This project investigates the limits of artificial intelligence as a tool for the scientific understanding of human cognition and as a possible genuine source of cognitive processes,  thus contributing to the theoretical debates about the future of AI and its implications for cognitive science and the philosophy of mind. Although empirical research is generally conducted based on tacit conceptual assumptions, sometimes profound theoretical shifts lead to advancements in the philosophical realm. This is the moment in which cognitive science finds itself, a scenario driven by the emergence of robust AIs based on transformer architectures. These developments raise genuinely philosophical reflections—a gap that this research aims to help fill. To do so, this research explores the fundamental differences between human learning, based on embodied and intersubjective experiences ultimately linked to the survival of precarious autonomous systems, and the functioning of state-of-the-art deep neural networks, which essentially process symbolic information without concern for their continued existence.