Strengthening the AI Research Ecosystem: Integrity, Critique, and Consensus

This panel brings together three recent challenging position papers from the NeurIPS 2025 platform that collectively spotlight structural vulnerabilities in the machine-learning research ecosystem and propose bold reforms. The first paper, “Stop DDoS Attacking the Research Community with AI‑Generated Survey Papers”, identifies the surge of AI-generated, mass-produced survey manuscripts as a form of “survey-paper DDoS” that threatens to flood and degrade the research record. The second, “Position: Machine Learning Conferences Should Establish a ‘Refutations and Critiques’ Track”, argues that major ML conferences currently lack a credible, high-visibility venue for rigorous critiques and corrections of prior work, and proposes the creation of a dedicated “Refutations & Critiques” track. The third paper, “NeurIPS should lead scientific consensus on AI policy”, makes the case that NeurIPS (and by extension the ML community) should play an active role in building scientific consensus on AI policy, filling an important gap in evidence-synthesis and decision-making.