2026 edition · Updated 5 November 2026

AI in the Workplace Statistics 2026: An executive summary of the McKinsey, BCG, Deloitte, PwC, Microsoft and Google reports

Name: AI in the Workplace Statistics 2026
Creator: AISA
License: https://creativecommons.org/licenses/by/4.0/

64% of workers used AI at work in the past year (PwC), yet only 36% feel properly trained (BCG) and just 14% are AI front-runners (PwC).

This page brings the six major 2026 reports on AI at work into one place: McKinsey, BCG, Deloitte, PwC, Microsoft, and Google with Ipsos. We start with an executive summary: where the six reports agree, and where they disagree and why. Then we introduce each report: who it surveyed, how far its numbers can be trusted, the findings only it has, and its PDF. From there we answer the seven questions leaders ask most, from how many people use AI to who is accountable when it fails, with the evidence behind each answer. We close with what measured AI skill adds, what to expect in the next 12 months, and all 105 statistics, sourced and ready to download.

It is the most comprehensive overview of the state of AI adoption in organisations today, so .

The short version

Executive summary

Read side by side, the six reports agree on five things and disagree on four. Each point links to the evidence further down.

Where do the 2026 AI reports agree?

01Use is broad. Fluency is rare.Most people now use AI at work, but only about one in five uses it every day, and a small minority use it well.PwC · BCG · Gallup · Google and Ipsos · Microsoft
02Training is falling behind.Only about a third of workers feel properly trained, access to learning is shrinking, and managers rate training far more highly than the people receiving it.BCG · PwC · Google and Ipsos · Wharton and GBK · Deloitte
03Strategy beats tools.Companies with a clear AI strategy and redesigned workflows get far more from AI than companies that only roll out tools.BCG · McKinsey · Microsoft · Deloitte
04The value sits with a few.A fifth of companies capture three quarters of AI’s returns, and fluent users save more than twice the time of occasional ones.McKinsey · PwC · Google and Ipsos · Microsoft
05Agents are outrunning governance.Agents reached the workflows of nearly a third of AI users in a year, while few companies running several agents together have full governance in place.BCG · Deloitte · McKinsey · Wharton and GBK · Microsoft

Where do they disagree, and why?

01How many people use AI: 15% to 74%It depends on how often counts and who was asked: daily use is a different claim from use a few times a year.See the figures ↓
02Whether AI training is adequate: 14% to 78%Leaders rate the training they provide far more highly than employees rate the training they get.See the figures ↓
03How many people are advanced AI users: 1.1% to 19%Each report draws the line in a different place, and only one measures skill instead of asking about it.See the figures ↓
04How fast AI is changing jobs: 14% to 82%Most figures are forecasts. The one count of actual cuts is under half of what the same survey predicted.See the figures ↓

Every report in one place

Download the 2026 reports

Every report in this review, free from its publisher. Each link goes straight to the publisher’s own report, so there is no need to hunt across six websites.

Report	Sample	Download
McKinsey State of AI 2026McKinsey & Company · August 2026	1,719 respondents	Download PDF
BCG AI at Work 2026BCG · June 2026	11,749 workers	Download PDF
Deloitte State of AI in the Enterprise 2026Deloitte · October 2026	3,500 leaders	Download PDF
PwC Hopes and Fears 2026PwC · September 2026	49,364 workers	Read the report
Microsoft Work Trend Index 2026Microsoft · May 2026	20,000 AI users	Download PDF
Google and Ipsos Path to AI Fluency 2026Google and Ipsos · February 2026	4,464 employees + 1,511 managers	Download PDF

Why these six

Hundreds of AI surveys are published every year. A report earned its place here only by meeting all four tests.

Scope
Large samples across many countries, covering the people AI is changing: employees, managers and the leaders who fund it.
Depth
Full reports with a published method, sample and fieldwork dates. Press releases and teaser pages do not qualify.
Value of insights
Findings a leader can act on: skills, training, how work is redesigned and what AI returns, not adoption counts alone.
Freshness
Fieldwork from late 2025 onwards, so every figure describes AI at work as it is used now.

Report by report

The six reports at a glance

Every report here is a survey, and a survey measures what people say. For each one: who was asked, what it found that no other report did, and how much weight its numbers can bear.

McKinsey & Company · August 2026

McKinsey State of AI 2026: key findings and PDF

The 2026 edition of McKinsey’s annual survey, a series that runs back to 2017. Its headline: adoption keeps climbing while the profit impact has stalled.

Who: 1,719 respondents across regions, industries and company sizes, most of them executives
Where: 97 countries
When: Fieldwork May to June 2026
Run by: McKinsey (QuantumBlack, AI by McKinsey)

The finding to take away

32% of respondents say their organisation dropped a software purchase because agentic coding tools let it build its own (McKinsey, 2026). (p. 9)

One in five respondents say their organisation limits AI use because of operating costs, including tokens (McKinsey, 2026). (p. 15)
55% of C-level AI users say AI helps them make better decisions, against 40% of individual contributors (McKinsey, 2026). (p. 11)

Strongest for

Enterprise adoption, scaling and returns, tracked year after year.

Read with care

Respondents skew senior: 195 of the 1,490 with a role shown are individual contributors. Returns are respondents’ own estimates. McKinsey sells AI transformation.

Download PDF

BCG · June 2026

BCG AI at Work 2026: key findings and PDF

BCG’s fourth annual survey of the workforce, split evenly between frontline employees, managers and leaders. Its headline is its title: strategy matters more than tools.

Who: 11,749 frontline employees, managers and leaders in equal thirds, at companies with $100 million or more in revenue
Where: 14 markets
When: Fieldwork dates not published
Run by: BCG

The finding to take away

A clear AI strategy matters more than better tools: 80% of users with strategic clarity but limited tools report measurable impact, against 60% of those with strong tools but little clarity (BCG). (p. 22)

42% of respondents say their organisation uses AI to build new business models and products, up from 22% in 2025 (BCG). (p. 8)
41% of regular AI users report more mental strain since adopting AI, rising to 48% of leaders (BCG). (p. 19)

Strongest for

The view from every level of the workforce, with frontline, manager and leader answers side by side.

Read with care

Only companies with $100 million or more in revenue, and “frontline” means white-collar staff without reports. Fieldwork dates and method are not published.

Download PDF

Deloitte · October 2026

Deloitte State of AI in the Enterprise 2026: key findings and PDF

Deloitte’s survey of senior leaders at companies already running AI, in its second 2026 edition. Its headline: returns are spreading while the appetite for transformation shrinks.

Who: 3,500 senior leaders, from directors to board members, at companies already using AI every day
Where: 25 countries
When: Fieldwork May to June 2026
Run by: Deloitte

The finding to take away

Only 25% of companies are using AI to deeply transform the business, down from 34% in January (Deloitte, October 2026). (p. 11)

Only 44% of companies track the per-unit cost of AI inference, per token or API call (Deloitte, October 2026). (p. 15)
86% of companies expect their AI investment to rise in the coming fiscal year (Deloitte, October 2026). (p. 14)

Strongest for

What leaders invest in, measure and govern, tracked across two 2026 editions.

Read with care

Leaders only, so figures about workers are leaders’ estimates. Every company surveyed already uses AI. Some questions changed wording between editions.

Earlier edition: January 2026 (3,235 leaders, fieldwork August to September 2025)

Only 20% of companies felt highly prepared on talent for AI, the lowest of the readiness areas Deloitte tracks (January 2026). (p. 29)

Download the January 2026 PDF

Download the October 2026 PDF

PwC · September 2026

PwC Hopes and Fears 2026: key findings and PDF

PwC’s annual survey of workers, the largest in this review, published as a web report. Its headline: the workforce is splitting into a small group pulling ahead and a large core left behind.

Who: 49,364 workers across 29 sectors
Where: 48 countries and regions
When: Fieldwork May to June 2026
Run by: PwC

The finding to take away

51% of PwC’s AI front-runners use generative AI daily, against 11% of the core workforce.

29% of AI front-runners say they are very or extremely likely to change employer in the next year (PwC).
Twice as many workers think their bargaining power has fallen over the past three years as think it has risen (PwC).

Strongest for

The broadest view of workers themselves, weighted to each working population.

Read with care

No per-question bases or question wording are published, and some headline figures come from other PwC studies. PwC sells AI and workforce consulting.

Read the report

Microsoft · May 2026

Microsoft Work Trend Index 2026: key findings and PDF

Microsoft’s annual report on how AI is changing work, the only one in this review that pairs its survey with usage data.

Who: 20,000 knowledge workers who use AI at work, 2,000 in each market
Where: 10 markets, from the US and UK to India, Japan and Brazil
When: Fieldwork February to April 2026
Run by: Edelman Data x Intelligence, for Microsoft

The finding to take away

What people get from AI depends more on their organisation than on themselves. In Microsoft’s analysis, organisational factors such as AI culture explain 67% of the impact AI users report, and their own mindset explains 32%. (p. 16)

49% of more than 100,000 Microsoft 365 Copilot chats support cognitive work such as analysing, reasoning and deciding (Microsoft). (p. 7)
43% of Microsoft’s most advanced AI users deliberately do some work without AI to keep their skills sharp, against 30% of other users. (p. 27)

Strongest for

How AI users work and how their organisation shapes the result, with telemetry alongside the survey.

Read with care

Only people who already use AI were surveyed, so shares describe users, not the workforce. Microsoft sells the tools it studies. Its impact analysis shows associations, not causes.

Download PDF

Google and Ipsos · February 2026

Google and Ipsos Path to AI Fluency 2026: key findings and PDF

Google and Ipsos’s study of the US workforce, drawn from a probability panel and weighted to the population, so it hears from people who don’t use AI as well as those who do. Its headline: fluency pays, and employees their employer has trained are far more likely to have it.

Who: 4,464 US employees and 1,511 US hiring managers, users and non-users of AI alike
Where: United States
When: Fieldwork December 2025
Run by: Ipsos, on its probability-based KnowledgePanel, for Google

The finding to take away

AI Fluent US employees say AI saves them a median 8 hours a week, against 3 hours for AI Explorers, who use it ad hoc (Google and Ipsos). (p. 4)

53% of US employees who don’t use AI say it isn’t relevant to their job, their most common reason (Google and Ipsos). (p. 10)
64% of US educators use AI at work, against 38% of other employees (Google and Ipsos). (p. 7)

Strongest for

The most representative picture of US employees, users and non-users alike.

Read with care

US only. “AI Fluent” is built from people’s own answers, and the group is small (215 people). Google sells AI tools and training.

Download PDF

Which report to trust for what

We reviewed every report for how its sample was drawn, how its figures were weighted and how far its analysis supports its conclusions, and cross-checked its findings on AI skills against AISA’s own measured assessment data. The largest samples are PwC’s 49,364 workers and BCG’s 11,749. The strongest sample design is Google and Ipsos’s probability panel, weighted to the US workforce, and PwC also weights to each country’s working population. McKinsey and Deloitte survey mostly senior people about their organisations: the best read on investment, returns and governance, and the weakest on what employees can do. Microsoft surveys only people who already use AI, and BCG does not publish its fieldwork dates or method. None tests skill: every report counts AI fluency in people’s own words.

The seven questions

How many employees use AI at work in 2026?

Most workers now use AI, but only about one in five uses it every day. How many people “use AI” depends almost entirely on how often counts.

Share of workers using AI, by how often counts

At all

Workers, in the past 12 months · PwC64%
US employees, a few times a year or more · Gallup52%
US employees who say they use AI at work · Google and Ipsos40%

Weekly or more

Frontline white-collar employees at large firms · BCG74%
US employees · Gallup30%

Every day

Workers using generative AI · PwC22%
US employees · Gallup15%

Why the numbers differ

Three things move the figure. How often counts: using AI a few times a year is a different claim from using it daily. Who was asked: BCG surveys white-collar staff at large firms, Gallup the whole US workforce, and managers report far heavier use (62% of US managers and above use AI daily, according to Wharton and GBK). And whether a report counts people or organisations: McKinsey’s 89% counts organisations using AI somewhere in the business.

Also worth knowing

52% of office-based US employees use AI at work, against 24% of employees who work outside an office (Google and Ipsos).
Frontline AI use trails in the US, France and Italy, at 62% each against a 74% average across 14 markets (BCG).
Writing is the most common weekly use of AI at work, named by 71% of US employees who use AI, ahead of learning (66%) and brainstorming (64%).
Of workers with sanctioned AI access, fewer than 60% used it in their daily workflow, largely unchanged on the year before (Deloitte, January 2026).
47% of US employees said their organisation had integrated AI tools by Q2 2026, up from 41% a quarter earlier (Gallup).

Sources: PwC Hopes and Fears 2026 · Gallup · Google and Ipsos Path to AI Fluency 2026, p. 6, 7, 8 · BCG AI at Work 2026, p. 13, 14 · Wharton and GBK Collective, p. 24 · McKinsey State of AI 2026, p. 5 · Deloitte State of AI in the Enterprise (January 2026), p. 8

How many employees get AI training at work?

Far fewer than need it, and the gap is widening. Only 36% of workers feel properly trained, and access to learning fell from 59% to 51% in a year.

Is AI training adequate? It depends who you ask

As leaders see it

US managers who say employees get adequate AI training · Wharton and GBK78%
Companies educating the workforce to raise AI fluency · Deloitte53%

As employees see it

US employees who say their AI training is adequate · Wharton and GBK56%
Workers with access to learning resources · PwC51%
Workers who feel properly trained in AI · BCG36%
US employees offered AI training in the past year · Google and Ipsos14%

Why the numbers differ

Leaders answer about the training they provide; employees answer about the training they received. The further down the organisation a survey goes, the worse training looks. Wharton and GBK asked managers and employees the same question and got answers 22 points apart.

Also worth knowing

US employees trained in AI by their employer were 2.5 times as likely to use AI and almost 3.5 times as likely to be AI Fluent (Google and Ipsos).
65% of employees at companies redesigning work around AI have been through a major reskilling initiative, against 47% where AI is only deployed (BCG).
Fewer than 40% of PwC’s “engine room”, the core 56% of the workforce, have access to learning resources, against nearly 80% of AI front-runners.
65% of US employees have some interest in formal AI training in the next 12 months, and 75% say it should be hands-on (Google and Ipsos).
46% of US employees allowed to use AI turn to colleagues or managers for guidance on it (Google and Ipsos).

Sources: Wharton and GBK Collective, p. 23 · Deloitte State of AI in the Enterprise (January 2026), p. 14 · PwC Hopes and Fears 2026 · BCG AI at Work 2026, p. 10, 11 · Google and Ipsos Path to AI Fluency 2026, p. 4, 23, 27, 28

How big is the AI skills gap in 2026?

Wider than any survey shows. By their own account, 5% to 19% of people are advanced AI users. Measured in conversation, 1.1% reach Expert.

Advanced AI users, by how each report draws the line

Self-reported

AI users in the Frontier zone of a readiness index · Microsoft19%
AI users reporting frontier practices · Microsoft16%
Workers with strong AI gains and in-demand skills · PwC14%
Organisations crediting AI with 5% or more of operating profit · McKinsey6%
US employees whose own account makes them AI Fluent · Google and Ipsos5%

Measured in conversation

Professionals at Expert level · AISA AI Fluency Index1.1%

Why the numbers differ

Each report draws the line in a different place and counts from a different base: Google and Ipsos and PwC count the whole workforce, Microsoft only people who already use AI, McKinsey organisations by the returns they report. Every one of them rests on what respondents say about themselves. Measured in a real conversation, the share is smallest.

Also worth knowing

Senior leaders named insufficient worker skills as the biggest barrier to integrating AI into existing workflows (Deloitte, January 2026).
Only 18% of US employees who use AI are extremely or very confident they get the most out of it (Google and Ipsos).
52% of workers have only a limited understanding of what AI agents are, including 61% of frontline staff and 39% of leaders (BCG).
AI users rank quality control of AI output (50%) and critical thinking (46%) as the human skills that matter most as AI does more (Microsoft, 2026).
Understanding how AI works and which tool fits the job is the weakest area for 44% of professionals, and in 12 of the 14 largest occupations (AISA AI Fluency Index, November 2026).

Sources: Microsoft Work Trend Index 2026, p. 8, 12, 23 · PwC Hopes and Fears 2026 · McKinsey State of AI 2026, p. 17 · Google and Ipsos Path to AI Fluency 2026, p. 8, 12 · AISA AI Fluency Index · Deloitte State of AI in the Enterprise (January 2026), p. 14 · BCG AI at Work 2026, p. 25

The surveys measure what people say about their AI skills.

AISA measures what they can do. A free, 20 to 40 minute conversation scores your AI fluency against the same rubric as thousands of professionals, with a full report and a certificate.

See my AISA score

Is AI paying off for companies in 2026?

People feel the gain; most companies do not yet see it in their numbers. The returns that do exist are concentrated in a small group.

80%

of respondents say AI has improved their own productivity (McKinsey)

37%

report any gain in operating profit from AI, unchanged from 2025 (McKinsey)

20%

of companies capture 74% of AI-driven returns (PwC)

Why the gain stalls before the bottom line

People save time, but organisations rarely tell them what to do with it. 66% of frontline employees who use AI regularly get little or no guidance on the time it saves, and 58% don’t reinvest it in more strategic work (BCG). The productivity stays with the individual.

Also worth knowing

28% of companies now report revenue gains from AI, up from 20% in Deloitte’s January edition (Deloitte, October 2026, 3,500 leaders).
74% of companies report efficiency and productivity gains from AI, up from 66% in January (Deloitte, October 2026).
28% of US managers and above report strongly positive ROI from their department’s AI investments so far (Wharton and GBK, 2026).
42% of frontline employees who use AI regularly save a full day or more a week, against 60% of leaders (BCG).
58% of AI users say they now produce work they couldn’t a year ago, rising to 80% of the most advanced users (Microsoft, 2026).

Sources: McKinsey State of AI 2026, p. 12, 14 · PwC Hopes and Fears 2026 · BCG AI at Work 2026, p. 3, 6, 7 · Deloitte State of AI in the Enterprise 2026, p. 7, 17 · Wharton and GBK Collective, p. 9 · Microsoft Work Trend Index 2026, p. 25

How is AI changing jobs and the way work is organised?

Expectations are running far ahead of reality. Leaders and workers expect deep change, but few companies have redesigned jobs in depth, and actual cuts trail the forecasts.

What people expect, against what has happened

Expected

Companies expecting 10% or more of jobs fully automated within three years · Deloitte82%
AI users who think agents could do half their job within three years · BCG61%
Respondents expecting AI to cut headcount next year · McKinsey39%

Has happened

Companies that have redesigned jobs around AI extensively · Deloitte15%
Respondents whose workforce shrank because of AI last year · McKinsey14%

Why the forecasts outrun reality

Most of these figures are forecasts. The one backward-looking count of cuts, McKinsey’s 14%, is under half of the 32% the same survey predicted a year earlier. And more workers fear the economy than AI for their job: 57% against 44% (PwC).

Also worth knowing

Nearly three in four AI high performers have fundamentally redesigned workflows around AI, up from 55%, against one in four other respondents (McKinsey).
70% of US hiring managers treat AI skills as a requirement or a preference when hiring (Google and Ipsos, 1,511 managers).
Workers with AI skills command a 62% wage premium, up from 57% in 2025, according to PwC’s AI Jobs Barometer, cited in Hopes and Fears 2026.
Only 26% of AI users say their leadership is clearly and consistently aligned on AI (Microsoft, 2026).
Only 13% of AI users say they are rewarded for reinventing work with AI even when the results don’t follow (Microsoft, 2026).

Sources: Deloitte State of AI in the Enterprise (January 2026), p. 12 · BCG AI at Work 2026, p. 26 · McKinsey State of AI 2026, p. 3, 18, 25 · Deloitte State of AI in the Enterprise 2026, p. 18 · PwC Hopes and Fears 2026 · Google and Ipsos Path to AI Fluency 2026, p. 20 · Microsoft Work Trend Index 2026, p. 13, 29

How far have companies got with AI agents?

AI agents moved into everyday work in 2026, faster than the rules for running them. Autonomy is spreading while governance lags.

30%

of regular AI users have agents in their workflows, up from 13% in 2025 (BCG)

40%

of companies let AI act without human approval, at least moderately (Deloitte)

of companies running several agents together call them fully governed (Deloitte)

Scale is concentrated in big companies

At organisations with $1 billion or more in revenue, 40% are scaling agents, up from 27% in a year, while smaller firms are flat at 22% (McKinsey).

Also worth knowing

67% of US departments say they use AI agents, but only 17% let agents act on their own, with a person reviewing afterwards or not at all (Wharton and GBK).
Active agents in Microsoft 365 grew 15 times over in a year, and 18 times over in large enterprises (Microsoft telemetry, 2026).
50% of workers say their company has no clear guidance on managing teams of people and AI together (BCG).
Only 21% of companies had a mature governance model for autonomous AI agents (Deloitte, January 2026).
About two in ten respondents say their organisation has reached the scaling phase with AI agents (McKinsey, 2026).

Sources: BCG AI at Work 2026, p. 26, 27 · Deloitte State of AI in the Enterprise 2026, p. 8, 20 · McKinsey State of AI 2026, p. 6, 7 · Wharton and GBK Collective, p. 9 · Microsoft Work Trend Index 2026, p. 17 · Deloitte State of AI in the Enterprise (January 2026), p. 20

Do employees trust AI at work, and who is accountable when it fails?

People check AI’s work carefully. What is slipping is trust in the leaders deploying it, and nearly a quarter of companies cannot say who answers when AI gets it wrong.

86%

of AI users treat AI output as a starting point, not a final answer (Microsoft)

52%

say AI has increased the time they spend reviewing and correcting its output (BCG)

28%

of frontline staff see leaders’ words and actions on AI line up (BCG)

Who answers when AI fails?

23% of companies say accountability is unclear when an AI system causes a bad outcome, and 10% put it on the employee who prompted the AI (Deloitte).

Also worth knowing

Trust in management fell 7 points in a year across PwC’s sample of nearly 50,000 workers.
Only 25% of US employees say their organisation has communicated a clear plan for integrating AI (Gallup, May 2026).
47% of midlevel managers and individual contributors report at least one negative effect from AI, against 31% of executives and senior managers (McKinsey).
65% of AI users fear falling behind if they don’t adapt quickly with AI (Microsoft, 2026).
63% of US employees who use AI say its benefits outweigh its concerns; 66% of those who don’t use it say the opposite (Google and Ipsos).

Sources: Microsoft Work Trend Index 2026, p. 27, 29 · BCG AI at Work 2026, p. 3, 17 · Deloitte State of AI in the Enterprise 2026, p. 8, 21 · PwC Hopes and Fears 2026 · Gallup · McKinsey State of AI 2026, p. 11 · Google and Ipsos Path to AI Fluency 2026, p. 9

AISA research

What measured AI skill shows that the surveys cannot

Every report in this review asks people about AI. AISA measures it. In a 20 to 40 minute conversation, each professional shows how they actually work with AI, and every score is tied to evidence in their own words. The November 2026 edition of the AISA AI Fluency Index, built on 3,400+ assessments, tells a starker story than any survey here.

Read the AISA AI Fluency Index, November 2026

Measured, not self-reported

46/100Average AI fluency score, measured in 3,400+ assessments 70%Of professionals score below Proficient +18Points by which three in four people overestimate their own AI fluency

See where you stand: a free conversation, a full report and a certificate.

See my AISA score

Where AISA agrees with the reports, and where it doesn’t

AgreeTraining is the gap that matters most.AISA’s data shows exactly where it bites. Understanding how AI works and which tool fits the job is the weakest of the five dimensions, at 41 out of 100, and the weakest area in 12 of the 14 largest occupations. It is also what most separates the top 10% from everyone else.
AgreeChecking AI’s work is the core skill.Critical thinking is one of the five dimensions every AISA score is built on, measured from how people actually handle AI’s answers rather than what they say about them. It averages 44 out of 100.
AgreeLeaders and employees see AI differently.The reports find leaders rating AI training far above the people receiving it. Measured skill adds a twist: managers and team leads score below senior individual contributors, while directors predict their own score within 4 points.
AgreeThe organisation sets the ceiling.People assessed through their employer score higher on all five dimensions and misjudge themselves by 5 points, against 19 for those who come on their own. What organisations build shows up in measured skill.
DisagreeSelf-reported fluency overstates the skill.Every survey here counts advanced users by what people say. Measured, three in four people overestimate their score, those who expect 80 or more average 50, and the least skilled overestimate by 41 points.
DisagreeAdoption is the wrong scoreboard.Daily use does show in the score: people who use AI several times a day average 60, against 33 for everyone else. But inside organisations with ten or more people assessed, the gap between the strongest and weakest scorer runs from 35 to 69 points. A usage count cannot see that spread.

More in the November 2026 Index

AISA publishes its research every quarter in the AISA AI Fluency Index, and the November 2026 edition is out now. It goes far deeper than this review can. Three findings from what people told us before they were assessed:

Average AI fluency score out of 100, and the share at Proficient or above

“Which AI tool do you use most?”

Claude55% Proficient+58.9

Everyone else22% Proficient+42.2

“How would your colleagues describe your AI usage?”

“Ahead of most” or “the one everyone asks for help”39% Proficient+52.6

Everyone else9% Proficient+34.9

“Do you pay for an AI tool?”

Yes, out of my own pocket38% Proficient+52.0

Everyone else14% Proficient+39.7

The edition also breaks scores down by occupation and seniority, maps the ten AI personas, and shows how scores change when people retake the assessment.

Read the full November 2026 Index

Conclusion

What happens to AI at work in the next 12 months?

AISA’s view of the year ahead, built on where these reports point and on what measured AI skill shows.

01Measurement shifts from usage to skill.With most workers already using AI, counting licences and logins stops telling leaders anything. Expect organisations to measure what people can do with AI, person by person, and to set targets on skill rather than adoption.
02Training becomes role-specific and hands-on.Employees ask for hands-on training from a trusted provider, and AI-leading firms are already 1.7 times as likely to offer ongoing, role-based learning (PwC). Generic awareness courses give way to practice built around each role’s real work.
03Agent governance moves to the top table.As agents take on decisions, named owners, approval rules and audit trails for agents become standard practice, led by the largest companies, which are scaling agents fastest.
04The frontier pulls further ahead.The most advanced users save more time, produce new kinds of work and get more of the learning on offer. Without deliberate training for everyone else, that lead compounds.
05AI skills become a hiring and pay signal, and proof replaces claims.With hiring managers already screening for AI skills and a wage premium attached to them, employers will ask for evidence: measured scores and verifiable credentials.
06Work redesign finally starts.The few companies capturing real returns are the ones that redesigned their workflows. The rest will follow, because that is where the returns are.

Frequently asked questions

What percentage of employees use AI at work in 2026?

64% of workers used AI at work in the past 12 months, according to PwC’s survey of 49,364 workers in 48 countries and regions, and 22% use generative AI daily. In the US, Gallup puts it at 52% of employees, with 15% using AI daily.

How many employees have been trained to use AI?

Only 36% of workers feel properly trained in AI, according to BCG’s survey of 11,749 workers, unchanged from 2025. Only 14% of US employees were offered AI training by their employer in the past year (Google and Ipsos), and access to learning fell from 59% to 51% (PwC).

Is AI paying off for companies?

For a minority. 37% of McKinsey’s respondents report any gain in operating profit from AI, unchanged from 2025, and 6% are high performers. Deloitte finds 28% of companies now report revenue gains from AI, up from 20% in January.

What is the biggest barrier to using AI at work?

Skills and strategy. Senior leaders in Deloitte’s January 2026 report named insufficient worker skills as the biggest barrier to integrating AI, and BCG finds a clear AI strategy matters more than tools: 80% of users with strategic clarity report measurable impact, against 60% with strong tools but little clarity.

Which 2026 AI at work report is the most reliable?

For scale, PwC’s 49,364 workers. For a representative picture, Google and Ipsos’s probability panel weighted to the US workforce. For returns and governance, McKinsey’s and Deloitte’s surveys of senior leaders. All of them survey what people report; none tests AI skill directly.

Where can I download the 2026 McKinsey, BCG and Deloitte AI reports?

Every report in this review is linked at the top of this page, under “Download the 2026 reports”, straight to each publisher’s own report.

How can I measure AI fluency in my team?

Surveys measure what people say about their AI skills. AISA measures what they can do: a 20 to 40 minute conversation scores each person’s AI fluency across five dimensions, and team results show exactly where the gaps are. The assessment is free for individuals.

The full list

All 105 statistics, with sources

Every figure in this review, grouped by question, with the report and page it comes from. Free to quote with its original source, or download the lot as a spreadsheet.

Show all 105 statistics

Statistic	Source
How many employees use AI at work in 2026?
64% of workers used AI at work in the past 12 months, up 10 points, and 22% now use generative AI daily, up from 14% (PwC, 49,364 workers).	PwC Hopes and Fears 2026
74% of frontline employees are now regular AI users, up 23 points in a year, against 88% of managers and 93% of leaders (BCG, 11,749 workers).	BCG AI at Work 2026, p. 13
52% of US employees used AI in their role in Q2 2026: 30% a few times a week or more, and 15% daily (Gallup, 22,573 employees).	Gallup
89% of McKinsey’s respondents say their organisation regularly uses AI in at least one business function, and 44% say AI is scaling across the enterprise.	McKinsey State of AI 2026, p. 5
40% of US employees said they use AI at work in Google and Ipsos’s survey of 4,464 employees, fielded in December 2025.	Google and Ipsos Path to AI Fluency 2026, p. 6
62% of US managers and above used AI daily in July 2026, up 16 points in a year, and 91% used it more than weekly (Wharton and GBK).	Wharton and GBK Collective, p. 24
47% of US employees said their organisation had integrated AI tools by Q2 2026, up from 41% a quarter earlier (Gallup).	Gallup
Of workers with sanctioned AI access, fewer than 60% used it in their daily workflow, largely unchanged on the year before (Deloitte, January 2026).	Deloitte State of AI in the Enterprise (January 2026), p. 8
52% of office-based US employees use AI at work, against 24% of employees who work outside an office (Google and Ipsos).	Google and Ipsos Path to AI Fluency 2026, p. 7
Frontline AI use trails in the US, France and Italy, at 62% each against a 74% average across 14 markets (BCG).	BCG AI at Work 2026, p. 14
Writing is the most common weekly use of AI at work, named by 71% of US employees who use AI, ahead of learning (66%) and brainstorming (64%).	Google and Ipsos Path to AI Fluency 2026, p. 8
53% of US employees who don’t use AI say it isn’t relevant to their job, their most common reason (Google and Ipsos).	Google and Ipsos Path to AI Fluency 2026, p. 10
64% of US educators use AI at work, against 38% of other employees (Google and Ipsos).	Google and Ipsos Path to AI Fluency 2026, p. 7
Professionals who use AI several times a day score 60 out of 100 on average, against 33 for everyone else (AISA AI Fluency Index, November 2026).	AISA AI Fluency Index
Professionals whose main AI tool is Claude score 58.9 out of 100 on average, against 42.2 for everyone else (AISA AI Fluency Index, November 2026).	AISA AI Fluency Index
People who pay for an AI tool out of their own pocket score 52.0 out of 100 on average, against 39.7 for everyone else (AISA AI Fluency Index, November 2026).	AISA AI Fluency Index
How many employees get AI training at work?
Only 36% of workers feel properly trained in AI, unchanged from 2025, while 88% expect to need major upskilling within five years (BCG).	BCG AI at Work 2026, p. 11
Only 14% of US employees had been offered AI training by their employer in the past 12 months (Google and Ipsos, 4,464 employees).	Google and Ipsos Path to AI Fluency 2026, p. 4
51% of workers say they have access to learning and development resources, down from 59% a year earlier (PwC).	PwC Hopes and Fears 2026
Fewer than 40% of PwC’s “engine room”, the core 56% of the workforce, have access to learning resources, against nearly 80% of AI front-runners.	PwC Hopes and Fears 2026
78% of US managers say employees receive adequate AI training; only 56% of entry- and mid-level employees agree (Wharton and GBK, 2026).	Wharton and GBK Collective, p. 23
53% of companies were educating their wider workforce to raise AI fluency, the most common talent response to AI (Deloitte, January 2026).	Deloitte State of AI in the Enterprise (January 2026), p. 14
65% of US employees have some interest in formal AI training in the next 12 months, and 75% say it should be hands-on (Google and Ipsos).	Google and Ipsos Path to AI Fluency 2026, p. 27
US employees trained in AI by their employer were 2.5 times as likely to use AI and almost 3.5 times as likely to be AI Fluent (Google and Ipsos).	Google and Ipsos Path to AI Fluency 2026, p. 28
65% of employees at companies redesigning work around AI have been through a major reskilling initiative, against 47% where AI is only deployed (BCG).	BCG AI at Work 2026, p. 10
AI-leading firms are 1.7 times as likely to provide ongoing, role-based AI learning, according to PwC’s AI Performance Study, cited in Hopes and Fears 2026.	PwC Hopes and Fears 2026
About half of AI users at every level, 51% to 55%, say using AI has helped them develop new skills (McKinsey, 2026).	McKinsey State of AI 2026, p. 11
Only 37% of US employees say their organisation gives guidance on how to use AI (Google and Ipsos).	Google and Ipsos Path to AI Fluency 2026, p. 4
46% of US employees allowed to use AI turn to colleagues or managers for guidance on it (Google and Ipsos).	Google and Ipsos Path to AI Fluency 2026, p. 23
People assessed through their employer misjudge their AI fluency by 5 points, against 19 for those who come on their own, and score higher on all five dimensions (AISA AI Fluency Index, November 2026).	AISA AI Fluency Index
How big is the AI skills gap in 2026?
Only 5% of US employees qualify as AI Fluent, against 40% who use AI at work (Google and Ipsos, 2026).	Google and Ipsos Path to AI Fluency 2026, p. 12
Only 14% of workers are AI front-runners, combining strong gains from AI with skills in high demand (PwC, 49,364 workers).	PwC Hopes and Fears 2026
16% of the 20,000 AI users in Microsoft’s 2026 survey qualify as Frontier Professionals, its most advanced group: 3,233 people.	Microsoft Work Trend Index 2026, p. 23
51% of PwC’s AI front-runners use generative AI daily, against 11% of the core workforce.	PwC Hopes and Fears 2026
Only 18% of US employees who use AI are extremely or very confident they get the most out of it (Google and Ipsos).	Google and Ipsos Path to AI Fluency 2026, p. 8
52% of workers have only a limited understanding of what AI agents are, including 61% of frontline staff and 39% of leaders (BCG).	BCG AI at Work 2026, p. 25
Senior leaders named insufficient worker skills as the biggest barrier to integrating AI into existing workflows (Deloitte, January 2026).	Deloitte State of AI in the Enterprise (January 2026), p. 14
Only 20% of companies felt highly prepared on talent for AI, the lowest of the readiness areas Deloitte tracks (January 2026).	Deloitte State of AI in the Enterprise (January 2026), p. 29
AI users rank quality control of AI output (50%) and critical thinking (46%) as the human skills that matter most as AI does more (Microsoft, 2026).	Microsoft Work Trend Index 2026, p. 8
20% of midlevel managers who use AI say it hurts their ability to think critically, against 9% of C-level users (McKinsey, 2026).	McKinsey State of AI 2026, p. 11
70% of professionals measured by AISA score below Proficient, and only 7.3% reach Advanced or Expert (AISA AI Fluency Index, November 2026).	AISA AI Fluency Index
Three in four professionals overestimate their own AI fluency, by 18 points on average (AISA AI Fluency Index, November 2026).	AISA AI Fluency Index
43% of Microsoft’s most advanced AI users deliberately do some work without AI to keep their skills sharp, against 30% of other users.	Microsoft Work Trend Index 2026, p. 27
19% of AI users have both strong AI skills of their own and an employer ready to use them, the Frontier zone of Microsoft’s readiness index.	Microsoft Work Trend Index 2026, p. 12
Only 1.1% of professionals measured by AISA reach Expert-level AI fluency (AISA AI Fluency Index, November 2026).	AISA AI Fluency Index
Understanding how AI works and which tool fits the job is the weakest area for 44% of professionals, and in 12 of the 14 largest occupations (AISA AI Fluency Index, November 2026).	AISA AI Fluency Index
People who say their colleagues see them as ahead of most, or the one everyone asks for help, score 52.6 out of 100 on average, against 34.9 for everyone else (AISA AI Fluency Index, November 2026).	AISA AI Fluency Index
Is AI paying off for companies in 2026?
37% of McKinsey’s respondents say AI has added to their organisation’s operating profit (EBIT), essentially unchanged from 2025.	McKinsey State of AI 2026, p. 12
Only 6% of respondents work at AI high performers, organisations that credit AI with 5% or more of operating profit and significant value, unchanged from 2025 (McKinsey).	McKinsey State of AI 2026, p. 17
20% of companies capture 74% of AI-driven returns, according to PwC’s AI Performance Study, cited in Hopes and Fears 2026.	PwC Hopes and Fears 2026
28% of companies now report revenue gains from AI, up from 20% in Deloitte’s January edition (Deloitte, October 2026, 3,500 leaders).	Deloitte State of AI in the Enterprise 2026, p. 7
74% of companies report efficiency and productivity gains from AI, up from 66% in January (Deloitte, October 2026).	Deloitte State of AI in the Enterprise 2026, p. 17
80% of respondents say AI has improved their individual productivity (McKinsey, 2026).	McKinsey State of AI 2026, p. 14
42% of frontline employees who use AI regularly save a full day or more a week, against 60% of leaders (BCG).	BCG AI at Work 2026, p. 3, 6
66% of frontline employees who use AI regularly get limited or no guidance on what to do with the time it saves (BCG).	BCG AI at Work 2026, p. 3
A clear AI strategy matters more than better tools: 80% of users with strategic clarity but limited tools report measurable impact, against 60% of those with strong tools but little clarity (BCG).	BCG AI at Work 2026, p. 22
28% of US managers and above report strongly positive ROI from their department’s AI investments so far (Wharton and GBK, 2026).	Wharton and GBK Collective, p. 9
AI Fluent US employees say AI saves them a median 8 hours a week, against 3 hours for AI Explorers, who use it ad hoc (Google and Ipsos).	Google and Ipsos Path to AI Fluency 2026, p. 4
58% of AI users say they now produce work they couldn’t a year ago, rising to 80% of the most advanced users (Microsoft, 2026).	Microsoft Work Trend Index 2026, p. 25
Only 25% of companies are using AI to deeply transform the business, down from 34% in January (Deloitte, October 2026).	Deloitte State of AI in the Enterprise 2026, p. 11
One in five respondents say their organisation limits AI use because of operating costs, including tokens (McKinsey, 2026).	McKinsey State of AI 2026, p. 15
55% of C-level AI users say AI helps them make better decisions, against 40% of individual contributors (McKinsey, 2026).	McKinsey State of AI 2026, p. 11
42% of respondents say their organisation uses AI to build new business models and products, up from 22% in 2025 (BCG).	BCG AI at Work 2026, p. 8
Only 44% of companies track the per-unit cost of AI inference, per token or API call (Deloitte, October 2026).	Deloitte State of AI in the Enterprise 2026, p. 15
86% of companies expect their AI investment to rise in the coming fiscal year (Deloitte, October 2026).	Deloitte State of AI in the Enterprise 2026, p. 14
49% of more than 100,000 Microsoft 365 Copilot chats support cognitive work such as analysing, reasoning and deciding (Microsoft).	Microsoft Work Trend Index 2026, p. 7
58% of frontline employees who use AI regularly don’t reinvest the time it saves in more strategic work, against 36% of leaders (BCG).	BCG AI at Work 2026, p. 7
How is AI changing jobs and the way work is organised?
39% of respondents expect AI to reduce their organisation’s headcount next year, up from 32% a year earlier (McKinsey, 2026).	McKinsey State of AI 2026, p. 3
Only 14% of respondents at AI-using organisations say AI reduced their workforce in the past year, under half the 32% who expected cuts (McKinsey).	McKinsey State of AI 2026, p. 25
Only 15% of companies have redesigned jobs around AI extensively or completely, essentially unchanged since January (Deloitte, October 2026).	Deloitte State of AI in the Enterprise 2026, p. 18
Nearly three in four AI high performers have fundamentally redesigned workflows around AI, up from 55%, against one in four other respondents (McKinsey).	McKinsey State of AI 2026, p. 18
82% of companies expect at least 10% of their jobs to be fully automated within three years; 36% expect it within one (Deloitte, January 2026).	Deloitte State of AI in the Enterprise (January 2026), p. 12
61% of regular AI users think AI agents could do at least half their job within three years, including 52% of frontline employees (BCG).	BCG AI at Work 2026, p. 26
More workers see economic volatility than AI as a risk to their job security: 57% against 44% (PwC).	PwC Hopes and Fears 2026
Workers with AI skills command a 62% wage premium, up from 57% in 2025, according to PwC’s AI Jobs Barometer, cited in Hopes and Fears 2026.	PwC Hopes and Fears 2026
70% of US hiring managers treat AI skills as a requirement or a preference when hiring (Google and Ipsos, 1,511 managers).	Google and Ipsos Path to AI Fluency 2026, p. 20
What people get from AI depends more on their organisation than on themselves. In Microsoft’s analysis, organisational factors such as AI culture explain 67% of the impact AI users report, and their own mindset explains 32%.	Microsoft Work Trend Index 2026, p. 16
Only 26% of AI users say their leadership is clearly and consistently aligned on AI (Microsoft, 2026).	Microsoft Work Trend Index 2026, p. 13
Only 13% of AI users say they are rewarded for reinventing work with AI even when the results don’t follow (Microsoft, 2026).	Microsoft Work Trend Index 2026, p. 29
18% of AI Fluent US employees say using AI has led to higher pay, against 4% of AI Explorers (Google and Ipsos).	Google and Ipsos Path to AI Fluency 2026, p. 17
29% of AI front-runners say they are very or extremely likely to change employer in the next year (PwC).	PwC Hopes and Fears 2026
Twice as many workers think their bargaining power has fallen over the past three years as think it has risen (PwC).	PwC Hopes and Fears 2026
Managers and team leads score below senior individual contributors, 48 against 51 out of 100, while directors predict their own score within 4 points (AISA AI Fluency Index, November 2026).	AISA AI Fluency Index
How far have companies got with AI agents?
30% of regular AI users say agents are integrated into their workflows, up from 13% in 2025, while 50% use them only in pilots (BCG).	BCG AI at Work 2026, p. 26
About two in ten respondents say their organisation has reached the scaling phase with AI agents (McKinsey, 2026).	McKinsey State of AI 2026, p. 7
At organisations with $1 billion or more in revenue, 40% are scaling agents, up from 27% in a year; smaller firms are flat at 22% (McKinsey).	McKinsey State of AI 2026, p. 6
40% of companies let AI execute decisions or actions without human approval, at least to a moderate degree (Deloitte, October 2026).	Deloitte State of AI in the Enterprise 2026, p. 8
Where several AI agents work together on one process, half of companies have no formal governance or only basic oversight, and 9% call it fully governed (Deloitte).	Deloitte State of AI in the Enterprise 2026, p. 20
67% of US departments say they use AI agents, but only 17% let agents act on their own, with a person reviewing afterwards or not at all (Wharton and GBK).	Wharton and GBK Collective, p. 9
Active agents in Microsoft 365 grew 15 times over in a year, and 18 times over in large enterprises (Microsoft telemetry, 2026).	Microsoft Work Trend Index 2026, p. 17
50% of workers say their company has no clear guidance on managing teams of people and AI together (BCG).	BCG AI at Work 2026, p. 27
Only 21% of companies had a mature governance model for autonomous AI agents (Deloitte, January 2026).	Deloitte State of AI in the Enterprise (January 2026), p. 20
32% of respondents say their organisation dropped a software purchase because agentic coding tools let it build its own (McKinsey, 2026).	McKinsey State of AI 2026, p. 9
Do employees trust AI at work, and who is accountable when it fails?
86% of AI users treat AI output as a starting point, not a final answer (Microsoft, 2026).	Microsoft Work Trend Index 2026, p. 27
52% of workers say AI has increased the time they spend reviewing and correcting its output (BCG).	BCG AI at Work 2026, p. 17
47% of midlevel managers and individual contributors report at least one negative effect from AI, against 31% of executives and senior managers (McKinsey).	McKinsey State of AI 2026, p. 11
Trust in management fell 7 points in a year across PwC’s sample of nearly 50,000 workers.	PwC Hopes and Fears 2026
Only 28% of frontline employees see strong alignment between what leaders say about AI and what the organisation does (BCG).	BCG AI at Work 2026, p. 3
Only 25% of US employees say their organisation has communicated a clear plan for integrating AI (Gallup, May 2026).	Gallup
23% of companies say accountability is unclear when an AI system causes a bad outcome (Deloitte, October 2026).	Deloitte State of AI in the Enterprise 2026, p. 8
10% of companies hold the employee who prompted the AI responsible when it causes a bad outcome (Deloitte, October 2026).	Deloitte State of AI in the Enterprise 2026, p. 21
65% of AI users fear falling behind if they don’t adapt quickly with AI (Microsoft, 2026).	Microsoft Work Trend Index 2026, p. 29
63% of US employees who use AI say its benefits outweigh its concerns; 66% of those who don’t use it say the opposite (Google and Ipsos).	Google and Ipsos Path to AI Fluency 2026, p. 9
Leaders estimate that only 13% of non-technical workers are highly enthusiastic about AI (Deloitte, January 2026).	Deloitte State of AI in the Enterprise (January 2026), p. 14
41% of regular AI users report more mental strain since adopting AI, rising to 48% of leaders (BCG).	BCG AI at Work 2026, p. 19

Download all 105 statistics (CSV)

How to cite this page

AISA (2026). AI in the Workplace Statistics 2026: An Executive Summary of the McKinsey, BCG, Deloitte, PwC, Microsoft and Google Reports. Updated 5 November 2026. https://aisa.to/ai-in-the-workplace-statistics

Also cited: Gallup, Organizational AI Adoption Jumps Six Points (Q2 2026 workforce study, 22,573 US employees) (July 2026) · Gallup, Global Indicator: Artificial Intelligence (May 2026) · Wharton and GBK Collective, The Adaptive Advantage: 2026 AI Adoption Report (1,007 US managers and above, 409 entry- and mid-level employees) (October 2026)

The Science Behind AISA

In 2026, Anthropic published the AI Fluency Index — the largest empirical study of AI fluency to date, analysing nearly 10,000 conversations. AISA covers 93% of the behaviours Anthropic identified as markers of AI fluency and goes even deeper with 4 additional criteria. The U.S. Department of Labor's AI Literacy Framework (TEN 07-25) defines what every worker needs to know about AI — AISA covers 100% of its 25 sub-competencies.Read our analysis: Anthropic's AI Fluency Study & AISA · DOL AI Literacy Framework & AISA

AISA's framework is developed by a team with deep roots in tech, behavioural science, and AI product leadership — the rubric is informed by backgrounds spanning the Metropolitan Police, Harvard, Crowdbotics (Silicon Valley), and the European School of Economics.

Official Member Of

UKAI Official Member — UK AI Trade Association

Featured on

Product Hunt

There's an AI for That