Merit-first evaluation policy for AI-generated manuscripts

Determine whether peer reviewers should evaluate AI-generated manuscripts strictly on scientific merit before any disclosure of AI authorship to mitigate evaluative bias in the peer-review process.

Background

The paper conducts a controlled paper of fully AI-generated manuscripts submitted to a peer-reviewed workshop and discusses ethical and policy questions surrounding AI-generated research. In articulating future community norms, the authors explicitly note uncertainty about whether reviewers should be informed of AI authorship before or after evaluating the work, due to potential bias.

This question is framed within broader considerations of transparency and responsible oversight, including disclosure requirements and timing, emphasizing the need for community consensus on fair review practices for AI-generated content.