A new PwC report attributes a widening workforce divide to differences in AI skill levels, according to HR Dive, putting capability gaps at the center of HR leaders' near-term agenda.

The Big Four firm's findings flag segmentation between workers who can apply artificial intelligence tools and those who cannot, a pattern with direct consequences for talent strategy, internal mobility and learning budgets.

What did PwC find?

The report, surfaced through HR Dive, identifies AI proficiency as the dividing factor between two groups of workers. PwC has tracked AI's workplace impact across multiple recent workforce surveys, making the firm a recurring reference point for reskilling benchmarks.

The publicly available coverage does not yet specify the survey's sample size, geographic scope, or whether the divide reflects self-reported skill or assessed competency. Those details will shape how seriously employers weight the findings when planning their own upskilling programs.

The findings arrive as enterprise adoption of generative AI tools continues to expand. Most large organizations have moved from pilot programs to wider deployment, putting pressure on HR functions to translate technical rollouts into workforce plans. PwC's framing of AI skill as a structural workforce variable — rather than a niche technical competency — pushes HR toward the same set of decisions other reskilling waves have triggered, only on a faster timeline.

Why does the divide matter for HR?

A workforce split between AI-fluent staff and those without those skills creates risk in several areas:

• Pay and promotion parity across roles
• Retention pressure on high-skill employees
• Productivity gaps between teams using AI and those that are not
• Unequal access to emerging AI-related positions

HR leaders typically respond by mapping current capabilities, expanding learning programs and reviewing promotion criteria. The effectiveness of those moves depends on how quickly training can be deployed at scale and whether managers actively direct employees toward AI-heavy projects.

The risk compounds when AI use becomes a de facto requirement for advancement. Workers shut out of those assignments early in the rollout may find themselves locked out of higher-paying roles later, widening internal pay gaps that HR functions are already being asked to monitor.

What should HR leaders do next?

The first step is measurement. Most organizations do not yet have a reliable inventory of which employees use AI tools, how often, and to what effect. Without that baseline, learning investments risk reaching the wrong audience or duplicating effort.

The second step is access. PwC's framing of the divide as a talent challenge, rather than a technology issue, points HR toward questions of equity: who gets trained, who gets deployed on AI projects, and who decides. Line managers typically hold that authority today, often without explicit guidance from HR.

The third is mobility. Internal job rotation remains one of the fastest ways to spread AI skills across an organization, particularly when formal training capacity is constrained. Pairing high-skill employees with newer users through structured mentoring can accelerate that transfer, though it adds to the workload of the most skilled workers, who may already be stretched thin.

What remains to be seen

HR Dive's coverage does not include direct quotes from PwC partners or named client employers. The absence of those voices leaves open how the firm expects HR functions to operationalize its findings, and whether the report benchmarks progress against earlier surveys. Sector breakdowns — whether the divide appears most sharply in financial services, technology or professional services — will also shape how individual employers read the data.

The report also does not address whether AI skill gaps correlate with tenure, role type or geography, all of which carry different policy implications for HR leaders planning targeted interventions.

What HR leaders can take from the coverage so far is a clear signal: the AI skill question is no longer theoretical, and the workforce is already sorting itself into groups based on who can answer it.