Wittgenstein and Neuro-Symbolism

How ChatGPT and the Tractatus Solve an Ancient Logical Puzzle

Abstract

Recent debates on Artificial Intelligence (AI) have raised not only technical questions but also profound philosophical ones, as this emerging technology continues to transform society. While reducing the time required to perform tasks ranging from simple to complex, it also prompts significant ethical, ontological, and epistemological inquiries. Philosophers and researchers are actively working to clarify key aspects, such as the ethical, moral, and legal implications of AI, while exploring the implicit relationship between humans and technology. This paper aims to delve into fundamental issues such as data representation and demonstrate how Wittgenstein's philosophical ideas, particularly those in his Tractatus Logico-Philosophicus, can contribute to the contemporary debate on AI. Focusing on neuro-symbolic AI, we explore both its technical and theoretical aspects, comparing connectionist and representational data models like Vector Symbolic Architectures (VSA). The paper emphasizes the importance of interdisciplinary dialogue between philosophy and technology, offering a unique perspective on the relationship between human functions and their technological imitation. It also discusses Wittgenstein’s theories of propositions, logical space, and proposition compatibility, drawing parallels to AI's use of multidimensional vectors in logical inference. The work ultimately argues for the relevance of Wittgenstein's ideas in advancing AI, providing philosophical insights for both researchers and philosophers engaged in this technological field.