Artificial Intelligence and the Problem of Representation: From Ontologies to Psychotherapy

This institutional scientific research report examines the problem of representation in contemporary artificial intelligence systems from an interdisciplinary perspective. Bringing together historical analysis, ontology engineering, cognitive theory, and psychotherapy-related frameworks, the report investigates the conceptual limits of AI-based representational models, with particular attention to large language models (LLMs).

The study is structured in three main parts: a historical reconstruction of artificial intelligence from its early logical foundations to contemporary data-driven systems; an analysis of functional ontologies and representation-centered architectures; and an inquiry into the epistemic and relational boundaries of AI-based chatbots in therapeutic contexts. Rather than proposing technical solutions, the report aims to clarify structural assumptions, methodological blind spots, and category errors that shape current debates on artificial intelligence.

This report is intended as a conceptual and theoretical contribution for researchers in artificial intelligence studies, philosophy of technology, cognitive science, and related social sciences.