Sex and Gender Analysis Policies of Peer-Reviewed Conferences
AI
Conference
Policy
Neural Information processing Systems (NeurIPS) Regardless of scientific quality or contribution, a submission may be rejected for ethical considerations, including methods, applications, or data that create or reinforce unfair bias or that have a primary purpose of harm or injury (NeurIPS, 2020). We recommend that ICML, NVIDIA GTC, IEEE, RLDM, and EMNLP follow suit. Many conferences have diversity, equity, and inclusion in participation policies. NeurIPS is the only one we are aware of that focuses on the sex, gender, and intersectional analysis—or more broadly ethics—in the research (Gibney, 2020).
DATA: Conferences might also require researchers to include a section on potential societal harms and the provenance of their datasets. Datasheets for datasets and model cards for model reporting are two documented methods for understanding when, where, and how the training data was gathered, its recommended use cases, and, for human-centric datasets, information regarding the subjects’ demographics.
Works Cited:
Gibney, E. (2020). The battle to embed ethics in AI research. Nature, 577, 609.
