在职场中过度使用生成式AI完成工作任务正在引发质量问题和认知能力下降的隐患。南非国家AI政策草案因包含AI生成的虚假引用而被撤回1，澳大利亚德勤公司也因向政府提交含有伪造引文的报告而被罚款，该报告耗资440,000澳元1。这些案例凸显了在实际工作中应用AI所面临的严重风险。
2025年墨尔本大学和毕马威联合开展的研究调查了超过48,000名受访者，覆盖47个国家，发现66%的人定期使用AI进行工作1。然而，调查也发现61%的受访者未曾接受过AI使用培训，60%的职场存在不当、自满和不透明的AI应用现象1。研究人员指出，在接纳AI的动力方面，48%的采用来自害怕被淘汰的担忧1。由旧金山AI安全中心在2026年7月进行的研究发现，顶级AI代理在约85%的项目上未能达到受委托工作的可接受质量标准1，进一步证实了当前AI应用面临的质量困境。
Heavy dependence on generative AI to complete workplace tasks carries significant risks, from compromised work quality to diminished cognitive abilities.1 A South African draft national AI policy was withdrawn after it contained AI-generated false citations, while Australian consulting firm Deloitte provided a partial refund to the government following the discovery of fabricated references in a report that cost A$440,000 to produce.1
Recent research sheds light on the prevalence and dangers of this trend. A 2025 study by the University of Melbourne and KPMG surveying over 48,000 respondents across 47 countries found that 66% of people regularly use AI, with 48% driven by fear of falling behind.1 However, adoption has outpaced safeguards: 61% of respondents reported receiving no AI training, and 60% described improper, complacent, and opaque AI use in their workplaces.1 More concerning still, research from San Francisco's AI safety center found that leading AI agents failed to meet acceptable quality standards on approximately 85% of commissioned projects.1
To protect cognitive capacity and work quality, experts recommend extending delivery timelines, conducting manual brainstorming sessions, and fostering a culture of critical evaluation.1
评论
还没有评论，欢迎留下第一条。