1. SafetyAnalyst: Interpretable, transparent, and steerable safety moderation for AI behavior
- Author
-
Li, Jing-Jing, Pyatkin, Valentina, Kleiman-Weiner, Max, Jiang, Liwei, Dziri, Nouha, Collins, Anne G. E., Borg, Jana Schaich, Sap, Maarten, Choi, Yejin, and Levine, Sydney
- Subjects
Computer Science - Computation and Language ,Computer Science - Computers and Society - Abstract
The ideal AI safety moderation system would be both structurally interpretable (so its decisions can be reliably explained) and steerable (to align to safety standards and reflect a community's values), which current systems fall short on. To address this gap, we present SafetyAnalyst, a novel AI safety moderation framework. Given an AI behavior, SafetyAnalyst uses chain-of-thought reasoning to analyze its potential consequences by creating a structured "harm-benefit tree," which enumerates harmful and beneficial actions and effects the AI behavior may lead to, along with likelihood, severity, and immediacy labels that describe potential impact on any stakeholders. SafetyAnalyst then aggregates all harmful and beneficial effects into a harmfulness score using fully interpretable weight parameters, which can be aligned to particular safety preferences. We applied this conceptual framework to develop, test, and release an open-source LLM prompt safety classification system, distilled from 18.5 million harm-benefit features generated by frontier LLMs on 19k prompts. On a comprehensive set of prompt safety benchmarks, we show that SafetyReporter (average F1=0.81) outperforms existing LLM safety moderation systems (average F1$<$0.72) on prompt safety classification, while offering the additional advantages of interpretability, transparency, and steerability.
- Published
- 2024