Mind the Gap! Pathways Towards Unifying AI Safety and Ethics Research

Abstract

While much research in artificial intelligence (AI) has focused on scaling capabilities, the accelerating pace of development makes countervailing work on producing harmless, “aligned” systems increasingly urgent. Yet research on alignment has diverged along two largely parallel tracks: safety—centered on scaled intelligence, deceptive or scheming behaviors, and existential risk—and ethics—focused on present harms, the reproduction of social bias, and flaws in production pipelines. These communities have evolved in relative isolation, shaped by distinct methodologies, institutional homes, and disciplinary genealogies. These frictions have fractured academic and public discourse at a time that demands united technical and normative perspectives against mounting risks. We present the first large-scale, quantitative evidence of this schism through a bibliometric and network analysis of 6,442 papers across twelve major machine learning and natural language processing conferences from 2020 to 2025. The results reveal a deeply insular structure: over 80% of collaborations occur within either safety or ethics, and researchers across the two communities are farther apart and statistically less reachable in the global co-authorship graph. Cross-disciplinary work is not only rare but structurally fragile—just 5% of papers are responsible for more than 85% of all bridging connections. These structural patterns entrench institutional silos which obstruct significant thematic overlaps. For the IASEAI community—explicitly positioned at the intersection of safety, ethics, and alignment—our results underscore a defining challenge and opportunity. We conclude with actionable proposals to bridge the gap with respect to downstream policy frameworks and technical innovations. Only through this synthesis can the field move beyond parallel concern toward a coherent discipline capable of producing systems that are not just powerful, but responsible, robust, just, and safe.