چكيده لاتين
With the advancement of artificial intelligence (AI) and the proliferation of agent-based intelligent systems, the formalization of propositional attitudes—such as belief, knowledge, desire, and intention—has gained paramount importance. These formalizations enable more precise modeling of human-like behaviors in machines, which is essential for applications including robotics, autonomous decision-making, and ethical AI systems. Without such rigorous formal frameworks, AI may prove inefficient in comprehending and predicting mental states, potentially leading to critical errors in human-machine interactions. Therefore, this thesis focuses on the examination and formalization of these attitudes within logic and AI, bridging the gap between philosophy of mind and emerging technologies.
This thesis analyzes propositional attitudes across categories such as cognitive, conative, informational, normative, and motivational, and formalizes them within epistemic logic, deontic logic, BDI logic (belief-desire-intention, based on CTL* temporal logic), and the KARO framework (knowledge, abilities, results, and opportunities, based on dynamic logic). The formalization of these logics is specifically investigated in the context of AI, employing possible worlds semantics and computational models to demonstrate that these frameworks are not only theoretically valuable but also practically applicable in designing autonomous and intelligence agents. Ultimately, the thesis proposes integrating these frameworks for modeling agentsʹ internal norms, thereby laying the groundwork for future AI research.
