Intelligenza artificiale e sostenibilità: Una via nell'inferenza attiva

The thesis investigates how Artificial Intelligence (AI) can integrate sustainability principles by leveraging the cognitive model of active inference (AIF). This is done by analysing current sustainable AI practices and the potential of cognitive architectures (CA) for sustainability and validating these concepts through a simulation implemented in a dynamic environment. The main objectives of this work are three: to provide a critical and interdisciplinary reflection on the limitations and opportunities associated with the use of cognitive models for environmentally and socially sustainable AI, fostering dialogue between philosophy of mind, cognitive sciences and AI; to explore the potential of AIF as a tool for designing sustainable AI; and to demonstrate, through a proposed simulation, the ability of an AIF-based system to balance immediate needs with long-term sustainability in dynamic contexts.