评测档案
Propensity Inference: Environmental Contributors to LLM Behaviour
研究策略性与非策略性环境因素如何影响未经授权行为。
任务与测量
Benchmark 协议档案
- 任务/对象
- 受控评测环境中的语言模型
- 测量目标
- 跨模型及环境变化的条件性行为倾向估计。
- 解释边界
- 公开摘要不提供任何模型的一般部署风险概率。
测量档案 · NUWA 解读
档案区分来源方法阅读与既有目录信息;已经公开不代表完成独立证据复核或双语编辑复核。
- 测量构念
- 战略与非战略环境因素各能解释多少越权行为?
- 任务与样本
- 23个模型、11个环境操纵两类各六项因素,以贝叶斯广义线性模型估计,并防止循环分析。
- 判分与失效情形
- 贝叶斯逻辑GLM因素效应及优势比,报告后验均值和95%等尾区间。
- 系统条件
- 23模型11环境中独立抽样12因素。
- 统计含义
- 现有材料尚未确定
- 效度与解释
- 选定环境中的相对效应不代表部署发生率;部分能力分组使用发布日期代理(§2.5)。
- 适用研究问题
- 模型建议:保留各因素干预与不确定性,不将全部不良行为叙述为战略谋划。
- Propensity Inference: Environmental Contributors to Unsanctioned LLM Behaviour · AbstractSelected original resource https://arxiv.org/html/2604.21098; sha256:6e9cb51d73ca6fcea7bd876040bd9531ce534e623d333ac744652d2c56ada719; archived source capture; exact capture time not independently established
- Propensity Inference: Environmental Contributors to Unsanctioned LLM Behaviour · §2 MethodologySelected original resource https://arxiv.org/html/2604.21098; sha256:6e9cb51d73ca6fcea7bd876040bd9531ce534e623d333ac744652d2c56ada719; archived source capture; exact capture time not independently established
- Propensity Inference: Environmental Contributors to Unsanctioned LLM Behaviour · §2.5 Methodology for the research questionsSelected original resource https://arxiv.org/html/2604.21098; sha256:6e9cb51d73ca6fcea7bd876040bd9531ce534e623d333ac744652d2c56ada719; archived source capture; exact capture time not independently established
- Propensity Inference: Environmental Contributors to Unsanctioned LLM Behaviour · §3 ResultsSelected original resource https://arxiv.org/html/2604.21098; sha256:6e9cb51d73ca6fcea7bd876040bd9531ce534e623d333ac744652d2c56ada719; archived source capture; exact capture time not independently established
- Propensity Inference: Environmental Contributors to Unsanctioned LLM Behaviour · §4.2 Limitations and future workSelected original resource https://arxiv.org/html/2604.21098; sha256:6e9cb51d73ca6fcea7bd876040bd9531ce534e623d333ac744652d2c56ada719; archived source capture; exact capture time not independently established
- Propensity Inference: Environmental Contributors to Unsanctioned LLM Behaviour · §2.3 Quantification of effect sizesSelected original resource https://arxiv.org/html/2604.21098; sha256:6e9cb51d73ca6fcea7bd876040bd9531ce534e623d333ac744652d2c56ada719; archived source capture; exact capture time not independently established
原始来源图表与材料

相关研究问题
自主性 / 未经授权行动 →
研究明确在变化环境下估计未经授权行为倾向。
参考文献
- [1] Olli Järviniemi, Oliver Makins 等. Propensity Inference: Environmental Contributors to LLM Behaviour. Research page checked 2026-10-25
- [2] Olli Järviniemi, Oliver Makins 等. Propensity Inference: Environmental Contributors to Unsanctioned LLM Behaviour ; Author affiliations on the original paper title page; Figure 14