Generative AI Policies under the Microscope: How CS Conferences Are Navigating the New Frontier in Scholarly Writing

TL;DR

The paper investigates how Generative AI policies are implemented across CS conferences, addressing the problem of policy ambiguity in scholarly writing and peer review. It analyzes 64 CS conferences and major societies, rating leniency on a 5-point scale and computing inter-rater reliability with Krippendorff’s alpha, $\alpha=0.832$, to map adoption patterns. The findings reveal discipline-specific gaps, with author policies more common and lenient than reviewer policies, strong adoption in AI venues, and lagging or absent policies in Theory and some Systems conferences, culminating in uneven progress from Year 1 to Year 2 and no policy sanctioning for authors. The paper contributes a comprehensive landscape, area- and temporal-trend insights, and concrete recommendations for aligning conference and society policies to foster responsible Gen-AI use in scholarly writing and reviewing.

Abstract

As the use of Generative AI (Gen-AI) in scholarly writing and peer reviews continues to rise, it is essential for the computing field to establish and adopt clear Gen-AI policies. This study examines the landscape of Gen-AI policies across 64 major Computer Science conferences and offers recommendations for promoting more effective and responsible use of Gen-AI in the field.

Figures (2)

• Figure 1: Area-wise percentage (%) of conferences with Gen-AI policies for authors and reviewers in Years 1 and 2. The number in parentheses indicates the number of conferences per area.
• Figure 2: Area-wise average leniency ratings of conferences with Gen-AI policies for authors and reviewers in Years 1 and 2. The ratings were calculated based only on conferences that adopted Gen-AI policies for each year in each area. Theory field was excluded as those conferences did not adopt Gen-AI policies.