AI skill levels drive workforce divide, PwC finds

More than half of workers surveyed said they are falling behind their AI-savvy colleagues, with proficiency bolstering a sense of job security.

You have probably sat in a meeting where someone casually mentions a prompt they wrote over breakfast, a workflow they automated last week or a model they fine-tuned to answer customer questions. The rest of the room nods politely, but half the people present have no idea what any of those words mean in practice. That gap is no longer a minor inconvenience. It is a fault line running through the middle of the workforce, and PwC's latest survey suggests it is widening.

More than half of workers surveyed said they are falling behind their AI-savvy colleagues. The finding points to a bifurcation that mirrors earlier technology waves, spreadsheets in the 1980s, the internet in the 1990s, mobile in the 2000s, but with a sharper edge. AI proficiency now bolsters a sense of job security, which means the people who cannot use these tools are not just slower; they feel vulnerable.

What the divide looks like

The survey did not measure lines of code or model accuracy. It measured perception, and perception drives behaviour. Workers who feel they are falling behind do not volunteer for new projects. They avoid tools they do not understand. They stop asking questions because they assume everyone else already knows the answer. The result is a self-reinforcing loop: the confident get more practice, the hesitant fall further back and the distance between the two groups grows.

AI proficiency is not a single skill. It is a bundle of capabilities that range from writing a decent prompt to understanding when a model is hallucinating, from knowing which tasks to automate to recognising when human judgement still matters. The workers who feel secure are not necessarily the ones with computer science degrees. They are the ones who have been given permission to experiment, access to tools and enough psychological safety to fail a few times without penalty.

The workers who feel left behind often work in organisations that treat AI as an IT project rather than a capability-building exercise. They hear about pilots and proofs of concept, but they do not get hands-on time. They attend webinars that explain what AI is, but they never write a prompt themselves. They are told the technology will make their jobs easier, but no one shows them how.

Why this matters for L&D

Learning and development teams are used to capability gaps. They have closed them before with onboarding programmes, compliance training and leadership development. But AI is different in two ways. First, the technology is moving faster than curriculum design. A course built in January may be outdated by June. Second, the gap is not just technical. It is psychological. Workers who feel they are falling behind are not just missing a skill; they are losing confidence, and confidence is harder to teach than syntax.

The organisations that are closing the gap are doing three things well. They are embedding AI into existing workflows rather than treating it as a separate subject. They are giving people permission to use tools in low-stakes environments where mistakes do not matter. And they are pairing technical training with storytelling that shows what good looks like, real examples from real colleagues, not vendor case studies or executive keynotes.

PwC's finding that AI proficiency bolsters job security is a signal that workers are already making their own calculations. They know that the next round of restructuring will favour the people who can do more with less, and they know that "more with less" increasingly means "with AI". L&D leaders who treat this as a nice-to-have rather than a strategic priority are setting their organisations up for a two-tier workforce: the people who can use the tools and the people who cannot.

The practical stake

The divide PwC describes is not inevitable. It is a design choice. Organisations that invest in broad-based AI literacy, not just for data scientists or product managers, but for everyone, will build workforces that adapt faster, innovate more consistently and feel less anxious about the future. Organisations that do not will end up with a small group of power users and a large group of people who feel left behind, and that is a recipe for disengagement, attrition and missed opportunity.

The question for L&D is not whether to build AI capability. It is whether to build it fast enough to keep the divide from becoming permanent. The workers who feel they are falling behind are not wrong. They are responding to a real gap, and that gap will only close if someone decides to close it.

Sources:

• HR Dive, 30 September 2026