超越生产力：衡量人工智能的真正价值

A new blueprint for the next era of AI-fueled organizational growth. Table of Contents Executive Summary3 A Hidden Productivity Drag HoldingBack Workforce Transformation4 The Demographics of the Drain:Heavy AI Users Pay the Highest Price5 The People Imperative: Reinvest AIGains into the Workforce9 Methodology 12 Executive summary. 3 things leadersshould know to getmore out of AI. The promise of AI in the workplace has largely been framed aroundproductivity—saving time, automating routine tasks, and helpingemployees do more in less time. New research from Workday showsthat while many organizations are realizing these gains, a substantialshare of that value is being quietly lost to rework and low-quality output. 1.Nearly 40% of AI’s promisedproductivity is silently lostto rework, reducing the netvalue of efficiency gains. 2.The most enthusiasticusers often carry thehighest burden, spendingdisproportionatetime verifying andcorrecting output. Therefore, productivity gains alone are not translating into betteroutcomes for most organizations.While AI is helping employeescomplete tasks faster, far fewer are using it to improve the quality oftheir work or support higher-value judgment and decision-making. Infact, only 14% of employees consistently achieve net-positive outcomesfrom AI use. 3.Organizations thatreinvest AI gains into theirpeople outperform thosethat reinvest primarilyin technology, achievingstronger outcomes andmore sustainable value. As a result, roughly 37% of the time saved through AI is being offsetby rework.Employees report spending significant time correcting,clarifying, or rewriting low-quality AI-generated content—essentiallycreating an AI tax on productivity. For every 10 hours of efficiencygained through AI, nearly 4 hours are lost to fixing its output. 1.5 weeks The amount of time lost to fixing AI outputs,per highly engaged employee, per year This hidden loss highlights a critical blind spot in how organizationsassess AI performance. Most leaders focus ongross efficiency—howmuch time AI saves. But this metric alone obscures the real picture.When time lost to rework is taken into account, the net value of AI isoften much lower than expected. To capture AI’s real return, organizations must move beyondmeasuring hours saved and begin accounting for outcomes achieved.Net value—time saved minus time lost—provides a more accurateview of whether AI is improving how work gets done, or simplyaccelerating activity without improving results. The research shows that low-quality AI output is not limited to aspecific industry or region. It appears wherever AI is adopted withoutcorresponding changes to skills, role design, and support. At the sametime, the data reveals clear patterns that distinguish employees whoconsistently generate net gains from those who absorb the cost ofrework—and points to specific actions organizations can take toclose that gap. A hidden productivitydrag holdingback workplacetransformation. The erosion of AI’s value is not abstract—it shows up in everyday work.Teams are moving faster but not necessarily further. Consider a common scenario: a team is preparing materials for ahigh-stakes meeting with senior leadership. A slide deck generated bya newly adopted AI tool arrives quickly and looks polished at first glance.But the narrative lacks context, the data sources are unclear, and thetone misses the audience. What should have been a time-saving startingpoint becomes a multi-day effort to verify, rewrite, and align the work. In moments like these, AI delivers speed—but not the strategic lift itpromised. Instead of reallocating time toward judgment, creativity, anddecision-making, employees spend it correcting low-quality output. Atscale, this pattern compounds, translating into millions of lost hourseach year in large organizations. This drag is structural, not individual. Employees are not misusing AI; theyare operating within systems that have not kept pace with its adoption. Skills, roles, and the sourceof rework. While two-thirds of leaders (66%) cite skills training as a top investmentpriority, that investment is not consistently reaching the employeesmost exposed to rework. Among employees who use AI the most,only 37% report increased access to training—a nearly 30-pointgap between stated intent and lived experience. As a result, manyemployees are expected to produce higher-quality outcomes with AIwithout the guidance or support needed to do so efficiently. of roles are reported asAI-ready in organizationsstruggling to achieve netproductivity gains The issue is compounded by lagging role design. Across the full sample,nearly 9 in 10 organizations report that fewer than half of their roles havebeen updated to include AI-related skills. AI has been layered onto rolesthat were never updated to accommodate it—forcing employees to use2025 tools within 2015 job structures. Rather than reducing effort, thismismatch often increases it, as employees are l