Scholarlink

<p>Artificial intelligence (AI) has become integral to organizational operations, yet its real-world deployment increasingly exposes critical points of failure. Despite growing interest, existing studies on AI failure remain fragmented, often focusing narrowly on technical flaws, interactional breakdowns, or ethical concerns. This scoping review analyzes 141 studies to consolidate current knowledge on AI failure. It identifies four research clusters and organizes them into three analytical categories: technical, interactional, and ethical. Building on this structure, the study develops a Subtypes–Causes–Mitigation (SCM) Framework that links failure subtypes, root causes, and mitigation strategies across the three categories. Finally, it outlines a research agenda to advance the understanding and management of AI failures. The findings provide a conceptual foundation for cumulative theorizing and offer practical guidance for improving the application of AI systems in real-world contexts.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Synthesizing AI Failure Research: A Scoping Review

  • 期刊论文

  • 发表日期：

    2025年10月27日

查看全文
  • Chenxi Li,
  • Yixun Lin,
  • Xinyi Tu,
  • Jing Elaine Chen,
  • Ziqi Zhao

摘要

Artificial intelligence (AI) has become integral to organizational operations, yet its real-world deployment increasingly exposes critical points of failure. Despite growing interest, existing studies on AI failure remain fragmented, often focusing narrowly on technical flaws, interactional breakdowns, or ethical concerns. This scoping review analyzes 141 studies to consolidate current knowledge on AI failure. It identifies four research clusters and organizes them into three analytical categories: technical, interactional, and ethical. Building on this structure, the study develops a Subtypes–Causes–Mitigation (SCM) Framework that links failure subtypes, root causes, and mitigation strategies across the three categories. Finally, it outlines a research agenda to advance the understanding and management of AI failures. The findings provide a conceptual foundation for cumulative theorizing and offer practical guidance for improving the application of AI systems in real-world contexts.