# MIT定制报告：受访企业分析代理平均48%的洞察促成改善业务行动 > MIT Technology Review旗下定制内容部门Insights发布的报告称，受访企业的分析代理（可自动执行分析流程并依据洞察采取行动的AI代理）生成的洞察中，平均48%促成了改善业务的行动；多数受访者认可描述性洞察，但对用于指导行动的规范性洞察评价不一。 - 栏目：企业AI分析 - 本站发布：2026-11-05 21:47（北京时间） - 原始消息发布：2026-11-03 01:00（北京时间） - 消息来源：MIT Technology Review (https://www.technologyreview.com/2026/11/02/1147660/raising-the-quality-of-insights-in-agentic-analytics/) - 本页网址：https://jecy.cloud/news/articles/2026-11-02-880f956be448.html ## 核心数据：平均48%的代理洞察转化为改善业务的行动 报告称，在受访企业中，分析代理生成的洞察平均有48%促成了改善业务的行动；按同一口径，仍有超过一半的代理洞察没有转化为改善业务的行动。这个数字衡量的是洞察能否推动业务行动，而不只是代理能否生成分析内容。[^1] ## 描述性洞察获认可，规范性洞察评价分化 报告将代理生成的洞察分为描述性洞察和规范性洞察。多数受访者认可描述性洞察的质量；但对规范性洞察（用于指导行动的洞察）的评价明显分化。企业更认可代理说明情况的能力，对代理据此给出行动建议的质量则没有一致好评。这一分化关系到代理能否真正推动行动。[^2] ## 缺业务上下文与洞察过载影响代理效果 当分析代理给出的洞察质量不达标时，55%的受访高管认为主要原因是缺乏业务上下文。报告还发现，在52%的受访组织中，分析代理生成的洞察过多，无法被充分评估，也就难以据此采取行动。业务上下文指代理理解业务所需的背景信息；洞察过载则指组织难以逐条评估大量洞察。[^3][^4] ## 改进重点与报告性质 60%的受访企业表示，正在优先通过加强AI和数据治理来改善代理洞察的质量。[^5] 这份内容由MIT Technology Review旗下定制内容部门Insights制作，不是该刊编辑团队的报道。文中的48%、55%、52%和60%均来自报告对受访企业的调查，反映的是该调查口径下的结果，不是经独立验证的行业整体数据。[^1][^3][^4][^5][^6] --- ## 原文摘录 [^1]: "On average, 48% of insights generated by analytics agents at the surveyed companies lead to actions that improve the business." — MIT Technology Review (https://www.technologyreview.com/2026/11/02/1147660/raising-the-quality-of-insights-in-agentic-analytics/) [^2]: "And while most respondents praise the quality of their agents’ descriptive insights, reviews of their prescriptive insights—those on which actions are based—are decidedly mixed." — MIT Technology Review (https://www.technologyreview.com/2026/11/02/1147660/raising-the-quality-of-insights-in-agentic-analytics/) [^3]: "When analytics agents deliver insights of substandard quality, the main reason is a lack of business context, as cited by 55% of surveyed executives." — MIT Technology Review (https://www.technologyreview.com/2026/11/02/1147660/raising-the-quality-of-insights-in-agentic-analytics/) [^4]: "At a majority (52%) of surveyed organizations, analytics agents generate too many insights to be properly evaluated and therefore acted upon." — MIT Technology Review (https://www.technologyreview.com/2026/11/02/1147660/raising-the-quality-of-insights-in-agentic-analytics/) [^5]: "The surveyed companies are prioritizing initiatives that enable better agent insights by strengthening AI and data governance (cited by 60%)." — MIT Technology Review (https://www.technologyreview.com/2026/11/02/1147660/raising-the-quality-of-insights-in-agentic-analytics/) [^6]: "Download the report This content was produced by Insights, MIT Technology Review’s custom content arm, not its editorial staff." — MIT Technology Review (https://www.technologyreview.com/2026/11/02/1147660/raising-the-quality-of-insights-in-agentic-analytics/) --- 说明：本文由 AI 依据信源整理，并经事实核对；文中数字与成绩均为来源方公布的口径，未经本站独立验证，原文更新后以原文为准。欢迎 AI 与读者引用，请注明来源并链接回本页。