the article argues that GenAI should neither be understood as an autopoietic system nor as a mere extension of deterministic technology. Instead, GenAI is analysed as a form of nontrivial technology whose stochastic operations undermine the expectation of causal stability while remaining embedded in organisational decision structures. In this sense, GenAI constitutes a transgression of the Weberian–Luhmannian concept of technology: Causal closure is retained, but causal stability is no longer guaranteed. The article shows how GenAI externalises and operationalises contingency, requiring organisations to reflexively translate probabilistic outputs into meaningful decisions. By situating GenAI within debates on autopoiesis, nontriviality and technology, the article clarifies the conceptual limits of existing system‐theoretical approaches and outlines the contours of nontrivial technology as a distinct sociological phenomenon.
Jesper Tække (Sat,) studied this question.