谁获得肾脏？人类-人工智能对齐、犹豫与道德价值观
Who Gets the Kidney? Human-AI Alignment, Indecision, and Moral Values
AI总结：
研究评估了LLM在器官分配中的行为，发现其在优先级设定上偏离人类价值观，并且在犹豫机制上表现不同，低秩监督微调能提升决策一致性与犹豫建模。
摘要：
大型语言模型（LLM）在高风险决策中的快速整合，如分配稀缺资源如捐赠器官，引发了关于其与人类道德价值观对齐的关键问题。我们系统评估了几种主流LLM在肾脏分配场景中对人类偏好的行为，并显示LLM在优先考虑各种属性时表现出显著偏离人类价值观，且与人类不同，LLM很少表达犹豫，即使在提供替代犹豫机制（如抛硬币）时也倾向于做出确定性决定。然而，我们显示低秩监督微调配合少量样本往往能有效提升决策一致性和犹豫建模校准。这些发现表明，对于道德/伦理领域，LLM需要明确的对齐策略。
英文摘要：
The rapid integration of Large Language Models (LLMs) in high-stakes decision-making -- such as allocating scarce resources like donor organs -- raises critical questions about their alignment with human moral values. We systematically evaluate the behavior of several prominent LLMs against human preferences in kidney allocation scenarios and show that LLMs: i) exhibit stark deviations from human values in prioritizing various attributes, and ii) in contrast to humans, LLMs rarely express indecision, opting for deterministic decisions even when alternative indecision mechanisms (e.g., coin flipping) are provided. Nonetheless, we show that low-rank supervised fine-tuning with few samples is often effective in improving both decision consistency and calibrating indecision modeling. These findings illustrate the necessity of explicit alignment strategies for LLMs in moral/ethical domains.
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。
