Here is a number worth paying attention to: 99%. That is the share of senior executives who say they have an approach to developing artificial intelligence (AI)-relevant skills in their workforce, according to a 2026 Economist Impact study supported by Kyocera Document Solutions. The study, which surveyed 639 decision-makers across five global financial hubs, also highlights a striking disconnect: while nearly all organizations are taking action, just 16% offer structured internal AI training. Almost everyone is doing something, almost no one is doing enough.

This gap defines the current state of enterprise AI readiness. Strategies are in place, and leadership alignment is growing. But when it comes to building the workforce capability required to operationalize AI at scale, most organizations are still relying on fragmented, low-impact approaches — mentorship schemes and self-directed learning platforms that rarely translate into real-world application.

The consequences are clear. While 88% of executives view AI skills as a source of competitive advantage, only 38% have a dedicated budget for AI-related skill development. Just 4% report that AI is embedded in core business processes delivering repeatable value.

The “Learn-and-Forget” Trap

The most common training approaches — mentorship and self-paced online learning — are easy to deploy but structurally limited. Without immediate opportunities to apply new skills, knowledge decays quickly. Employees complete modules, but behavior does not change.

Solving this is not about better content. It is about redesigning how learning happens. AI capability-building must be applied, embedded and accountable.

That starts with a structured skills audit. Rather than relying on self-reported confidence, organizations need a role-by-role mapping of the capabilities required to use AI effectively — prompt engineering, output validation, data governance and change management — alongside a clear assessment of current proficiency.

This is critical given the scale of the gap: fewer than 30% of organizations report high proficiency in core “AI-for-all” skills such as prompt engineering and output interpretation.

From Learning to Capability

Closing this gap requires a shift from awareness to application. Training programs must be designed around real work tasks. Employees should not just learn how AI tools function — they should use them to improve workflows, rewrite documents and enhance decision-making in their day-to-day roles.

Two structural changes make the difference. First, protected learning time must be built into workflows, not treated as optional. Second, organizations need low-risk environments for experimentation, where employees can build confidence using AI tools in context.

This is particularly important given that low confidence and fear of misuse remain significant barriers, especially in regulated industries.

The Middle Management Bottleneck

One of the most overlooked barriers to AI adoption sits with middle management. Nearly half of executives say managers have minimal responsibility for AI skills development, creating a disconnect between strategy and execution.

This is fundamentally an accountability issue. Managers prioritize what they are measured on. If AI capability-building is not embedded into performance expectations, it will remain peripheral and genuine skills-building will stall.

Effective programs therefore treat manager enablement as a core workstream — equipping leaders not just with technical knowledge, but with the ability to guide, support and create opportunities for applied learning within their teams.

Beyond Technical Skills

AI readiness is often framed as a technical challenge. In practice, it is equally a human one. Skills such as critical thinking, creativity and judgement are increasingly important as AI automates routine tasks, yet only a minority of organizations report strength in these areas.

Leading organizations are responding by integrating technical and human skill development, creating cross-functional learning environments and embedding reflection and evaluation into workflows.

Measuring What Matters

Finally, measurement must evolve. Most organizations still focus on short-term productivity gains or course completion metrics. These fail to capture true readiness.

A more meaningful approach tracks behavior: whether employees are using AI effectively, applying sound judgement and contributing to scaled deployment.

AI readiness is not a one-off initiative. It is an operational capability. Organizations that treat it as such — embedding learning into roles, workflows and management structures — will be the ones that move from experimentation to impact.