The Semantic Gap and Limits of AI Agency: Buddhist Dependent Origination, Emptiness, and Wittgenstein’s Philosophy of Language

This paper investigates the limits of artificial intelligence (AI) with respect to semantic understanding and moral agency by bringing Buddhist philosophy and Wittgenstein’s philosophy of language into a comparative framework. Contemporary deep-learning systems learn by optimizing large parameter spaces through supervised learning, backpropagation, and gradient descent. Although such systems generate highly fluent outputs, their operations remain at the level of syntactic pattern processing rather than genuine semantic understanding. Searle’s “Chinese Room” argument makes this gap explicit: rule-based symbol manipulation, however sophisticated, does not amount to understanding. This finding resonates with Wittgenstein’s later view that meaning is not an intrinsic property of signs but arises from use, practice, and forms of life. Buddhist thought, particularly the doctrines of dependent origination (pratītyasamutpāda) and emptiness (śūnyatā), deepens this analysis. According to Madhyamaka philosophy, all phenomena—including linguistic meanings—are empty of intrinsic nature and arise only in relational networks of conditions. From this perspective, AI outputs do not possess meaning in themselves; rather, their meaning originates in a contingent nexus of training data, model architecture, human interpreters, and socio-linguistic contexts. The paper then examines the “responsibility gap” in AI decision-making through Buddhist ethics, arguing that intentionality (cetanā) is the basis of moral responsibility. Because current AI systems lack consciousness and intention, they cannot be moral agents; responsibility must be ascribed to the humans and institutions that design, deploy, and regulate these systems. In conclusion, the study briefly considers, as a purely speculative horizon, whether a future “Robo-Dasein” equipped with embodiment, second-person relationality, and self-reflection could transform these assessments. Overall, the paper proposes a relational, cross-cultural framework for rethinking meaning, agency, and responsibility in the age of AI, integrating insights from analytic philosophy, Buddhist thought, and the philosophy of technology.