The release of first-round review comments for ICLR 2027 has set off a wave of complaints across the AI community. The reviews have surfaced a range of problems: some demanded that authors explain standard deviations; some followed templates so closely that leftover AI prompt fragments remained in the text; some questioned papers simply because authors had disclosed AI-assisted writing, as they were required to do; and some reviewers admitted that their own expertise was insufficient to evaluate the work. Critics argue that AI's involvement in review has clear drawbacks: it can readily churn out a long list of limitations, yet cannot distinguish ordinary limitations from critical issues that undermine a paper's core conclusions—and in doing so, it often buries the very flaws that most deserve attention. At bottom, these problems reflect a deeper tension in the AI era: academic review resources simply cannot keep pace with the surge in submissions. The ICLR 2027 Program Chairs had previously introduced measures such as capping the number of submissions, strengthening the reciprocal review system, and permitting reviewers to use AI-assisted review only to a limited extent, subject to disclosure requirements. The first-round reviews have only just been released; authors may now raise questions in communication with reviewers. Final acceptance decisions will be announced on December 16.