AI Skills Gap: 68% of Workers Use AI but Lack Training Time

Artificial intelligence tools have quickly shifted from experimental software into permanent fixtures across enterprise offices. According to the 2026 State of Skills Intelligence Report published on September 23 by skills assessment platform Workera, 80% of enterprise organizations believe they are on track for an AI-enabled future in 2026, climbing from 67% in 2025. However, while organizational optimism and tooling availability have accelerated rapidly over the past twelve months, enterprise investment in employee training time has struggled to keep pace.

The research surveyed 1,000 full-time salaried professionals working at organizations with 5,000 or more employees across the United States. Conducted in July 2026 using market research platform Pollfish and benchmarked against comparable findings gathered in March 2025, the study outlines a persistent structural divide: practical access to artificial intelligence software is widespread, but corporate structures rarely grant staff the operational bandwidth to master these systems.

The Adoption Surge and the Workplace Time Crunch

The daily use of artificial intelligence in enterprise environments has expanded significantly over the past year. Workera discovered that 67.8% of surveyed employees now use AI tools beyond ChatGPT at least a few days each week, representing a substantial climb from the 39.9% reported in 2025. Concurrently, formal training offerings have increased: 58.3% of workers stated they were offered AI-specific training opportunities in the last twelve months, more than double the 24.8% measured in the previous benchmark.

Despite the wider distribution of training programs, daily scheduling realities prevent meaningful upskilling. A majority of respondents, 56.4%, reported that no time is specifically allocated during work hours for skill development. Furthermore, 84.3% of employees spend five hours or less per week on training, while 42.5% identified a distinct lack of relevant learning materials as a primary obstacle to building their capabilities.

I would advise any leader right now to think about giving their people time to upskill. Don’t think that they will just figure it out without you actually carving out time. With time also comes the psychological safety; give them the psychological safety to experiment.

This friction has created an environment of unguided experimentation. Confronted with limited internal materials and tight operational deadlines, substantial segments of the corporate workforce have turned to external resources to build proficiency independently.

• Shadow learning: 46.5% of surveyed employees use skill development tools not supplied by their employers, with 74.6% of that subgroup turning to mainstream platforms such as ChatGPT, Claude, and Gemini.
• Evaluation inaccuracies: 42.5% of respondents reported being mis-rated, passed over, or held back professionally due to managerial misjudgments of their actual technical skills.
• Management coaching deficits: Only 26.6% of enterprise employees receive regular coaching from a direct manager, while 21.5% receive no coaching at all.

Human Judgment, Capability Limits, and Skills Measurement

While employees are actively using generative systems, their confidence in human workplace capabilities remains resilient. The survey found that 76.6% of workers believe they can perform their jobs better than artificial intelligence. Only 9.5% of respondents stated that an automated AI agent could effectively execute more than half of their current daily responsibilities.

A similar dynamic governs performance evaluation. Approximately 69.1% of employees believe a human manager evaluates their professional skills more fairly than an automated AI tool, compared to just 14.7% who prefer algorithmic assessment. Furthermore, only 7% of workers believe AI is already better than humans at evaluating workplace skills, while 38.5% maintain that automated systems will never surpass human evaluators in this domain.

To bridge the gap between static reviews and actual technical execution, Workera is testing Ambient, an AI-native skills assessment system with over 10,000 signups on its waitlist designed to measure capabilities directly within workflows. The study showed that 41.1% of employees would participate in continuous skills measurement if they owned their data and controlled sharing permissions, while 31.2% remained undecided.

Operational Takeaways for Enterprise Leaders

For enterprise organizations, these findings indicate that purchasing software licenses and publishing optional training catalogues is insufficient to secure workforce readiness. Because 46.5% of staff engage in unsanctioned shadow learning and 42.5% face career penalties from flawed skill assessments, businesses risk accumulating low-quality AI outputs and misallocating internal talent.

Closing this operational gap requires leadership teams to formally define internal AI competency standards, protect dedicated weekly learning time on employee calendars, and deploy objective, transparent measurement frameworks. By pairing continuous skill baselines with regular managerial coaching and explicit incentives, organizations can accelerate learning velocity while ensuring workforce capabilities align directly with business goals.