Self-Programming Dynamics: Exploring Generative AI-User Interactions through Luhmann’s Systems Theory

Abstract

This paper explores the self-programming dynamics between generative AI and its users, drawing on Luhmann’s systems theory to frame the interaction. By tracing the development of generative AI from early models to advanced architectures like foundation models, the paper illustrates how these systems approximate contextual understanding and respond dynamically to user inputs. The concept of self-programming is extended to users, who adapt their engagement strategies through repeated interactions, developing prompt scripts and styles that enhance the usefulness of AI outputs. The study highlights the recursive feedback loop between AI and users, wherein both entities influence and modify each other’s operations. Additionally, the paper investigates the structural coupling of these interactions, emphasising how user inputs and AI responses form a shared textual medium that evolves over time. Despite advancements, the research identifies persistent paradoxes in AI-mediated interactions, particularly the balance between specificity and generality, and the tension between novel and familiar information. It calls for further examination of these paradoxes to foster more meaningful and adaptive forms of AI-user interactions.