Riscopriamo l'ONTOLOGIA per capire l'IA | FILOSOFIA con Luciano Floridi

Metaphysics is the theory of the "thing in itself," of what reality is regardless of whether anyone observes or knows it, what Kant called the noumenon, that is, the object separated from all our conceptualizations or models. The problem is that without the act of knowledge, metaphysics risks becoming a set of arbitrary stories, fascinating narratives but devoid of objective criteria. On the other hand, we have ontology, which for us today represents the description of the world as it is perceived, modeled, and internalized by the human species. Ontology is a theory of the model, not of the system itself, and can vary depending on our purpose or our senses. But what does all this have to do with artificial intelligence? Well, we will discover that the extraordinary success of today's generative AIs comes not from their supposed intelligence, but from the fact that between the machine and the world we have inserted a massive interface made up of all our semantic capital: texts, images, films, and translations. AI doesn't read the world, but reads our translation of it. We'll also discover how new avenues of AI research seem to ignore the lessons of history and philosophy, risking spending billions to achieve something impossible or unnecessarily narrow. Unfortunately, as often happens, a little more reading would be enough to avoid repeating mistakes that have already proven to be very short-lived.