From intelligence to autopoiesis: rethinking artificial intelligence through systems theory
Frontiers in Communication May 19, 2025 Benedikt Zönnchen, Mariya Dzhimova, Gudrun Socher
Large language models (LLMs) generate complex linguistic patterns that challenge the distinction between machine computation and human understanding. This article analyzes AI from a systems-theoretical perspective, arguing that classical Turing machines are not sense-making systems because they lack self-reference and the ability to make contingent selections. Artificial neural networks exhibit a novel, loosely coupled interaction with social systems by extracting patterns from societal communication. The paper proposes understanding LLMs as producing a new form of artificial meaning—a recursive reflection of socially shaped linguistic patterns—rather than as purely technical tools or genuine cognitive entities. This perspective calls for critical reflection on how AI transforms understanding of communication and cognition.