What Is AI Safety? What Do We Want It to Be?

TL;DR

The paper confronts the lack of consensus on what AI safety should encompass and critiques the focus on future catastrophic risks and purely engineering-based approaches. It advocates The Safety Conception, defining AI safety as harm prevention and mitigation across both social and existential harms, analyzed through the lens of conceptual engineering. The authors articulate two aims: (A) provide a unifying explanation for why diverse AI-safety topics belong to the same field, and (B) promote practical harm reduction. They argue that this conception yields better research integration and policy impact, suggesting cross-domain venues, broader funding, and collaborative governance to address the full spectrum of AI harms.

Abstract

The field of AI safety seeks to prevent or reduce the harms caused by AI systems. A simple and appealing account of what is distinctive of AI safety as a field holds that this feature is constitutive: a research project falls within the purview of AI safety just in case it aims to prevent or reduce the harms caused by AI systems. Call this appealingly simple account The Safety Conception of AI safety. Despite its simplicity and appeal, we argue that The Safety Conception is in tension with at least two trends in the ways AI safety researchers and organizations think and talk about AI safety: first, a tendency to characterize the goal of AI safety research in terms of catastrophic risks from future systems; second, the increasingly popular idea that AI safety can be thought of as a branch of safety engineering. Adopting the methodology of conceptual engineering, we argue that these trends are unfortunate: when we consider what concept of AI safety it would be best to have, there are compelling reasons to think that The Safety Conception is the answer. Descriptively, The Safety Conception allows us to see how work on topics that have historically been treated as central to the field of AI safety is continuous with work on topics that have historically been treated as more marginal, like bias, misinformation, and privacy. Normatively, taking The Safety Conception seriously means approaching all efforts to prevent or mitigate harms from AI systems based on their merits rather than drawing arbitrary distinctions between them.