arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2602.21889cs.AIcs.LG
收藏

2-Step Agent: 一个用于决策者与AI决策支持交互的框架

2-Step Agent: How a Bayesian Decision Maker Learns from AI-Decision Support

Department of Medical Informatics, Amsterdam UMC University of Amsterdam(医学信息学系,阿姆斯特丹大学医学中心,阿姆斯特丹大学);Institute for Logic, Language and Computation, University of Amsterdam(逻辑、语言和计算研究所,阿姆斯特丹大学);Department of Mathematics and Earth Sciences, University of Trieste(数学与地球科学系,特里埃斯特大学)

Otto Nyberg, Davide Tugnoli, Fausto Carcassi, Giovanni Cinà

AI总结:

本文提出2-Step Agent框架,用于研究决策者如何学习和利用基于机器学习的决策支持,并揭示了即使在理想条件下,ML-DS也可能导致更严重的负面影响。

摘要:

机器学习模型的预测支持人类在多个领域做出决策,包括高风险领域如医疗和司法。然而,我们仍然缺乏对决策者如何从基于机器学习的决策支持(ML-DS)中学习的清晰理解。在本文中,我们介绍了一个通用的计算框架,即2-Step Agent,以捕捉这一过程。由于机器学习模型的预测包含关于训练数据的信息,预测也可以用于推断。我们的框架模型了(i)新的观察预测如何影响理性贝叶斯代理的信念,以及(ii)这种信念变化如何影响因果效应的估计、下游决策和后续结果。除了框架本身外,我们还做出了三个贡献。首先,在线性高斯设定下,我们推导出了解决我们引入的具有挑战性的贝叶斯推断问题的可计算解,即代理从ML预测中推断。其次,我们通过实验确定了ML-DS有益的条件。第三,我们证明了即使ML模型是良好规范的,且代理是完全理性的,单个不一致的先验信念也可能使ML-DS导致比没有决策支持更差的下游结果。因此,即使在理想条件下,ML-DS也可能造成更大的伤害。

英文摘要:

Prognostic predictions from ML models support human decision making in several fields, including high-stakes ones such as healthcare and the judiciary. Yet, we still lack a clear understanding of how decision makers learn from ML-based decision support (ML-DS). In this paper, we introduce a general computational framework, the 2-Step Agent, to capture how a prediction from an ML model can also be used as a source of useful information about the training data and the causal mechanism that generated it. Our framework models (i) how a prediction from an ML model affects the beliefs of a rational Bayesian agent, and (ii) how this change in beliefs affects the estimation of the causal effect, the downstream decision for a new observation, and the subsequent outcome. In addition to the framework itself, we make three contributions. First, for the multivariate linear-Gaussian setting, we derive a tractable solution for the Bayesian inference problem introduced, i.e. one in which the agent infers information about a population from an ML prediction. Second, we experimentally identify conditions under which ML-DS is beneficial compared to unassisted decision making. Third, we show that a single misaligned prior belief of the user can be sufficient for ML-DS to lead to worse downstream outcomes compared to no decision support even when the prognostic model is well-specified and the agent is perfectly rational. Hence, even under ideal conditions, ML-DS can do more harm than good.

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。

↑