Most workers quietly hide using AI at work — so how would any company know if it's actually helping?
Why do most organizations lack reliable data on AI's actual impact on productivity?
This explores why companies struggle to get trustworthy numbers on whether AI is actually making their people more productive, and what in the corpus explains that gap.
This explores why companies struggle to get trustworthy numbers on whether AI is making their people more productive. The corpus points to a problem with several layers. The tools people use to measure, the people being measured and the place where gains are supposed to appear all skew the result, and each pushes in a different direction.
Start with the people being measured. Many workers hide their AI use. In one interview study, most professionals and creatives said AI saved them time, yet about 70% actively concealed or downplayed using it because of workplace stigma and worries about their professional identity Why do workers hide productivity gains from AI use?. Controlled experiments explain why: people expect colleagues and managers to see them as less competent and less diligent if they admit to using AI, so they tell them less Do people fear judgment when they use AI at work?. An organization can't measure what its workers won't report. Much of the real usage is invisible to leadership.
When people do report, their reports don't match what gets measured. A randomized trial with experienced open-source developers found AI tools made them 19% slower, even though the developers predicted beforehand that they would be 24% faster. Outside experts in economics and machine learning made the same mistake Do AI coding tools actually speed up experienced developers?. At the executive level, a survey of 750 leaders found that the gains people perceived were larger than the measured ones, partly because revenue lags behind day-to-day operational improvements Do AI productivity gains feel larger than they actually measure?. Most corporate data on AI comes from self-report surveys, and self-report is the measure the corpus shows to be least reliable.
The less obvious problem is that gross time saved and net time saved are very different numbers. Workday's survey found that nearly 40% of AI time savings are lost to fixing errors and checking outputs, and only 14% of employees consistently come out ahead Where does AI's time savings actually go in practice?. Zapier found the same pattern: 92% of users report a productivity boost, yet the average worker spends 4.5 hours a week cleaning up AI output. The heaviest, best-trained users clean up the most How much time do workers really spend fixing AI mistakes?. Adding tools can make the picture worse. BCG found that self-reported productivity peaks at three AI tools and then falls, because the work of overseeing them piles up Does using more AI tools always boost worker productivity?. Mollick argues that much of this cost comes from interface design, not model quality. Chatbot interfaces add mental overhead that cancels out real gains, especially for less experienced staff Is the AI capability gap really an interface problem?.
Finally, organizations may be measuring at the wrong level. The strongest positive studies measured people doing tasks they already knew how to do. When people used AI to learn something new, the gains disappeared When does AI actually boost worker productivity?. A single number for the whole workforce blends those two very different situations. Narayanan and Kapoor argue that AI compresses only the middle 'execution' part of knowledge work, while deciding what to do and delivering the result stay the same size or grow. A faster task may therefore never show up as a more productive job Does AI really compress all layers of knowledge work equally?. Microsoft argues the real frontier is team productivity, which almost no one tracks, though its report offers no evidence for that claim yet Can AI boost how teams work together?. Economists can't agree on the national numbers either. Brynjolfsson sees a 2.7% productivity surge in 2025 data, while others say AI still has no clear signature in jobs, productivity or earnings outside the tech leaders Is AI productivity finally showing up in economic data?. If economists studying the whole economy can't settle the question, it's unsurprising that a single firm, relying on surveys of workers who conceal their use, can't either.
Sources 12 notes
In a 1,250-person interview study, 86% of general workers and 97% of creatives said AI saved them time, yet 69–70% actively hid or downplayed their use due to workplace stigma and concerns about professional identity and economic displacement.
Across four experiments with 4,439 participants, people using AI expected others to judge them as less competent and diligent, and reported lower willingness to disclose AI use to managers and colleagues. The gap suggests a social cost that users foresee and act on.
A randomized controlled trial of 16 developers on 246 real tasks found completion times increased 19%, despite developers forecasting a 24% speedup beforehand. Experts in economics and ML also overestimated gains; slowdown factors included over-optimism, low AI reliability, and developers' deep familiarity with mature codebases.
A survey of 750 executives found that perceived AI productivity gains exceed measured ones, likely because revenue lags operational improvements. Effects concentrate in high-skill services and finance, with labor reallocating rather than shrinking overall.
A Workday-commissioned survey of 3,200 active AI users found that while 85% save 1–7 hours weekly, almost 40% of those savings disappear into correcting errors and verifying outputs. Only 14% of employees consistently see positive net outcomes, with success tied to organizations that retrain staff and redesign roles rather than simply deploying tools.
Show all 12 sources
A Zapier survey of 1,100 enterprise AI users found 92% report productivity boosts, yet the average worker spends over half a day weekly revising AI-generated work. Trained, heavy users report the largest gains but also spend the most time on cleanup.
BCG's survey found self-reported productivity rose with up to three AI tools but fell sharply with four or more. Workers experiencing this 'brain fry' showed 34% quit intent versus 25% without it, driven by oversight burden rather than tool count alone.
Mollick argues that better interfaces—not better models—will drive perceived capability leaps. Evidence includes a cognitive-load study showing financial professionals gained productivity from GPT-4 but lost it to chatbot design's cognitive overhead, especially hurting less experienced users.
Studies showing AI productivity gains measured tasks within workers' existing domains. When workers used AI to learn new skills, productivity gains disappeared and learning suffered, suggesting prior findings do not generalize to skill acquisition.
Narayanan and Kapoor argue AI narrows only the middle execution layer of knowledge work while decide and deliver layers persist or grow. Translation and legal work show stable or expanding employment despite AI gains, suggesting task-level compression doesn't shrink occupational demand.
Microsoft's 2025 report argues the next AI frontier is collective productivity, requiring systems built around shared goals and collaboration norms rather than individual tools. The claim frames this as a deliberate design mandate, though the excerpt provides no empirical evidence of collective-productivity gains.
Brynjolfsson argues that slower job growth alongside GDP expansion in 2025 indicates a productivity surge of 2.7%, suggesting AI has moved from experimentation to structural utility. However, other economists dispute this reading, noting AI lacks clear signature in employment, productivity, and earnings data outside tech leaders.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- What 81,000 people told us about the economics of AI
- Toward Measuring AI's Effects on Skill Formation: The Stock-Formation Gap
- Beyond Productivity: Measuring the Real Value of AI
- How AI Impacts Skill Formation
- Adoption of Generative AI in the Workplace: Increasing and Shifting the Balance of Productivity and Communication Activity
- Artificial Intelligence, Productivity, and the Workforce: Evidence from Corporate Executives
- We are Changing our Developer Productivity Experiment Design
- Firm Data on AI