OpenAI 调查显示，在约 9,000 名、遍及 100 间公司的受访员工中，使用其生成式 AI 工具的上班族每天可在专业工作上平均节省约 40 至 60 分钟。资料科学、工程、传播与会计等职能节省时间最多，约有四分之三的员工表示 AI 有提升产出速度或品质。

然而，即使在所谓「 AI 热潮」展开三年后，对于生产力提升程度的质疑仍然存在。麻省理工学院研究指出，多数企业在生成式 AI 投资上几乎没有回报；哈佛与史丹佛学者更提出「workslop」概念，认为大量 AI 内容只是看似体面却缺乏实质贡献，加深市场对 AI 泡沫的忧虑。

为回应质疑，OpenAI 与竞争对手积极公布自家研究以凸显经济效益。Anthropic 宣称，其 AI 工具 Claude 可将完成特定工作任务所需时间削减 80%，这一估计来自对 10 万笔用户对话的分析；OpenAI 则表示已有逾 100 万家企业付费采用，ChatGPT 职场产品的付费席次达 700 万，企业导入速度甚至在部分情况下快于终端消费者。

OpenAI reports that among roughly 9,000 employees across 100 companies using its generative AI tools, workers save an average of about 40 to 60 minutes per day on professional tasks. Time savings are highest in data science, engineering, communications, and accounting, and about three quarters of employees say AI improves either the speed or quality of their output.

Despite three years of the AI boom, doubts about productivity gains remain. A Massachusetts Institute of Technology study found that the vast majority of organizations saw essentially zero return on their generative AI investments. Harvard and Stanford researchers warn about workslop, AI-produced content that looks polished but does not meaningfully advance the underlying task, fueling concerns about an AI investment bubble.

In response, leading AI firms are releasing their own impact studies. Anthropic claims its AI assistant Claude can reduce the time to complete certain work tasks by 80%, based on an analysis of 100,000 user conversations, though these reports are not peer-reviewed. OpenAI says more than 1 million businesses now pay for its enterprise products, with 7 million paid workplace ChatGPT seats, and that enterprise adoption is accelerating, sometimes even faster than consumer uptake.