Research Summary: Governing AI Research Through Peer Review: A Mixed-Methods Study of the Longitudinal Effects of Ethics Flags Across Resubmissions

Research indicates that the implementation of ethics flags in selective Artificial Intelligence (AI) conferences primarily leads to revisions in how research projects are presented rather than significant changes to the underlying research agendas. This suggests that current mechanisms designed to steer AI research towards safer and more responsible practices may not be achieving their intended depth of impact on research direction.

Why it matters

This finding highlights a critical gap in the effectiveness of current governance mechanisms for Artificial Intelligence research. While aiming for responsible innovation, the current approach may only be superficial, potentially allowing underlying ethical risks in research methodologies or objectives to persist. This requires a strategic re-evaluation of how to genuinely influence research direction towards more robust ethical considerations.

Key insights

• Selective AI conferences have introduced ethics flags and related review requirements to promote safer and more responsible research practices.
• The primary observed effect of ethics flags on rejected or withdrawn submissions is the revision of project presentation.
• Authors are less likely to redirect their fundamental research agendas following ethics flags.
• The study tracks longitudinal effects of ethics flags on resubmissions after the initial review process concludes.

Source

arXiv — Computers and Society — https://arxiv.org/abs/2609.10740