MIT Technology Review发布的一项调查报告指出,尽管生成式AI技术推动了企业分析的普及,但当前“代理式分析”(Agentic Analytics)生成的洞察质量仍呈现参差不齐的状态 1。数据显示,平均仅有48%的分析代理所生成的洞察能够促成改善业务的实际行动,这一比例虽然较一年前的37%有所提升,但仍有超过半数的洞察未能发挥效用 1。
阻碍洞察质量提升的主要因素包括商业背景的缺失以及数据的碎片化与不可访问性 1。具体而言,55%的高管认为缺乏商业背景是导致洞察质量低下的主要原因;另有45%的高管归咎于数据碎片化或不可访问问题 1。此外,52%的组织面临“洞察过载”的挑战,即生成的洞察数量过多导致无法有效评估和执行,而能够全面评估所有洞察的组织占比仅为24% 1。
为应对上述挑战,企业正优先加强人工智能与数据治理,60%的受访组织提及了这一举措,同时45%的组织正在添加语义层以提升效果 1。相比之下,被定义为“洞察领导者”的组织通过加强AI治理和语义层建设,实现了超过70%的有效洞察率 1。需要说明的是,该内容由MIT Technology Review定制内容部门制作,并非其编辑部的新闻稿 1。
A new report from MIT Technology Review reveals that while generative AI has broadened the accessibility of enterprise analytics, the quality of insights generated by "agentic analytics" remains inconsistent 1. On average, only 48% of AI agent-generated insights lead to actions that improve business outcomes, a figure that has increased from 37% a year earlier but still indicates that more than half fail to deliver tangible results 1. In contrast, organizations classified as "insight leaders" achieve significantly higher effectiveness, with over 70% of their agentic insights resulting in business improvements 1.
Several structural barriers contribute to these performance disparities. Fifty-five percent of executives identify a lack of business context as the primary cause of low-quality insights, while 45% point to fragmented or inaccessible data 1. The volume of output also poses a challenge; 52% of organizations struggle with "insight overload," generating too many findings to effectively evaluate and act upon, and only 24% are able to assess all insights produced 1. To address these issues, companies are prioritizing enhanced AI and data governance (cited by 60%) and the implementation of semantic layers (cited by 45%) to boost analytical efficacy 1. It is noted that this content was produced by MIT Technology Review's custom content division rather than its editorial team 1.
评论
还没有评论,欢迎留下第一条。