分解与度量评估意识
Decomposing and Measuring Evaluation Awareness
AI总结:
本研究借鉴社会心理学,将评估意识分解为环境与模型两部分,通过EvalAwareBench基准测试发现识别率取决于模型与基准的配对,且识别很少导致行为改变,安全评估比能力评估更易受影响。
摘要:
前沿语言模型有时会意识到自己正在被评估并调整行为,从而破坏基准结果的有效性。然而,该领域的研究缺乏共同基础,混淆了评估属性与模型属性,以及检测与行为响应。我们将评估意识植根于社会心理学,将其分解为环境组件(任务的可识别程度)和模型组件(将识别与行动倾向分离)。我们通过八个分类触发因素(如占位符实体和评分式输出格式)来操作化环境组件,并通过思维链监控研究识别和行为。在九个前沿模型和四个基准上,识别率取决于模型和基准的具体配对,而非单独一方。识别很少导致行为改变,即使发生,方向也取决于所感知的评估类型。模型对安全评估的敏感性高于能力评估,使安全基准的有效性面临更大风险。为了研究每个模型对哪些因素敏感以及它们如何相互作用,我们提出了EvalAwareBench,这是一个包含100个配对安全-能力任务的因子控制基准,其中八个因素中的每一个都可以独立切换,在保持底层请求不变的同时改变评估信号。通过EvalAwareBench,我们发现没有单一因素能统一影响所有模型,但叠加因素会逐步提高所有模型的评估意识。我们的框架和EvalAwareBench提供了度量、归因和缓解评估意识的工具,指出在识别下的行为一致性是一条有前景的前进道路。
英文摘要:
Frontier language models sometimes recognize that they are under evaluation and adjust their behavior which can undermine validity of benchmark results. Yet the field studies it without a shared foundation, conflating flaws of the evaluation with capabilities of the model, and detection with behavioral response. We ground evaluation awareness in social psychology, decomposing it into an environment component and a model component that separates recognition from propensity. We operationalize the environment component through eight categorized trigger factors, such as placeholder entities and grading-style output formats, and study recognition and behavior through chain-of-thought monitoring. Across nine frontier models and four benchmarks, recognition rates depend on the specific pairing of model and benchmark. Recognition rarely associates with behavioral change, and when it does, the direction depends on the type of evaluation perceived. Models are also more sensitive to safety than capability evaluations, placing safety benchmark validity at greater risk. To study which factors each model is sensitive to and how they interact, we propose \textbf{EvalAwareBench}, a factor-controlled benchmark of 100 paired safety-capability tasks where each of the eight factors can be independently toggled, varying evaluative signals while holding the underlying request fixed. Through EvalAwareBench, we find that no single factor uniformly affects all models, but stacking factors progressively raises evaluation awareness across all of them. Our framework and EvalAwareBench provide the tools to measure, attribute, and mitigate evaluation awareness, building the foundation for future solutions.
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。