Most learning and development programmes spend their AI training budget teaching people to write better prompts for drafting emails and summarising documents. That is not AI readiness. It is a marginally more efficient version of what people were already doing.
A 2026 Deloitte study found that 84% of organisations have introduced some form of AI upskilling, but fewer than 23% report that employees can meaningfully integrate AI into complex decisions and workflows. The training is happening. The capability transfer is not.
The problem is rarely the content. It is the frame. And until L&D shifts the frame, organisations will keep investing in AI training that produces compliance rather than genuine capability change.
Shift 1: From Tool Adoption to Mindset Architecture
The question most L&D programmes are trying to answer is: how do we teach people to use AI tools? The more important question is: how do we help people fundamentally change how they decide which tasks need a human and which do not?
The difference matters because the first question leads to skill transfer tied to specific tools, features, and use cases. The second builds a durable capability that works across every tool, every context, and every change in the AI landscape over the next decade.
In practice, this means building what some practitioners call a delegation-first habit: before starting any knowledge task, the trained response is to ask whether an AI agent can do a version of this better, faster, or at lower cognitive cost, and only proceeding personally where the answer is clearly no.
This is not about removing human judgement from the process. It is about clearing the cognitive path to it. The managers and professionals who will create the most value over the next decade will not be those who work harder within existing workflows. They will be those who have learned to direct AI effectively, operating as the decision-maker and quality controller rather than the executor.
Shift 2: Reframe Career Development Around What Is Genuinely Human
Traditional career development frameworks ask: what skills does this person need to advance? The more pressing question, and the one L&D professionals need to be asking on behalf of their organisations, is: what does this person do that genuinely requires a human?
Oxford Economics' 2025 research identifies the capabilities with the lowest automation probability as those requiring complex judgement in ambiguous situations, deep human connection, creative synthesis across domains, and the ability to navigate social and organisational complexity. These are precisely the capabilities that traditional L&D treats as peripheral.
Emotional intelligence, leadership presence, stakeholder influence, the ability to hold a difficult conversation and come out with a stronger relationship than you started with: these are not soft skills. In an AI economy, they are the premium capability. Building career development frameworks around these human strengths is not a defensive posture. It is the highest-return investment an organisation can make in its people right now.
PwC's 2026 CEO Survey found that 76% of CEOs consider emotional intelligence more important than technical skill for leadership effectiveness in the current environment. The research direction is clear. Most L&D investment is still pointing the other way.
Shift 3: Prepare People to Manage AI Systems, Not Just Use Them
A significant proportion of professional roles over the next five years will shift from execution to oversight. McKinsey Global Institute's 2026 research projects that by 2030, 65% of current knowledge work activities will be automatable. But this does not mean those roles disappear. It means they transform.
The knowledge worker of 2030 will spend less time producing content and analysis, and more time setting direction, evaluating AI output quality, handling exceptions, and exercising the human judgement that consequential decisions require.
This is a different professional identity, and it requires deliberate development that most current L&D portfolios do not provide. The capabilities that matter here include designing effective AI workflows, evaluating outputs critically rather than deferring to them, knowing when to override AI recommendations, and communicating clearly when a decision was AI-assisted and how human judgement was applied.
Treating AI like a capable employee you manage, setting direction, reviewing outputs, correcting course, and holding it to quality standards, is a better mental model than treating it as a sophisticated search engine. The organisations building that model into their management development programmes now are ahead.
Shift 4: Invest in What Compounds as AI Advances
Efficiency has historically been the primary frame for performance development: how do we get more output from the same input? That frame is being disrupted. AI handles efficiency gains at a scale and speed that human development cannot match. The new frame for L&D investment is adaptability: the ability to learn, unlearn, and redirect faster than the environment changes.
The World Economic Forum's Future of Jobs 2025 report found that 39% of current skill sets will be transformed by 2030, a figure Deloitte's 2026 analysis suggests is already accelerating ahead of that projection. The average employee will need reskilling in five to six new capabilities over the next five years. The half-life of specific technical skills continues to shorten. The capabilities that compound in value as AI advances are deeply human: intellectual curiosity, grit, the ability to tolerate ambiguity, genuine empathy, the social skill to influence without authority, and moral courage under pressure.
These are learnable. They are also developable in ways that matter for organisational performance. Gallup's 2026 research found that teams led by managers with high empathy and interpersonal clarity show 28% lower turnover and 23% higher productivity than those led by managers focused primarily on task and output.
The L&D programmes that will deliver sustained value over the next decade are those that invest in the capabilities that become more valuable as AI becomes more capable, not those chasing the latest tool release.
Shift 5: Set Impossible Problems, Not Routine Tasks
When organisations train AI skills through routine tasks, email drafting, document summarisation, slide generation, they are implicitly communicating that AI is a productivity tool for the existing list of work. This limits the imagination of learners to work they already know.
The more transformative approach is to set genuinely difficult challenges that would previously have been out of reach for an individual or small team, and ask people to use AI to attack them. A strategist who discovers she can produce a competitive landscape analysis that would previously have required weeks of team effort, or a programme designer who finds he can prototype ten learning interventions in the time it used to take to build one, does not need to be persuaded that AI changes what is possible. They have experienced it directly.
MIT Sloan Management Review research found that workers who used AI for complex, novel tasks reported 47% higher skill acquisition than those using AI for routine work. Challenge-based AI learning produces not just skill but a qualitatively different kind of confidence, the kind that actually changes behaviour back in the organisation.
The design principle is straightforward: identify something your learners would previously have considered impossible within their role or domain, and give them the tools, time, and permission to attempt it. The experience of succeeding at the previously impossible is what shifts professional identity at the level that L&D exists to change.
The Implication for L&D Strategy
These five shifts are not incremental refinements to existing programmes. They represent a genuine restructuring of what L&D exists to do in an AI-augmented organisation.
The function that teaches tools is a service function. The function that develops the human capabilities that determine what AI-human collaboration actually achieves is a strategic one. The tools change every six months. The human capabilities they are built on persist for careers.
The organisations that get this right will be those where L&D moves deliberately from the first position to the second, before the window to do so becomes the gap between them and the competition.
Try This
Run an internal audit of your current L&D portfolio. For each programme or intervention, ask a single question: is this developing a capability that AI is likely to make more or less valuable over the next five years? Be honest. The answer will tell you exactly where the strategic investment is, and where you are spending development budget on diminishing returns.
Then pick one team and set them an impossible problem. Not an AI training exercise with a defined answer, a genuine challenge they have never attempted because it was previously out of reach. Give them AI access and give them time. What they discover about their own capability will be more instructive than any training content you could put in front of them.
References
Deloitte (2026) Global Human Capital Trends. London: Deloitte Insights.
Gallup (2026) State of the Global Workplace. Washington, DC: Gallup.
McKinsey Global Institute (2026) A new future of work: The race to deploy AI and raise skills. New York: McKinsey & Company.
Noy, S. and Zhang, W. (2023) 'Experimental evidence on the productivity effects of generative AI', Science, 381(6654), pp. 187-192.
Oxford Economics (2026) Future-proofing the workforce: Automation, skills and the human advantage. Oxford: Oxford Economics.
PwC (2026) 29th Annual Global CEO Survey. London: PwC.
World Economic Forum (2025) Future of Jobs Report 2025. Geneva: WEF.