On 29 September 2026, the consultancy PwC published the latest edition of its annual Global Workforce Hopes and Fears Survey — and the headline finding should worry every executive steering an AI strategy. Artificial intelligence is spreading through the workplace faster than ever: 64% of workers say they used AI at work in the past 12 months, up 10 percentage points in a year, and daily use of generative AI has climbed from 14% to 22%. But the machinery that turns AI exposure into AI capability — training — is moving in the opposite direction. Only 51% of workers say they have access to the learning and development resources they need, down from 59% a year ago.
Put those two trend lines on one chart and you get a workforce dividing into two: a minority racing ahead with AI, and a majority at risk of being left behind. The survey — 49,364 workers across 48 countries and regions and 29 sectors, surveyed in May and June 2026 and weighted to be broadly representative of working populations in each market — finds companies increasingly exposed to a double risk: losing the AI-savvy employees they can least afford to lose, while the bulk of their workforce stagnates.
The four tribes of the AI workplace
To map the divide, PwC segmented respondents along two dimensions: how scarce their skills are, and how far along the AI learning curve they sit. The result is four groups that barely resemble one another.
The largest, the "engine room", accounts for 56% of workers — employees whose skills are considered less scarce and who are not far along the AI learning curve. Only two in five of these workers say they have access to the learning and development resources they need. The "AI insurgents", at 18%, have less scarce skills but are using AI to deliver more. The "indispensables", at 11%, possess scarce skills that employers value highly but have made less progress with AI. And the "front-runners", at 14%, combine scarce skills with strong AI capabilities.
Note the asymmetry: the smallest group holds the scarcest skills and the strongest AI fluency, while the largest group holds neither. That imbalance is the structural problem every other finding keeps circling back to.
Half the workforce is moving at a different speed
The daily experience of the two ends of the divide barely overlaps. Workers who use AI daily report markedly better outcomes across the board: 68% feel confident about their job security, compared with 57% of infrequent users. They are 12 percentage points more likely to ask for a promotion, 15 points more likely to trust senior management, and 21 points more confident in their ability to learn new skills. AI use, the survey suggests, does not make workers feel replaceable — it makes them feel capable. Daily users report that AI makes them feel energised and empowered rather than tired and powerless, by margins of around five to one.
"Workers are under growing pressure from rising workloads, rapid change and external uncertainty," said Pete Brown, PwC's global workforce leader. "But those using AI every day are more confident, more ambitious and more positive about their careers and prospects."
The flip side is the engine room. These workers — the majority carrying out most organisations' day-to-day work — are less likely to be rewarded for using AI, learning new skills or challenging established ways of working. Their skills are not treated as scarce, their AI adoption lags, and their access to training is thinnest. PwC's warning is explicit: the global workforce is "starting to move at different speeds", and more than half of workers are not yet benefiting from AI and skills in the same way.
The retention threat at the top
The cruel irony for employers is that the workers they most need to keep are the ones most likely to leave. Front-runners are the most rewarded group in the data — 1.5 times more likely than the global average to be rewarded for challenging existing ways of working, and 1.2 times for learning new skills — and 33 percentage points more likely to ask for a promotion. Yet almost a third (29%) say they are very or extremely likely to change employer in the next year.
That is a retention problem hiding inside a training problem. Companies are underinvesting in AI learning precisely among the workers who most need it, while the employees who have already cracked AI on their own are shopping for employers who value them more. The result is a slow sorting of talent: the capable drift toward the companies that invest in them, and the underinvested majority drifts further behind.
"The challenge for leaders is to keep developing those people — but not at the expense of the rest of the workforce," Brown said. His closing warning was blunt: "There is a real risk that the global workforce is starting to move at different speeds. More than half of workers are not yet benefiting from AI and skills in the same way, while those with scarce skills and strong AI capabilities are becoming more confident and more mobile. Leaders need to think hard about where they invest in skills, how they give people the opportunity to adapt, and how they hold on to the capabilities they most need."
The middle matters too: insurgents and indispensables
The two middle segments show how the divide might narrow — or harden. The AI insurgents, at 18% of the workforce, are the hopeful story: workers whose skills are not considered scarce but who are actively using AI to deliver more. With training and encouragement, insurgents are the natural pipeline into front-runner status — ambitious, self-directed adopters who need only institutional backing.
The indispensables, at 11%, are the quiet risk. They hold scarce skills employers value highly, but have made little progress with AI. In a labour market where AI fluency is increasingly the multiplier on top of domain expertise, rare skills without AI augmentation may not stay rare — or as valuable — for long. Employers who assume scarcity alone will keep these workers competitive are making the same mistake as those who assume the engine room will catch up on its own.
The pressure cooker underneath
The AI divide is forming against a backdrop of mounting pressure. Only a third of workers report having money left at the end of the month, down 8 percentage points year on year. Almost a third say they have significantly less bargaining power than three years ago — more than double the number who disagreed — and among fully remote workers that figure is 10 points higher still.
Strikingly, workers fear the economy more than the technology. When asked about the biggest risks to their job security, 57% cited economic volatility, compared with 44% who worried about AI taking on more of their tasks. Gen Z workers are more anxious across all of these measures.
The pace of change is also accelerating. Of those who experienced workplace change, 58% said there had been more of it in the past year than previously, and 27% said fatigue or burnout was limiting their productivity. Two-fifths of workers (40%) said they had needed to apply new skills in their jobs to a large or very large extent in the past year — rising to 51% of senior executives and 46% of managers, versus 29% of non-managers and 38% of entry-level employees. Change, in other words, is being felt most intensely at leadership levels, while the engine room gets the least support to keep up.
What this means for employers
The practical takeaway is threefold. First, audit where learning investment actually goes: if training budgets flow disproportionately to the already-capable, the divide compounds with every funding cycle. Second, give the engine room deliberate on-ramps — 59% of workers expect their AI use to increase over the next 12 months, so the appetite exists; the access does not. Structured, job-relevant AI training for the majority is the single highest-leverage move in this data. Third, treat the front-runners as a flight risk, not a solved problem: their skills are scarce, their confidence is rising, and nearly a third are open to leaving.
The deeper lesson of the survey is that AI does not create a skills divide on its own — employers do, through where they invest and who they reward. Companies have spent the past two years racing to adopt AI tools. The PwC data suggests the next race is the one that matters: building a workforce where the whole organisation can use them. The firms that keep developing their best AI talent without leaving the rest of the workforce behind will be the ones writing the next chapter of this story — everyone else will be reading it.
Sources
