Ontological Drift in Generative Mediation: A Working Paper on Semiotic Preservation, Human-AI Interaction, and Don Quixote Adaptations

This working paper presents a public and synthetic version of an ongoing doctoral research project in Communication Sciences at the University of Beira Interior, LabCom. The project investigates how Generative Artificial Intelligence systems preserve, transform, or displace the ontological identity of cultural artefacts during processes of reproduction, adaptation, and reinterpretation. The research develops three core conceptual contributions: ontological drift, meta-artefact, and linear indexing. Ontological drift designates the cumulative displacement of an artefact’s semiotic identity during generative mediation. Meta-artefact defines the generative output as a second-order object produced from previously indexed cultural artefacts, rather than from a direct material relation with the empirical world. Linear indexing designates the computational reduction of complex semiotic relations, narrative fields, and axiological tensions into processable units within generative pipelines. The project also proposes two methodological instruments: the Recurrent Semantic Matrix and the Longitudinal Coherence Index. These instruments are designed to analyse the preservation or loss of semiotic identity across Cervantes’s Don Quixote, selected intermedial adaptations, and generative AI outputs. The research is positioned primarily within Human-Computer Interaction, with secondary alignment to Responsible AI in HCI, Human-AI Collaboration, Generative Artificial Intelligence, Natural Language Processing, semiotics of culture, and intermedial adaptation studies.