Unpacking Connectionist Networks in Large Language Models: A Kristevan Analysis of Metalogues

Potential Abstract:
In this study, we investigate the interplay between connectionist networks and large language models within the framework of Julia Kristeva’s theory of language and semiotics. By analyzing metalogues – dialogues with a self-reflective nature – generated by advanced AI systems, we aim to uncover the intricate layers of meaning construction and intertextuality embedded within these models. Drawing on Kristeva’s concepts of the semiotic and symbolic realms, as well as her notions of language as a process of subject formation and representation, we delve into the ways in which these cutting-edge technologies exhibit characteristics of linguistic creativity, ambiguity, and narrative complexity. Through a qualitative analysis of the generated metalogues, we tease apart the underlying processes at play in the production of text by these models, shedding light on the ways in which they navigate linguistic structures, context, and coherence. Our findings contribute to a deeper understanding of the inner workings of connectionist networks in language models and their implications for language learning, creativity, and communication in educational settings.

Potential References: