AI Upskilling Gap Widens As 67.8% Of Workers Use Tools Weekly

AI training has moved from an adoption problem to an operating gap inside large companies, ZDNET reported, citing a Workera survey that found 67.8% of employees now use AI tools beyond ChatGPT at least a couple of days each week while 56.4% say their employers give them no work time to build the skills those tools require.

That split is the useful signal in the research.

Companies increasingly believe they are closer to an AI-enabled future, but the people expected to deliver that future are still fitting training around the edges of their jobs.

Workera’s 2026 State of Skills Intelligence Report found that 80% of companies consider themselves more likely to be on track for AI readiness in 2026, up from 67% last year.

The survey covered 1,000 salaried U.S. professionals at organizations with at least 5,000 employees.

Pollfish conducted the 2026 fieldwork in July, and Workera compared those results with a similar March 2025 survey.

The year-over-year shift shows AI use spreading quickly: regular use of non-ChatGPT AI tools rose from 39.9% last year to 67.8% this year, nearly a 30-point gain.

The training structure has not kept pace.

Nearly 43% of employees pointed to a shortage of relevant learning materials, and 84.3% said they spend five hours or less each week on skill development.

Workera founder and CEO Kian Katanforoosh argued that managers should make time and psychological safety explicit, rather than assuming employees will teach themselves while normal workloads continue.

The result is a shadow-skills market inside the enterprise.

Just under half of workers, 46.5%, have used tools their employer did not provide to develop skills.

Among that group, 74.6% used mainstream systems such as ChatGPT, Claude or Gemini.

That pattern gives companies faster experimentation, but it also weakens visibility into which skills are improving, which practices are safe and which teams are merely accumulating unsupervised habits.

Training availability has improved in one respect.

Nearly six in 10 employees said they had been offered AI-specific training in the past 12 months, a contrast with last year’s finding that most had not received such opportunities.

The remaining problem is execution: courses and access do not guarantee practice time, measurement or incentives tied to role-specific standards.

Workera’s answer is to define what AI-ready means, measure employees against that standard and connect learning to incentives.

That approach also explains the company’s interest in continuous assessment.

Its Ambient tool, now in pilot with more than 10,000 waitlist signups, is designed to measure skills in the flow of work rather than as a separate classroom exercise.

Employee appetite for that kind of measurement is cautious rather than closed.

About 41.1% said they would opt into continuous skills measurement if they owned the data and controlled what could be shared, while 31.2% were undecided.

The privacy condition matters because skills analytics can help target training, but it can also feel like surveillance if workers have no control over the record.

The research also cuts against the idea that employees see AI as an immediate substitute for their own roles.

Almost 76.6% said they can do their jobs better than AI, and only 9.5% believed an AI agent could do more than half their work effectively today.

On assessment itself, just 7% said AI can already judge skills better than humans, while 38.5% said it will never surpass human evaluators.

For employers, the numbers point to a practical next step rather than a philosophical one.

AI tools are already embedded in weekly work for a majority of respondents; the open question is whether companies can turn that usage into measured capability.

Without dedicated time, relevant materials and trusted assessment, the adoption curve may leave workers improvising the skills strategy on their own: 67.8% are using broader AI tools regularly, while 56.4% still report no work-hour time to upskill.