Abstract
Generative artificial intelligence has introduced a discontinuity in the long history of human meaning-making: for the first time, language is no longer an exclusively human medium for producing, circulating, and legitimizing knowledge. Building on Francisco Varela’s notions of emergence and enaction, Neil Postman’s technological critique, and Yanis Varoufakis’ technopolitical lens, this paper offers a critical reading of contemporary generative systems as epistemic actors rather than mere tools. We contrast rule-based, “trivial” systems with today’s large, parameterized, multimodal models—closer to Heinz von Foerster’s non-trivial machines and Edgar Morin’s complexity—in which knowledge is encoded in dynamic, high-dimensional spaces rather than explicit rules. This shift destabilizes inherited epistemological assumptions in at least three ways: (i) it obscures the locus of authorship and accountability, since linguistic output emerges from socio-technical ensembles; (ii) it blurs the human/machine distribution of cognitive labor in academic and scientific practices; and (iii) it challenges institutional frameworks that still presuppose human-centered, linear models of validation. We argue that, under the 2026 special focus on human-centered AI transformations, these developments require re-grounding research and educational practices in enactive, relational, and institutionally supported approaches to knowledge-making. The paper concludes by examining whether LLM self-improvement processes can be meaningfully framed through second-order cybernetics and (limited) autopoietic metaphors, and what this implies for governing AI-mediated knowledge.
Presenters
José Luis CarrascoProfessor and Researcher, Departamento de Electrónica e Informática, Universidad Técnica Federico Santa María, Bío-Bío, Chile Israel Figueroa
Student, PhD in Artificial Inteligence, Universidad Catolica de la Santisima Concepcion, Bío-Bío, Chile
Details
Presentation Type
Paper Presentation in a Themed Session
Theme
2026 Special Focus—Human-Centered AI Transformations
KEYWORDS
Generative artificial intelligence, Epistemic destabilization, Non-trivial machines, Human–AI co-authorship
