A bibliometric analysis of 6,442 papers from major machine learning and NLP conferences between 2020 and 2025 found that over 80% of collaborations occur within either the AI safety or AI ethics community, with minimal cross-disciplinary exchange. The researchers found that just 5% of papers account for more than 85% of the bridging connections between the two fields, suggesting the divide is structural and institutional rather than merely conceptual. The authors argue for integrating technical safety research with normative ethics through shared benchmarks, joint venues, and mixed methodologies to build AI systems that are both robust and just.
