NeurIPS Mandates Societal Impact and Financial Disclosure for AI Papers
NeurIPS now requires all submitted AI research papers to include statements on societal impact and financial conflicts of interest. Papers flagged for ethical concerns face additional review by ethics experts before acceptance.
NeurIPS, one of the world's largest AI research conferences, now requires all submitting researchers to disclose the potential broader societal impacts of their work and any financial conflicts of interest. This groundbreaking mandate forces researchers to confront both the positive and negative outcomes of their AI models. Additionally, the financial disclosure requirement sheds light on the influence that big tech companies and industry funding have on machine learning research.
These new ethical and financial statements appear in the final published versions of the papers so they do not compromise the double-blind review process. Reviewers initially assess submissions based solely on technical contributions. However, if a paper raises potential ethical concerns, it goes to a secondary panel of reviewers who possess expertise in both ethics and machine learning, and final acceptance depends on their positive assessment.
Conference organizers acknowledge that the specific norms for these societal impact statements are still developing and will evolve over future conferences. While NeurIPS previously discussed requiring researchers to calculate and state the carbon footprint of their AI models, this specific environmental metric is not explicitly mandated in the current submission requirements. Organizers note that this climate concern remains important but is not universally relevant since many conceptual papers do not require large-scale computational resources.