Abstract: Whether peer review suppresses disruptive work is a crucial topic in the field of scientometrics and technology policy. Using publicly available review data from the International Conference on Learning Representations (ICLR), a top-tier conference in the field of artificial intelligence, this study analyzes the correlation between the disruptiveness of scientific papers and their scores in peer review. We further explore whether disruptive works are accompanied by significant dissensus. It finds a significant positive correlation between paper disruption and review scores, indicating that more disruptive work is more likely to receive high recognition from peers. This suggests that the peer review system does not suppress disruptive work. Furthermore, no obvious dissensus was found to be associated with highly disruptive papers. This indicates that in the rapidly advancing frontier fields of artificial intelligence, open peer review can effectively identify outstanding work with disruptive potential. This study provides theoretical basis and empirical support for science and technology management departments to optimize the structure of review panels, encourage original innovation, and build a more open and inclusive research evaluation system.

Key words: Disruptive work, Original innovation, Peer review, Dissensus, ICLR