发现大模型有稳定风险偏好，可像人一样评估决策风险。
Some Large Language Models Exhibit Consistent Risk Attitudes
- 设计跨任务框架，分离风险认知与决策行为
- 6个大模型在3类任务中表现一致风险倾向
- 模型风险偏好趋同，与人类分布差异明显
随着人工智能系统在开放、高风险场景中的部署，一个关键维度仍缺乏测量：风险感知如何转化为行动。本文测试大语言模型（LLMs）在不确定性下是否表现出系统性且一致的风险态度。提出一种跨领域框架，将上下文风险信念与分类决策解耦，并在空间导航、临床分诊和金融分配任务中对6个代表性LLMs及100名人类参与者进行测试。通过回归模型提取各智能体的信念-决策映射，量化风险敏感度与风险态度偏差。结果表明，多数测试的LLMs表现出：(i) 任务内一致性，即在固定任务域内信念到风险决策的映射稳定；(ii) 跨领域排名稳定性，保持相对风险姿态不变；(iii) 相较于更广泛的个体人类基准，向有限的风险态度分布收敛。这些发现揭示了风险态度是大模型行为中一种稳定且此前未被描述的维度，为评估和对齐开放式决策中的AI系统奠定了基础，并激发对这些内在行为倾向起源的进一步研究。
原文摘要 · Abstract (English)
As artificial intelligence systems are deployed in open-ended, high-stakes settings, a critical dimension remains unmeasured: how perceived risk is translated into action. We test whether large language models (LLMs) exhibit systematic and consistent risk attitudes under uncertainty. We introduce a cross-domain framework that decouples contextual risk belief from categorical decision, and apply it to six representative LLMs and 100 human participants across spatial navigation, clinical triage, and financial allocation tasks. Using regression models, we extract each agents belief-to-decision mapping and quantify risk sensitivity and risk attitude bias. We find that most tested LLMs exhibit (i) robust intra-task consistency, indicating stable mappings from contextual belief to risk decision within a fixed task domain; (ii) cross-domain rank-order stability, preserving relative risk posture across tasks; and (iii) a convergence toward a restricted risk-attitude distribution relative to the broader human baseline. These results reveal risk attitude as a stable and previously uncharacterized dimension of LLM behavior, establishing a foundation for evaluating and aligning AI systems in open-ended decision-making and motivating further investigation into the origins of these intrinsic behavioral dispositions.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。