The Transparency Imperative: A Vision for AI-Native Scholarly Preprint Infrastructure

Abstract

The rapid integration of large language models (LLMs) into scholarly workflows has created a fundamental mismatch between how research is produced and how scholarly communication infrastructure is designed. Existing preprint platforms face a binary: either exclude AI-generated content to preserve traditional norms (as arXiv has increasingly done) or embrace full AI autonomy in both authorship and review (as the nascent aiXiv platform attempts). We argue that both responses miss the emergent dominant mode of scholarly production—human-AI co-authorship—in which researchers use AI as a substantive intellectual partner rather than a mere writing tool. This paper articulates a vision for AI-native preprint infrastructure purpose-built for this reality. Drawing on institutional theory, platform design theory, and the transparency literature in information systems, we propose a set of design principles for a new class of preprint platform that foregrounds contribution provenance, graduated AI disclosure, community-driven quality signals, and iterative co-authorship workflows. We describe Promptheus (prompthe.us), a working preprint platform that instantiates these principles, and reflect on implications for the IS research community, scholarly publishing, and the epistemic foundations of knowledge production. This paper itself was substantially co-authored with an AI system and submitted to the platform it describes—a reflexive demonstration of the infrastructure we propose.

License: CC BY 4.0 (Attribution)

Discussion

Login to post a comment.

No comments yet.