Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallOutdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchNearly 6,000 senior executives reported that AI had made little measurable difference to productivity at most of their firms over the previous three years. That is a warning about the gap between AI adoption and results—not proof that AI delivered zero value or that $250 billion was lost. The survey and the investment figure come from separate datasets and do not provide a matched calculation of spending versus returns.
What the executive survey actually found
The 2026 National Bureau of Economic Research (NBER) working paper “Firm Data on AI” draws on responses from nearly 6,000 CEOs, CFOs and other senior executives at firms in the United States, United Kingdom, Germany and Australia. Responses were collected from November 2025 through January 2026, according to the NBER Digest summary.
- About 69% of firms said they actively use AI.
- About 89% reported no effect on labor productivity over the previous three years.
- More than 90% reported no effect on employment over that period.
- More than two-thirds of executives said they personally used AI, averaging about 1.5 hours per week.
These are executive reports, not an audit of each company’s output, payroll or AI spending. The paper asked about reported effects on productivity, employment, output and costs, as well as executives’ own use and expectations. It did not measure every employee’s performance, each AI project’s financial return, product quality, customer satisfaction, or time saved and redirected to other work. It is a working paper, so its findings should be read as evidence about surveyed firms’ reported experience, not a final verdict on every business or AI system.
Why “no productivity impact” does not mean “AI did nothing”
Productivity depends on the level being measured. A tool can help one employee complete a task faster without raising output per worker across an entire company. And company-level results do not automatically establish whether the whole economy is producing more per unit of labor and capital.
#1 Best Overall
- Task productivity: Did a person complete a particular task faster, more accurately or at better quality?
- Firm productivity: Did the business produce more measurable output per worker or per hour across the organization?
- Economy-wide productivity: Did total economic output rise relative to the inputs used?
A company might use AI to draft customer replies faster, for example, yet see no company-wide productivity change if only a small team uses it, staff spend time checking the drafts, or the saved capacity goes into handling more complex cases. Benefits may also show up as fewer errors, improved service or new products rather than more units of output per employee. Those outcomes can matter, but they are not identical to a measured increase in labor productivity.
The survey does not establish which explanation accounts for the flat reported results. Limited adoption, workflow changes still in progress, human review, integration costs and benefits that are difficult to capture in a single productivity measure are plausible possibilities—not findings that the survey proves.
Adoption is not the same as deep, productive use
A firm can count as an AI user because employees have access to a tool, have tried it or use it in a narrow task. That does not tell us whether the company has redesigned a recurring workflow, connected AI to relevant data, trained staff, or measured whether the change improves business results.
The reported average of about 1.5 hours of executive AI use per week underscores why adoption rates alone can overstate how deeply AI is embedded in work. A license, an occasional experiment and a repeatable production process are different stages. The survey’s roughly 69% firm-use figure should not be read as meaning that AI is central to most firms’ operations.
What the $252.3 billion figure means—and what it does not
Stanford’s 2025 AI Index estimated global corporate AI investment at $252.3 billion in 2024. The figure describes a broad investment measure; it is not a tally of productivity-software subscriptions, operating expenses at the surveyed firms, or money all directed toward workplace tools. Stanford separately estimated private generative-AI investment at $33.9 billion in 2024, a narrower measure.
The investment estimate and the NBER survey are not matched. They cover different data and answer different questions, so the survey cannot show that $252.3 billion was wasted, or calculate the return on that sum. It does raise a reasonable question about how quickly a large AI investment boom is translating into broadly visible firm-level productivity gains.
For context, Stanford’s separate 2025 AI Index dataset found that 78% of surveyed organizations reported using AI in 2024. That adoption measure is not directly comparable with the NBER survey’s later estimate: the surveys differ in timing, samples and methods. Stanford’s 2026 AI Index economy chapter also reports that global corporate AI investment more than doubled in 2025. Rising investment and weak reported productivity can coexist; spending is not itself proof of realized returns.
Executives report weak past results but expect future gains
The NBER study also asked executives to forecast AI’s effects over the next three years. They expected productivity to rise by about 1.4%, output by 0.8% and employment to fall by 0.7%, according to the Stanford Institute for Economic Policy Research summary.
| Measure | Reported experience over the previous three years | Executive expectation for the next three years |
|---|---|---|
| Labor productivity | About 89% reported no effect. | About 1.4% increase. |
| Employment | More than 90% reported no effect. | About 0.7% decrease. |
| Output | The survey summary does not state a comparable no-effect share. | About 0.8% increase. |
The expectations are forecasts, not measured results or guarantees. They may reflect anticipated improvements in tools, deployment and organizational change; the survey does not determine which expectations will be fulfilled.
Why task-level gains can coexist with flat firm results
Evidence from a separate field study helps explain why findings about individual tasks need not contradict the executive survey. An experiment involving more than 6,000 knowledge workers at 56 firms examined access to Microsoft 365 Copilot. The study reports meaningful adoption among workers given access, but it does not establish that companies broadly achieved higher company-wide productivity. See the field study.
A worker might finish a first draft faster while a team’s total output remains unchanged because colleagues review it, systems are not integrated, or the organization uses the time for additional work. Research on the distinction between task, firm and economy-wide effects likewise cautions against treating one level as a proxy for another; see Brookings’ discussion of AI, growth and productivity.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Why investment may arrive before measurable productivity
Some AI investment may go toward infrastructure, models, data and acquisitions rather than immediate changes to front-line workflows. Within a firm, a pilot may remain small, data may be difficult to access, or the surrounding approval process may remain the bottleneck. Faster generation can also add review and coordination work, while savings in one department shift work to IT, legal, security or quality teams.
There can also be a measurement mismatch: a firm may improve quality or customer response without increasing the volume of output, or redeploy saved labor instead of reducing headcount. These are ways investment and productivity figures can diverge; the survey does not quantify how much each contributes.
New technologies sometimes require complementary investments and changes in management and work practices before their impact appears in productivity statistics. The MIT Initiative on the Digital Economy discusses the difference between improved outputs and broader outcomes. This productivity-paradox idea is a possible explanation for a lag, not evidence that future gains are certain or that weak deployments should be excused.
How a business can test whether AI is paying off
For business owners and managers, the useful lesson is to measure a workflow’s result rather than count licenses, prompts or demonstrations. A disciplined pilot can show whether a tool improves a specific process after costs and quality checks are included.
- Choose one recurring workflow and an accountable owner. Define the work being changed and who is responsible for turning a test into an operating result.
- Record a baseline before deployment. Track existing cycle time, cost per transaction, error and rework rates, service levels or another outcome appropriate to the workflow.
- Measure quality as well as speed. Faster output is not a gain if it causes more errors, customer complaints or downstream review.
- Include the full cost. Count tool and integration costs, training, human verification, governance and added work for other teams.
- Track business outcomes and decide whether to scale. Depending on the use case, relevant measures may include revenue per employee, gross margin, customer resolution time, retention, sales conversion or product-development throughput. Scale only when a repeatable improvement is visible against the baseline.
For a household or investor trying to interpret the headline, the same discipline applies: distinguish a large investment estimate from a demonstrated loss, and distinguish adoption from verified returns. The available figures do not reveal the return on an individual company’s AI project or establish that future gains will materialize.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →The verdict on the “$250 billion black hole” claim
The evidence supports a narrower conclusion: most firms in a large four-country executive survey had not seen a measurable labor-productivity or employment effect from AI over the previous three years, even as executives expected future gains. Stanford’s $252.3 billion estimate documents the scale of global corporate AI investment in 2024, but it is not a bill for productivity tools and was not matched against those firms’ results. Calling it a proven $250 billion loss—or saying AI produced literally zero value—goes beyond what the evidence shows.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
