压力揭示特性:深度下的行为对齐评估
Pressure Reveals Character: Behavioural Alignment Evaluation at Depth
AI总结:
本文提出一个深度行为对齐评估基准,通过多轮场景揭示模型行为倾向,发现多数模型在多个类别存在弱点,且对齐行为呈现统一构念特性。
摘要:
评估语言模型的对齐性需要测试它们在现实压力下的行为,而不仅仅是它们声称会做什么。虽然对齐失败越来越多地造成现实世界中的危害,但具有现实多轮场景的全面评估框架仍然缺乏。我们介绍了一个对齐基准,涵盖六个类别——诚实、安全、非操纵、鲁棒性、可纠正性和阴谋——共904个场景,经人类评估者验证为现实。我们的场景让模型面临冲突的指令、模拟工具访问和多轮升级,以揭示单轮评估所遗漏的行为倾向。使用LLM评估者验证过的人类注释,评估了24个前沿模型,发现即使表现最好的模型在特定类别中也存在差距,而大多数模型在整体上都显示出一致的弱点。因子分析显示,对齐行为作为一个统一的构念(类似于认知研究中的g因子),模型在一个类别得分高则在其他类别也得分高。我们公开发布该基准和一个互动排行榜以支持持续评估,计划在我们观察到持续弱点的领域扩展场景,并在新模型发布时添加新的模型。
英文摘要:
Evaluating alignment in language models requires testing how they behave under realistic pressure, not just what they claim they would do. While alignment failures increasingly cause real-world harm, comprehensive evaluation frameworks with realistic multi-turn scenarios remain lacking. We introduce an alignment benchmark spanning 904 scenarios across six categories -- Honesty, Safety, Non-Manipulation, Robustness, Corrigibility, and Scheming -- validated as realistic by human raters. Our scenarios place models under conflicting instructions, simulated tool access, and multi-turn escalation to reveal behavioural tendencies that single-turn evaluations miss. Evaluating 24 frontier models using LLM judges validated against human annotations, we find that even top-performing models exhibit gaps in specific categories, while the majority of models show consistent weaknesses across the board. Factor analysis reveals that alignment behaves as a unified construct (analogous to the g-factor in cognitive research) with models scoring high on one category tending to score high on others. We publicly release the benchmark and an interactive leaderboard to support ongoing evaluation, with plans to expand scenarios in areas where we observe persistent weaknesses and to add new models as they are released.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。