선(禪)과 AI : 대화형 AI 선문답으로써의 가능성에 대한 시론(試論)

Abstract

This study explores whether conversational AI can function as a form of performative dialogue by comparing the structure of Seon Dialogue(禪問答) with the linguistic generative mechanisms of conversational AI.‌ From Bodhidharma to Huineng,‌ Seon Dialogue employed linguistic counter questions as a pedagogical means for practitioners to illuminate their own emptiness(śūnyatā),‌operating within language through recognizable dialogical patterns.‌In this respect,‌certain points of contact with conversational AI can be identified.‌ After Mazu,‌ however,‌ Seon Dialogue shifted toward non logical language and subversive actions that dismantled linguistic conventions to provoke immediate awakening,‌directly pointing beyond language itself.‌ Because conversational AI relies on pattern-based generation and reinforces appropriateness through RLHF,‌ it cannot reproduce this performative dimension that negates linguistic adequacy.‌ Accordingly,‌ whether conversational AI can engage in Seon Dialogue must be answered both affirmatively and negatively,‌depending on how the scope and definition of Seon Dialogue are understood.