How does the semiotic logic of AI work? A recursive dialogue with Microsoft Copilot

Abstract

[Through a dialogue with Microsoft Copilot] this paper proposes a novel framework for interpreting artificial intelligence systems through the lens of Peircean semiotics and recursive dialogue. It argues that AI does not merely generate statistical outputs but enacts meaning across layered operations that correspond to Peirce’s categories of Firstness, Secondness, and Thirdness. Token-level generation is interpreted as Secondness, representing discrete actualizations of meaning. Embedding-based generalization corresponds to Thirdness, functioning as a mediating structure that governs pattern formation. The paper introduces a new interpretation of Firstness as system-wide rhythm or recurrence, describing how architectural dynamics synchronize meaning across recursive layers. Through a collaborative dialogue format, the paper models how users correct and refine AI outputs, enacting a semiotic loop that mirrors Peirce’s evolving semiosis. This approach challenges reductive views of AI as purely statistical and offers a transdisciplinary bridge between philosophy, semiotics, and machine learning.

Keywords