Prompt Jams: How Online Collaboration Shapes the Use of Text-to-Image Generative AI

Abstract

Text-to-image generation (TTIG) has transformed digital creativity, yet collaboration within these communities remains under-explored. While traditional theories of online collaboration, such as those for Open Source Software, typically focus on convergence towards a single functional solution, TTIG collaboration prioritizes aesthetic divergence and rapid iteration. Drawing on semi-structured interviews with active TTIG community members and thematic analysis, we identify four collaboration modes: Direct, Indirect, Inspirational, and Functional collaboration. A primary contribution is the definition of "Prompt Jams" as collaborative episodes of prompt-based co-creation with a shared creative trajectory, iterative reciprocal responsiveness within a near-synchronous or session-based time window, and action-oriented feedback enacted through successive prompt revisions and generations. We show that collaboration in TTIG communities involves complex social performances and ethical tensions regarding ownership and remixing. We extend knowledge collaboration theory by conceptualizing prompts as a fluid, conversational artifact rather than a static object, offering critical insights for future platform design.

This document is currently not available here.

Share

COinS