AI and Human Workforce Statistics 2026

Key Takeaways

McKinsey's 2025 survey found that 88% of respondents said their organizations regularly used AI in at least one business function, but only about one-third had begun scaling AI programs enterprise-wide.
Gallup found in Q2 2026 that 52% of U.S. employees used AI at work at least a few times a year, while 30% used it at least a few times a week.
An NBER field study of 5,179 support agents measured a 14% average productivity increase from an AI assistant, including 34% for novice and lower-skilled workers.
The ILO and NASK estimated in 2025 that 25% of global employment was in occupations with some generative AI exposure, with job transformation more likely than full automation.
The World Economic Forum reported that 77% of surveyed employers planned to reskill or upskill workers to operate more effectively alongside AI by 2030.

AI is common in business, but deep adoption is not. The clearest AI and human workforce statistics 2026 show a gap between access to tools and redesign of work. Employees often report help with specific tasks. Far fewer say AI has changed how their organization operates.

That distinction matters. An adoption rate does not measure productivity, and occupational exposure does not predict a job loss. This article keeps those measures separate. It draws on 2023 to 2026 surveys, experiments, and task-level labor research, with the date and population stated for each major figure.

AI and human workforce statistics at a glance

Measure	Finding	Population and definition
Organizational AI use	88%	Respondents to McKinsey's 2025 global survey who said their organization regularly used AI in at least one business function
Enterprise AI scaling	About one-third	Respondents in the same 2025 survey who said their organization had begun scaling AI programs across the enterprise
U.S. employee AI use	52%	U.S. employees in Gallup's Q2 2026 measure who used AI in their role at least a few times a year
Frequent U.S. employee use	30%	U.S. employees using AI at least a few times a week in Q2 2026
Support-agent productivity	14% average increase	Issues resolved per hour in a field study of 5,179 customer support agents
Novice and lower-skilled agent productivity	34% increase	Issues resolved per hour for that subgroup in the same field study
Global occupational exposure	25% of employment	Jobs in occupations with some potential generative AI exposure in the ILO and NASK 2025 index
High-income occupational exposure	34% of employment	Same index, limited to high-income countries
Reskilling response	77% of employers	Employers planning reskilling or upskilling so staff can work more effectively with AI by 2030
Governance coverage	Nearly 90%	Managers using algorithmic management who said their firm had at least one governance measure

Enterprise adoption is broad, while scaling remains limited

McKinsey's November 2025 global survey found that 88% of respondents reported regular AI use in at least one business function, up from 78% in its previous survey. The measure covers any AI use in one function. It does not mean that 88% of companies have rebuilt their operations around AI.

The same survey found that only about one-third of organizations had begun scaling AI programs across the enterprise. Company size mattered. Nearly half of respondents from businesses with more than $5 billion in revenue reported reaching the scaling phase, compared with 29% from businesses with less than $100 million in revenue.

AI agents were earlier in the adoption cycle. McKinsey reported that 23% were scaling an agentic AI system somewhere in the enterprise and another 39% were experimenting. No individual business function had more than 10% of respondents reporting scaled agent use.

The figures describe different stages:

Regular use means AI appears in at least one business function.
Scaling means the organization is expanding deployment beyond a limited experiment or pilot.
Enterprise value means the use has a measurable organization-wide result.

McKinsey found that 39% attributed some EBIT impact to AI, and most of that group attributed less than 5% of EBIT to it. Access is now widespread. Large financial effects are not.

Employee use is growing unevenly

Gallup's Q2 2026 U.S. workforce data offers a worker-level view. It found that 52% of employees used AI in their role at least a few times a year, 30% used it at least a few times a week, and 15% used it daily. These are self-reported usage rates, not employer software logs.

The most common uses among workers who used AI were writing or editing at 51%, search or research at 49%, and general assistance or problem-solving at 39%. Coding assistance and process automation were each reported by 16% of AI users. That mix helps explain why many productivity gains remain tied to tasks instead of whole workflows.

Gallup also found an adoption gap by work setting at the end of 2025. Among employees in remote-capable roles, 66% used AI at least a few times a year and 40% used it frequently. The corresponding rates in roles that were not remote capable were 32% and 17%.

Businesses adding AI to administrative or customer workflows still need accountable people to interpret context, handle exceptions, and check outputs. An AI virtual assistant can support that division of work when the task boundary and escalation route are explicit.

Productivity gains depend on the task and the worker

The strongest evidence comes from studies that measure output rather than ask people how productive they feel. A field study published by the National Bureau of Economic Research followed 5,179 customer support agents during the staggered rollout of a conversational AI assistant. Access to the tool increased issues resolved per hour by 14% on average. The increase was 34% for novice and lower-skilled workers, with little effect on experienced and highly skilled workers.

That result is narrower than a claim that AI makes every employee 14% more productive. It applies to one support operation, one assistance system, and a specific output measure. It also shows why workforce averages can hide large differences. The tool helped newer agents apply patterns associated with more experienced colleagues.

Gallup's April 2026 survey of 23,717 U.S. employees found a similar task-level pattern through self-reporting. Among employees in organizations that had adopted AI, 65% said it improved their productivity or efficiency. Yet only about one in 10 strongly agreed that it had transformed how work was done across the organization.

In Gallup's Q2 2026 findings, 77% of workers using AI for coding assistance and 77% using it for process automation reported a somewhat or extremely positive productivity effect. The comparable figures were 68% for writing and editing and 65% for search or research. These associations do not establish that the tools caused the gains. Workers may keep using applications that already suit their jobs.

Exposure usually means tasks will change

The ILO and Poland's NASK published a refined global index in May 2025 based on nearly 30,000 occupational tasks, expert review, model-based scoring, and harmonized labor data. It estimated that 25% of global employment was in occupations with some generative AI exposure. The share reached 34% in high-income countries.

Exposure is a technical estimate of whether generative AI can perform parts of an occupation. It is not an observed automation rate. The ILO concluded that transformation was more likely than replacement because most occupations still contain tasks that require human involvement.

The exposure is not distributed evenly. In high-income countries, the ILO and NASK estimated that occupations in the highest exposure category accounted for 9.6% of female employment and 3.5% of male employment. Clerical occupations remained the most exposed. The 2025 index also found growing exposure in highly digitized work such as software, finance, and media roles.

A March 2026 ILO and World Bank study extended the analysis to 135 countries covering about two-thirds of global employment. It found that developing economies generally had lower aggregate automation exposure than advanced economies but a more even distribution of augmentation potential across income groups. The authors warned that limited digital infrastructure may prevent some countries from capturing the productivity gains even when disruption reaches workers.

Reskilling demand is larger than current AI use

The World Economic Forum's Future of Jobs Report 2025 surveyed more than 1,000 employers representing over 14 million workers. It found that 86% expected AI and information-processing technologies to transform their business by 2030.

The planned workforce response centers on skills. By 2030, 77% of surveyed employers planned to reskill or upskill existing workers to operate more effectively alongside AI. Another 69% planned to recruit people who could design or improve AI tools, while 62% expected to hire people with skills for working with AI. Almost half, 47%, planned to move employees from AI-disrupted roles into other positions.

Employer plans also include job reductions. The same survey found that 41% expected to reduce workforce numbers where AI could replicate roles. That figure is an employer intention through 2030, not a forecast of the percentage of jobs that will disappear.

Skills are already a barrier. In the World Economic Forum's separate survey of more than 11,000 executives, 50% cited a lack of skills as a barrier to AI adoption and 43% cited a lack of management vision. Training needs to cover the work around the tool: deciding when to use it, testing the result, protecting data, and escalating uncertain cases.

Worker sentiment combines useful experience with job concern

Gallup's February 2026 survey found that 18% of U.S. employees believed their job was somewhat or very likely to be eliminated by AI or automation within five years. The share was 23% among employees at organizations that had adopted AI.

Those concerns coexist with reported gains. Sixty-five percent of employees in AI-adopting organizations said the technology improved their productivity or efficiency, while fewer than one in 10 reported a negative effect. Neither response proves a future employment outcome. Together they show that workers can find a tool useful and still worry about how employers will use it.

Staffing reports were also mixed. Employees at AI-adopting organizations were more likely than those at other organizations to report both expansion, 34% compared with 28%, and reduction, 23% compared with 16%. Both groups had a net tendency toward expansion. Among organizations with at least 10,000 employees, however, staff at AI adopters reported reductions slightly more often than expansion, 33% compared with 30%.

The survey is cross-sectional, so it cannot show that AI caused either staffing direction. Companies making major investments may also be changing headcount for other reasons.

Human oversight is part of job design

AI oversight is broader than checking a generated paragraph. It includes responsibility for hiring decisions, work allocation, performance monitoring, scheduling, and escalation when a system is wrong.

An OECD study published in February 2025 surveyed more than 6,000 firms in France, Germany, Italy, Japan, Spain, and the United States about algorithmic management. The category included software that fully or partly automated tasks traditionally performed by managers, and the OECD noted that not all of those tools used AI.

Among managers whose firms used the tools, nearly two-thirds reported at least one trustworthiness concern. Unclear accountability for a wrong decision was cited by 28%. Inability to follow the logic of a decision or recommendation and inadequate protection of worker health were each cited by 27%.

Most firms had put some governance around these systems. Nearly 90% of managers reported at least one measure, such as a risk assessment, software audit, implementation guideline, complaint channel, ethics function, or worker consultation. Almost two-thirds reported worker consultation. The survey measured whether a control existed, not whether it worked well.

A practical human-oversight design names five things:

The tasks AI may perform or assist with.
The person accountable for the final outcome.
The conditions that require human review before action.
The evidence retained for review, correction, and audit.
The route workers or customers can use to challenge a decision.

For teams that need flexible execution capacity, a virtual assistant service can keep a person responsible for quality while AI handles bounded research, drafting, routing, or data-entry steps.

What the 2026 data supports

The evidence supports three restrained conclusions. AI use has spread faster than enterprise scaling. Measured productivity gains can be meaningful, but they vary by task and experience. Most occupational exposure points to changed task bundles rather than immediate removal of whole jobs.

The data does not support one universal productivity percentage or a precise count of jobs that AI will replace. Employers can make better decisions by measuring AI at the workflow level: output per paid hour, error and rework rates, escalation quality, employee experience, and the share of cases that still require judgment.

Frequently asked questions

What percentage of companies use AI in 2026?

The answer depends on the definition and survey population. McKinsey's 2025 global survey found that 88% of respondents said their organizations regularly used AI in at least one business function. Only about one-third said their organizations had begun scaling AI programs across the enterprise. Gallup's Q2 2026 U.S. worker survey found that 47% of employees said their organization had integrated AI tools, while 20% were unsure.

How many workers use AI at work?

Gallup reported that 52% of U.S. employees used AI in their role at least a few times a year in Q2 2026. Thirty percent used it at least a few times a week, and 15% used it daily. These figures are U.S. self-reports and should not be treated as global workforce rates.

Does AI improve worker productivity?

It can for defined tasks. The NBER study of 5,179 customer support agents measured a 14% average increase in issues resolved per hour, with a 34% increase for novice and lower-skilled agents. Gallup found that 65% of U.S. employees at AI-adopting organizations reported better productivity or efficiency, but only about one in 10 strongly agreed that AI had transformed work across the organization.

Is AI replacing jobs or augmenting workers?

The ILO and NASK concluded in 2025 that job transformation was the more likely effect. Their index placed 25% of global employment in occupations with some generative AI exposure, but full automation remained limited because many tasks still required people. Exposure measures technical potential, not actual layoffs.

How much reskilling will AI require?

The World Economic Forum reported that 77% of surveyed employers planned to reskill or upskill existing workers to work more effectively with AI by 2030. Fifty percent of executives in a separate survey cited skills shortages as an adoption barrier.

What should remain under human oversight?

People should retain clear authority over decisions with legal, financial, safety, employment, or customer consequences. The review threshold should reflect the cost of an error. Organizations should also give workers and customers a route to challenge automated or AI-assisted decisions.

Sources and definitions

Book a free 15-min match call

Tell us what role you're filling. We'll match you with a pre-vetted virtual assistant - or tell you honestly if we're not the right fit.

Book a free call →