Abstract

Artificial intelligence is currently reshaping academic publishing at a pace that far exceeds the response time of institutional governance frameworks. This crisis stems from the lack of a shared definition for ethical AI use throughout the research lifecycle. Recent evidence indicates that while roughly 13.5% of biomedical papers in 2024 involved LLM processing (Kobak et al. 2025), 64% of journals still provide no AI guidance for peer reviewers (Wang et al. 2026). Without intervention, these sociotechnical systems will continue to distribute power asymmetrically, following established patterns of digital inequality (Warschauer 2003) and compounding the structural disadvantages faced by early-career and under-resourced scholars. This talk argues that existing governance arrangements compose an Asymmetric Burden Problem where the costs of compliance concentrate more on scholars with the least institutional capacity. We identify four compounding mechanisms driving this trend. Policy Fragmentation creates a landscape where varying guidelines across disciplines burden those who lack institutional support. The Multiplier Effect allows well-resourced institutions to capture disproportionate productivity gains, which widens structural inequality even as absolute output rises (Merton 1968). Furthermore, the Honest Reviewer Trap creates an environment where ethical reviewers who rely less on AI assistance complete fewer assignments and might lose critical network access. Finally, the Temptation Architecture imposes an ethics tax where the burden of ambiguous standards increases as career security decreases. To address these systemic failures, we propose a four-layer governance framework: define, train, equalize, and protect. This response includes the creation of a joint AIS and COPE discipline-tiered taxonomy covering the full research lifecycle and a cross-jurisdiction co-authorship annex. It further advocates for career-stage-differentiated AI literacy programs that acknowledge the 87% adoption rate among early-career researchers. To level the playing field, we call for symmetric reviewer disclosure requirements and an equity access fund for AI tools at under-resourced institutions. Finally, the framework seeks to protect scholars through conflict arbitration pathways and tenure criteria adjusted for AI-inflated submission environments. The IS community is uniquely positioned to lead this effort by applying its decades of research on sociotechnical power to its own publishing infrastructure (Orlikowski 2000).

Download

Share

COinS