
AI支出呈上升趋势。斯坦福大学2026年人工智能指数报告显示,2025年,88%的受调查机构在至少一项业务职能中使用了人工智能,而KPMG最新的美国人工智能季度脉搏调查发现,领导者预计未来12个月将在人工智能上平均投资1.73亿美元。
Spending on AI is on an upward trajectory. Stanford University’s 2026 AI Index reports that 88% of surveyed organizations used AI in at least one business function in 2025, while KPMG’s latest US AI Quarterly Pulse Survey found that leaders expect to invest an average of $173 million in AI over the next 12 months.
随着这一增长,围绕衡量的新问题无疑随之而来。生产力提升是人工智能收益中最容易量化的之一,因此一直主导着讨论。但假设这些收益本身就代表良好的投资回报率存在风险。此外,在如此高的支出水平下,领导者现在被要求证明人工智能在绩效、增长和更长期的业务变革方面的影响,而不仅仅是成本节约。
With this increase undoubtedly comes new questions around measurement. Productivity gains are among the easiest AI benefits to quantify, which is why they’ve tended to dominate the discussion. But there’s risk in assuming those gains alone represent a good return on investment. Plus, with this level of spending, leaders are now being asked to justify the impact of AI across performance, growth, and longer-term business change, not just cost savings.
“如果领导者仅从成本削减的角度评估人工智能,他们可能会忽略客户关系、创新步伐、决策质量以及全新收入来源中的价值,”KPMG美国人工智能企业转型负责人拉赫桑·希尔斯表示。
“If leaders evaluate AI solely through the lens of cost reduction, they risk overlooking value in customer relationships, the pace of innovation, decision-making quality, and entirely new sources of revenue,” says Rahsaan Shears, AI Enterprise Transformation Leader at KPMG US.
认识到人工智能的一些最重要影响可能不那么直接或不那么容易衡量,也意味着认识到它们可能不会整齐地体现在短期财务结果中。对于领导者来说,这带来了一个新的挑战:如何使用比传统衡量标准所能提供的更广泛的价值观来评估人工智能。
Understanding that some of AI’s most significant effects may be less immediate or straightforward to measure also means recognizing they may not show up neatly in short-term financial results. For leaders, that creates a new challenge around how to evaluate AI using a broader view of value than traditional measures alone can provide.
KPMG认为,现代化人工智能评估的一种方法是将投资视为一个投资组合。与其将人工智能作为一个单一的、同质的类别来衡量,不如将不同的应用场景分开,这可以让领导者更清晰地了解价值在哪里创造,以及应该如何衡量。
KPMG argues that one way to modernize AI evaluation is to treat investments as a portfolio. Rather than measuring AI as a single, homogeneous category, separating different use cases can give leaders a clearer picture of where value is being created, and how it should be measured.

这是因为有些举措旨在推动运营改进并带来短期回报,而另一些则专注于通过加速产品开发或改善用户体验来释放增长。最雄心勃勃的努力更进一步,彻底改变整个运营模式,即产品和服务交付给客户的方式。
That’s because some initiatives are designed to drive operational improvements and generate near-term returns, while others focus on unlocking growth by accelerating product development or improving the user experience. The most ambitious efforts go even further, revolutionizing entire operating models, or the way products and services are delivered to customers.
“创造最大价值的组织不会在人工智能上押单一赌注,”希尔斯说。“他们在多个时间维度上平衡投资,并以不同方式衡量成功。”一旦领导者了解人工智能投资旨在解决的具体问题,成功的正确衡量标准就会变得更加清晰。面向客户的人工智能系统可能通过留存率或销售额来衡量,而内部工具则可能根据节省的时间或产能进行评估。与此同时,产品开发系统可能会根据上市速度来评判。将指标与使用场景相匹配,能让领导者更准确地了解一项投资是否实现了其设计目标。
“The organizations creating the most value aren’t making a single bet on AI,” Shears says. “They’re balancing investments across multiple horizons and measuring success in different ways.”Once leaders understand the specific problem an AI investment is intended to solve, the right measure of success becomes clearer. A customer-facing AI system might be measured through retention or sales, while an internal tool could be assessed on time saved or capacity. A product-development system, meanwhile, might be judged on speed to market. Matching the metric to the use case gives leaders a more accurate view of whether an investment is delivering the outcome it was designed to achieve.
但技术本身并不决定人工智能是否能创造价值。员工如何使用它,可能会对企业从投资中获得的结果产生重大影响。正如希尔斯所说,“优势不在于仅仅拥有人工智能的使用权。优势来自于知道如何与它协作。”德勤会计师事务所与德克萨斯大学奥斯汀分校的一项联合研究分析了2597名员工在八个月内的超过140万次工作场所人工智能互动。研究人员发现,最强的使用者并非仅仅是技术专长最强的人。他们更有可能仔细界定问题、质疑输出结果、优化回复,并在行动前运用自己的判断。
But technology alone doesn’t determine whether AI delivers value. How employees use it can have a significant impact on the results a business gets from its investment. As Shears says, “The advantage isn’t simply having access to AI. The advantage comes from knowing how to work with it.”A joint study by KPMG and The University of Texas at Austin analyzed more than 1.4 million workplace AI interactions from 2,597 employees over eight months. The researchers found that the strongest users were not simply those with the most technical expertise. They were more likely to frame problems carefully, challenge outputs, refine responses, and apply their own judgment before acting.

这有助于解释为什么劳动力技能正在成为人工智能战略中越来越重要的一部分。毕马威第三季度2026年人工智能脉搏调查发现,65%的领导者预计人工智能代理将推动将人工智能协作技能纳入工作流程和职位要求,而制定与人工智能代理协作的职位特定指南的比例已从21%上升至55%,同比增长。
That helps explain why workforce skills are becoming a bigger part of AI strategy. KPMG’s Q3 2026 AI Pulse Survey found that 65% of leaders expect AI agents to drive the integration of AI collaboration skills into workflows and role requirements, while the share developing role-specific guidelines for working with AI agents has risen from 21% to 55% year over year.
毕马威本身一直通过其aIQ计划在内部应用这一原则,该计划旨在将人工智能融入日常工作。负责aIQ的希尔斯表示,该举措将人工智能工具与培训和治理相结合,包括aIQ学院,并利用内部学习成果来指导其与客户的工作。
KPMG itself has been applying that principle internally through aIQ, its program for integrating AI into everyday work. Shears, who heads up aIQ, says the initiative combines AI tools with training and governance, including an aIQ Academy, and uses what it learns internally to inform its work with clients.
随着人工智能支出的增加,理解其回报意味着认识到不同的应用创造不同的价值。明确目标,选择合适的衡量标准,并确保员工具备实现目标所需的技能。这可以让企业更清晰地了解人工智能在哪些方面发挥作用,以及下一步应在哪里投资。
As AI spending rises, understanding its return means recognizing that different applications create different kinds of value. Define the outcome, choose the right measure, and make sure employees have the skills to deliver it. That can give businesses a clearer view of where AI is working and where to invest next.
