Recommended Free Tools
iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
AI is spreading beyond technology firms, but adoption is not the same as a return on investment, higher company-wide productivity, or fewer jobs. The evidence through 2026 shows rising use, task-level gains in some settings, and mixed results once businesses account for implementation, workflow changes, and employment. The practical answer is to evaluate AI against a defined task and business outcome, then measure what happens after deployment—not to assume that a faster task automatically means a more productive or profitable company.
How much are non-tech businesses adopting AI?
Adoption is rising, including in industries outside technology, but headline percentages describe different populations and kinds of use. The figures below should be read as separate snapshots, not as one continuous trend line.
| Measure | Reported result | What it measures |
|---|---|---|
| OECD firm-use series, 2023–2025 | 8.7% in 2023; 14.2% in 2024; 20.2% in 2025 | Firms reporting AI use in the OECD series. The OECD’s 2025 figures also show 57.3% use among ICT firms and 36.8% among professional and scientific services firms. |
| OECD enterprise implementation, 2024 | 13.9% of enterprises in the OECD area; 13.5% in EU27 | Enterprises that implemented AI, as reported by the OECD in 2025. This is a separate measure from the firm-use series. |
| U.S. Census Bureau AI supplement, November 2025–January 2026 | 18% of firms; 32% on an employment-weighted basis | Firms using AI in at least one business function, according to a 2026 Census Bureau working paper. The employment-weighted figure gives more weight to firms with more workers. |
| U.S. Census Bureau biweekly estimates, September 2023–February 2024 | 3.7% to 5.4% | Business Trends and Outlook Survey estimates of firm AI use. The paper’s expectation of about 6.6% by early fall 2024 was a forecast, not an observed result. |
These numbers cannot be lined up as if they measured the same thing. They differ by geography, survey, reference period, denominator, and definition: “using AI,” “using AI in a business function,” and “implementing AI” are not interchangeable. In particular, the OECD enterprise implementation figure and the U.S. Census business-function figure do not establish a direct comparison between regions.
Industry figures also need careful interpretation. In the OECD’s 2025 firm-use data, use grew year over year by 62.5% in accommodation and food services and 59.1% in construction. Those are growth rates in use, not adoption rates or percentage-point increases. The figures indicate rapid change from each sector’s prior level; they do not mean that 62.5% of accommodation and food firms, or 59.1% of construction firms, used AI.
#1 Best Overall
Technology firms remain ahead in the OECD’s 2025 figures, but adoption outside technology is not limited to a single kind of work. The Census Bureau’s 2026 supplement found that 57% of adopting firms used AI in three or fewer business functions, and 66% of users relied on it solely to augment tasks. That points to many businesses adopting AI in a narrow part of operations rather than transforming the whole organization.
What does AI adoption mean for ROI?
There is no universal AI ROI figure in this evidence. A company’s return depends on what it deploys, where it fits into the workflow, how much implementation costs, and whether a local improvement turns into a measurable business outcome.
ROI is a measurement problem before it is a vendor claim. Before a pilot, define the business result you want and record a baseline. A faster drafting or search task may be useful, but it does not by itself prove better margins, more output, or improved service. Measure the outcome the organization actually values.
Build an evaluation around the whole workflow
- Choose a bounded use case. Specify the task, the people who perform it, and whether AI is intended to assist workers or automate part of the task. A narrow, well-defined use case makes it easier to compare results with the existing process.
- Set a baseline before rollout. Record current time, throughput, quality or reliability, and any relevant costs. Pick measures suited to the task; for example, time saved is not an adequate measure if review work or error correction increases elsewhere.
- Count the full cost of implementation. Include the effort to integrate AI into the process, train staff, prepare data or infrastructure, and oversee the system. The evidence identifies complementary investment and organizational change as relevant, but does not provide a single cost schedule or universal ROI formula.
- Track task results and business results separately. Measure whether the task became faster or better, then whether that change improved a wider result such as output, service reliability, or profitability. A bottleneck elsewhere in the process can absorb task-level time savings.
- Reassess after the process has settled. A pilot-period result and a longer deployment can differ. Allow for adjustment, workflow redesign, and training; do not treat an early result as proof of mature, company-wide performance.
One indication of why firms adopt AI is not the same as evidence that they received a return. In the U.S. Census Bureau’s 2025 summary of the 2023 Annual Business Survey, 45.8% of surveyed AI adopters cited improving the quality or reliability of processes or methods as a motivation. The survey covers businesses’ reported technology and workforce changes from 2020 to 2022; this percentage describes a reason for adoption, not a measured ROI.
Do AI task gains translate into higher company productivity?
Sometimes, but not automatically. The International Labour Organization’s May 2026 brief reports task-level productivity gains typically in a 10%–70% range across the studies it reviewed, with stronger effects for less-experienced workers and well-defined, text-intensive tasks. That wide range summarizes heterogeneous task-level evidence; it is not a forecast, a guaranteed result, or a firm-wide productivity estimate.
A person completing one task faster may create capacity without increasing total company output. Work may shift to a later step, require additional review, or change in quality. If the business does not redesign the process or use the freed capacity productively, a local time saving may never appear in company-wide productivity statistics.
The ILO’s companion brief describes this as an “aggregation paradox”: strong improvements at the task level have not yet become a clear AI-driven increase in aggregate productivity in official statistics. Its 2026 review says reported gains are real but uneven and often unverified; worker-reported time savings have not consistently become higher measured output, earnings, or employment. Complementary investment, skills, broader diffusion, and organizational change help explain why task results and economy-wide results can diverge.
Evidence from manufacturing illustrates the adjustment problem, but should not be generalized to every sector. A U.S. Census Bureau working paper on industrial AI finds short-run performance losses before longer-run gains in the firms it studies. Results vary with firm characteristics and production practices. The finding supports looking at the deployment period and operating context; it does not establish that every manufacturing adoption, or every non-tech AI project, follows the same path.
Is AI replacing workers in non-tech industries?
The evidence does not support a blanket answer that AI is either replacing workers or creating jobs across all non-tech industries. Survey findings often show no overall change in worker counts during the periods studied, while projections and emerging evidence point to possible changes in labor demand and job opportunities.
The U.S. Census Bureau’s 2025 summary of its 2023 Annual Business Survey says adopting businesses most often reported no overall change in worker counts between 2020 and 2022. The summary’s analyst, Rachel Arledge, said, “The adoption of new technology like robotics and artificial intelligence (AI) had little impact on the number or skills of workers that businesses employ.” This is a description of the survey’s reported changes over that period—not evidence that AI will never affect employment or that every adopting business had the same outcome.
In the Census Bureau’s 2026 AI supplement, AI-related employment decreases were rare at the firm level. The working paper also found different associations depending on whether AI was integrated across business functions or used in workers’ tasks. Those differences caution against treating “AI adoption” as a single exposure: a business using AI in one narrow workflow is not equivalent to one changing how work is organized across multiple functions.
The ILO’s 2026 review likewise describes large-scale displacement as limited in the evidence it synthesizes so far, while flagging possible pressure on opportunities for younger workers and changes in work organization. “Limited so far” describes the evidence available to that review; it is not a guarantee about future effects or a prediction for every occupation.
Rank #3
Labor projections are a different kind of evidence. The U.S. Bureau of Labor Statistics projects that healthcare and social assistance will grow 8.4% and add roughly 2.0 million jobs from 2024 to 2034. That is a sector-wide labor-market projection, not an AI-caused increase. Separately, BLS expects growing AI adoption and productivity gains to dampen labor demand in some fields, including sales, design, and administrative support. These are forecasts about demand, not observed employment outcomes attributable to AI.
Why do adoption and results vary by industry?
Industries do not share one adoption path because their work, firms, and operating constraints differ. A digital task with clear inputs and outputs may be easier to test than a process tied to specialized equipment, service delivery, or multiple handoffs. Adoption rates alone do not say how deeply AI is integrated or whether it improves results.
The OECD’s 2025 figures show different levels across industries: 57.3% of ICT firms and 36.8% of professional and scientific services firms reported use. Meanwhile, accommodation and food services and construction recorded strong year-on-year growth in use. These observations show variation in levels and growth, but they do not explain the performance of an individual business or establish which sector has the best ROI.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitchesWhen comparing businesses or industries, keep five distinctions in view:
- Definition and denominator: Is the figure about any AI use, use in a business function, or implementation of an AI solution? Is it firm-weighted or employment-weighted?
- Industry and firm profile: Sector, organization size, digital maturity, and production practices can affect both adoption and outcomes.
- Task scope: AI may augment or automate work in writing, search, analysis, marketing, or production. The number and type of functions involved matter.
- Outcome and time horizon: Task time, quality, output per worker, profitability, reliability, and employment measure different outcomes. A pilot and a mature deployment are different stages.
- Implementation complements: Training, workflow redesign, data and cloud infrastructure, production practices, and oversight can influence whether use scales.
A 2026 BLS working paper offers another adoption-related finding, but not a causal result: a one-standard-deviation increase in AI exposure is associated with a 4–11 percentage-point higher adoption probability. After year and industry controls, the association falls to 1–8 percentage points. The authors characterize exposure as informative but incomplete; the estimates describe an association in a working paper, not proof that exposure causes adoption or improves performance.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What should a business decision-maker conclude?
Use adoption statistics to understand diffusion, task studies to identify possible local gains, and firm-level measures to assess business performance. Do not substitute one for another. The evidence shows that AI use is increasing, that some workers and tasks can become more productive, and that firm-wide gains depend on implementation and organizational conditions.
For a decision about a particular deployment, the most useful question is not whether AI is “good for productivity” in general. It is whether this defined use case improves the target outcome after the organization accounts for cost, quality, workflow effects, and the time needed to adapt—and whether that result persists beyond an initial pilot.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Frequently Asked Questions
Does AI help less-experienced workers more than experts?
The ILO’s May 2026 brief says reviewed studies found stronger task-level effects for less-experienced workers, particularly on well-defined text-intensive tasks. That does not establish that every novice benefits more, or that the difference carries through to company-wide output.
Why do some AI adoption statistics use an employment-weighted rate?
An employment-weighted rate gives more influence to firms that employ more people, so it answers a different question from the share of firms adopting AI. In the U.S. Census Bureau’s November 2025–January 2026 supplement, that distinction separates the 18% firm figure from the 32% employment-weighted figure.
Does AI exposure cause a business to adopt AI?
The BLS working paper reports an association between measured exposure and adoption probability, not a causal effect. Exposure is also described by the authors as informative but incomplete, so it should not be treated as a full measure of which work AI can affect.
Are employment projections evidence that AI has already reduced jobs?
No. A labor projection describes an expected future pattern under its assumptions, while observed survey responses describe what businesses reported during a past period. The BLS’s sector growth and labor-demand projections therefore should not be read as counts of jobs created or lost because of AI.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →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.
