Digital Cognitive Capital and Generative AI.

RAG Socrate as a Methodological Environment for Epistemic Awareness

Autors

Monica Di Domenico; Pio Alfredo Di Tore.

Abstract

If Artificial Intelligence can be understood as an epistemic infrastructure that redefines the production and validation of knowledge, and if LLMs operate as statistical systems for the simulation of language, then the issue is not only what they answer, but when and why their answers are accepted as knowledge. The research draws on the understanding of media as cognitive environments capable of reorganising perception, language and epistemic authority (McLuhan, 1964), while conceiving digital cognitive capital as a dynamic set of critical, interpretive and metacognitive resources that enable subjects to orient themselves within informational environments (de Kerckhove, 2025). Within this framework, generative AI is examined as a socio-technical network of subjects, models, sources and interfaces (Latour, 2005), and as a threshold between cognitive extension and cognitive delegation (Clark, 2008). The contribution presents the design and first data of a pedagogical study conducted with the orchestrated RAG system Socrate, powered by Tiresia, understood as a methodological environment for making visible the conditions that produce the authority and plausibility of AI-generated answers.

Keywords

Generative artificial intelligence; Digital cognitive capital; Retrieval-augmented generation.

Bibliografia

Clark, A. (2008). Supersizing the mind: Embodiment, action, and cognitive extension. Oxford University Press.

de Kerckhove, D. (2025). L’uomo quantistico. Mente, società, democrazia: Dove ci porterà la prossima rivoluzione digitale. Rai Libri.

Latour, B. (2005). Reassembling the social: An introduction to actor-network-theory. Oxford University Press.

McLuhan, M. (1964). Understanding media: The extensions of man. McGraw-Hill.

Starting in September, the full article will be available on ojs.gsdjournal.it