跳转至

文章背景与核心概要

可解释人工智能（XAI）领域虽然已经开发出众多的解释技术，但由于传统工具各自为战的孤岛特性，用户在实际应用中依然难以感受到它们的实际效用。本文从哲学视角切入，将“选择问题”进行了形式化定义，即XAI界面系统性地无法弥合用户自然语言中的不确定性与解决该问题所需的特定解释技术之间的鸿沟。作者采用逻辑前提-结论的形式，论证了传统界面提出了一个不合理的前提条件：要求用户必须将个人的不确定性转化为具体的技术选择。

为了解决这一痛点，本文提出了一种结构性解决方案，即利用多智能体大语言模型（LLM）编排工具，自动将用户的自然语言查询映射到对应的XAI解释技术上。该研究不仅深化了对人机交互中XAI可用性瓶颈的理解，也为构建更加智能、直观的可解释AI系统提供了切实可行的技术路径。

Addressing the Selection Problem in Explainable AI

Summary

Explainable AI (XAI) research has successfully developed numerous explanation techniques, yet users continue to struggle to find them practically effective due to the siloed nature of conventional tools. This paper explores this issue through a philosophical lens, formalizing the selection problem—the systematic failure of XAI interfaces to bridge the gap between a user's natural-language uncertainty and the specific explanation technique required to resolve it. Using a logical premise-conclusion format, the authors demonstrate that traditional interfaces demand an unreasonable prerequisite: users must translate their personal uncertainty into a technical technique selection. To solve this, the paper proposes a structural solution utilizing a multi-agent Large Language Model (LLM) orchestration tool that automatically maps natural-language user queries to the appropriate XAI explanation technique.

Explainable AI (XAI) research has successfully developed numerous explanation techniques, yet users continue to struggle to find them practically effective due to the siloed nature of conventional tools. This paper explores this issue through a philosophical lens, formalizing the selection problem—the systematic failure of XAI interfaces to bridge the gap between a user's natural-language uncertainty and the specific explanation technique required to resolve it. Using a logical premise-conclusion format, the authors demonstrate that traditional interfaces demand an unreasonable prerequisite: users must translate their personal uncertainty into a technical technique selection. To solve this, the paper proposes a structural solution utilizing a multi-agent Large Language Model (LLM) orchestration tool that automatically maps natural-language user queries to the appropriate XAI explanation technique.

Paper Metadata

  • Title: Addressing the Selection Problem in Explainable AI
  • Authors: Claire Vlases, Katelyn Morrison
  • Submitted On: August 23, 2026
  • Primary Subject: Artificial Intelligence (cs.AI)
  • Secondary Subjects: Human-Computer Interaction (cs.HC)
  • Conference Acceptance: Accepted to the Workshop on Explainable Artificial Intelligence at the International Joint Conference on Artificial Intelligence 2026 (XAI@IJCAI26)
  • Identifiers:
  • arXiv: 2608.22356 [cs.AI]
  • DOI: 10.48550/arXiv.2608.22356

Paper Metadata

  • Title: Addressing the Selection Problem in Explainable AI
  • Authors: Claire Vlases, Katelyn Morrison
  • Submitted On: August 23, 2026
  • Primary Subject: Artificial Intelligence (cs.AI)
  • Secondary Subjects: Human-Computer Interaction (cs.HC)
  • Conference Acceptance: Accepted to the Workshop on Explainable Artificial Intelligence at the International Joint Conference on Artificial Intelligence 2026 (XAI@IJCAI26)
  • Identifiers:
  • arXiv: 2608.22356 [cs.AI]
  • DOI: 10.48550/arXiv.2608.22356

Abstract

Explainable AI (XAI) research has produced a plethora of explanation techniques, yet user studies repeatedly show that available explanations are not effective in practice. We argue that, given the siloed nature of conventional XAI, users are struggling to select the appropriate XAI technique. Viewing XAI through a philosophical lens, we offer a formalization of what we call the selection problem: the systematic failure of XAI interfaces to bridge the gap between a user's natural-language uncertainty and the explanation technique that resolves it. Following a logical premise-conclusion format, we show that conventional interfaces require users to translate their uncertainty into a technique selection, a challenging prerequisite to meet. We also propose a structural solution: a multi-agent LLM orchestration tool that translates the user's query to the proper XAI explanation technique. We provide an example of how this structural solution could be instantiated to address the selection problem.

Explainable AI (XAI) research has produced a plethora of explanation techniques, yet user studies repeatedly show that available explanations are not effective in practice. We argue that, given the siloed nature of conventional XAI, users are struggling to select the appropriate XAI technique. Viewing XAI through a philosophical lens, we offer a formalization of what we call the selection problem: the systematic failure of XAI interfaces to bridge the gap between a user's natural-language uncertainty and the explanation technique that resolves it. Following a logical premise-conclusion format, we show that conventional interfaces require users to translate their uncertainty into a technique selection, a challenging prerequisite to meet. We also propose a structural solution: a multi-agent LLM orchestration tool that translates the user's query to the proper XAI explanation technique. We provide an example of how this structural solution could be instantiated to address the selection problem.

Access and Resources

Access and Resources