Survey Finds Broad Support for Greater Openness in AI Peer Review

Nihar B Shah, Andrew McCallum

OpenReview was launched in 2013 with the goal of bringing greater openness and constructive dialogue to scientific communication. In the first year of the ICLR conference, we conducted an initial survey that found measured but meaningful support among participants for elements of open peer review, including perceptions of improved fairness and review quality. Now, twelve years later, we have conducted a new survey of the broader machine learning and AI peer-review ecosystem to better understand community perspectives on open reviewing policies.

The survey received 2,385 responses from reviewers, authors and other participants. Overall, respondents expressed strong support for various degrees of openness in the peer-review process. More than 80% of respondents supported releasing reviews for accepted papers and allowing public comments after acceptance. In contrast, support in the broader community for releasing rejected manuscripts was much lower, at 27%.

Respondents identified several perceived benefits of open reviewing, including:

• Improved public understanding of research by providing additional context around papers
• Educational value for novice reviewers and authors
• Increased fairness and accountability in editorial decision-making
• Stronger incentives for writing careful, high-quality reviews

At the same time, respondents also raised concerns about perceived challenges, including:

• The risk of resubmission bias if reviews are publicly available
• Concerns about reviewer de-anonymization and potential professional repercussions
• The possibility of noise, low-quality comments, or abuse in public commenting systems

However, some respondents also noted that releasing rejected manuscripts could help discourage submissions that are not yet ready for peer review, including AI slop submissions, as well as low-quality or illegitimate automated reviews. So it seems that views on this issue are still evolving.

A detailed analysis of the survey results is available in the full report:
https://arxiv.org/pdf/2511.23439