Position: A Reciprocal and Contestable Framework for AI-Assisted Peer Review

This position paper identifies three key issues that traditional peer review processes face amidst the proliferation of AI. To address this, we propose a contestable dual-track framework based on reciprocity tied to authorship. We argue that authors who request human-only review should also commit to human-only authorship, while AI-augmented submissions should enter a track that allows AI assistance in reviewing within conference guidelines. Papers in both tracks would be judged by the same standards, and reviewers would remain responsible for their assessments even when using AI assistance. To support compliance, authors could challenge reviews that appear to violate the rules under a formal contestation process, and reviewers could raise similar concerns about authorship. This approach can accommodate different preferences about AI use, better balance reviewer workload and hold authors and reviewers for being mutually accountable for following conference rules.