Leadership development is a $370 billion global industry. The estimate comes from research firm Josh Bersin (2019), and by most accounts the number has grown since. Yet meta-analyses of leadership development programs consistently find modest, short-lived effects — Lacerenza et al. (2017) reviewed 335 studies and found an average effect size of d = 0.63, with most gains fading within months absent reinforcement. The industry is large; the results are mixed. This is what the evidence actually says about leadership skills for the future — and which investments are most likely to pay off as AI reshapes the professional environment.

What McKinsey Says About Future Leadership Skills

McKinsey's research on future-ready leadership (2023) identifies four core capabilities most in demand across high-performing organizations:

Sense-making — the ability to synthesize ambiguous, contradictory, or incomplete information into actionable direction. This is not the same as having more information or better analytical tools. It is the human capacity to find patterns, assign meaning, and create coherent narratives from noise — a capability that becomes more, not less, important when information volume increases.

Relating — building genuine trust across diverse teams and stakeholder groups. Not networking in the superficial sense, but the deep interpersonal competence that allows people from different backgrounds, functions, and perspectives to coordinate effectively around shared goals.

Visioning — the ability to articulate credible and compelling futures that people want to move toward. Particularly important in environments of high uncertainty, where teams need a directional signal without false certainty about what the path will look like.

Inventing — designing new approaches under uncertainty, where standard playbooks don't apply. This is the leadership equivalent of creative problem-solving applied to organizational challenges.

These four capabilities share a common feature: they are not well-supported by hierarchical authority models. They require genuine human engagement, contextual judgment, and relational trust. They are the leadership skills for the future precisely because they are the dimensions of leadership AI cannot replicate.

The Leadership Skills That Are Becoming Obsolete

Being direct about obsolescence is useful for prioritization. Several traditional leadership competencies are declining in value:

Expertise as the primary authority source. When a leader's authority derived primarily from knowing more than subordinates, AI changes the equation. AI systems have broader factual knowledge than any human in most domains. Leadership authority grounded solely in expertise is structurally eroding.

Information control. In hierarchical organizations, leaders often held advantage by controlling the flow of information. Transparency, information symmetry, and networked communication have already degraded this model; AI accelerates the trend.

Hierarchical coordination. The managerial layer whose primary value was coordinating information flow between levels is being compressed by flatter organizational structures and AI coordination tools. Middle management roles defined by information relay are among the most exposed.

Annual performance management. The traditional annual review cycle was already poorly suited to dynamic organizations. In environments where work, teams, and priorities shift quarterly, annual feedback loops are too slow to be useful guidance systems.

Leadership Skills for the Future: The Evidence-Based List

Four capabilities have strong evidence bases for high-return development investment:

Directing AI effectively. This is an entirely new leadership competency with no direct historical precedent. It requires understanding AI capabilities well enough to know when to deploy them, how to evaluate their outputs, and where they fail. Leaders who cannot do this will increasingly make decisions based on AI outputs they cannot assess — a structural governance failure. This is one of the most important leadership skills for the future precisely because leadership programs have not yet caught up to it.

Creating psychological safety. Amy Edmondson's research at Harvard Business School (1999, 2018) remains among the most replicated findings in organizational behavior: teams where members feel safe to raise concerns, admit errors, and challenge assumptions significantly outperform teams where they don't. The leader's primary structural contribution to team performance is creating the conditions for psychological safety. This is measurable and developable.

Meaning-making under uncertainty. The ability to help teams function effectively through ambiguity without manufacturing false certainty. This means communicating honestly about what is unknown, maintaining direction without premature closure, and helping people tolerate the discomfort of not knowing while still moving forward. Research on organizational resilience (Weick, 1995) identifies meaning-making as the central leadership function under crisis conditions.

Cross-domain synthesis. Connecting insights across disciplines and domains — the capacity to see structural similarities between apparently different problems and apply solutions from one context to another. AI amplifies this capability: a leader who can direct AI across multiple domains compounds the value of the tool far beyond what a narrow specialist can.

How to Develop These Skills

Psychological safety is developed through specific behavioral patterns, not intention. Edmondson's research identifies three leader behaviors that reliably increase it: demonstrating genuine curiosity rather than judgment when someone raises a concern; explicitly modeling uncertainty ("I don't know the answer to that"); and creating clear structures where raising problems is an expected part of work, not an exception to it.

Sense-making develops through structured scenario exercises — deliberately working through ambiguous situations with multiple valid interpretations and practicing narrative construction. Many senior leadership programs include some version of this; most junior development programs do not.

Directing AI effectively requires deliberate engagement with AI tools — not casual use, but serious deployment in real work contexts combined with explicit reflection on where the outputs failed. The leaders who develop this skill fastest are those who treat AI as an object of study, not just a tool of convenience.

These are the leadership skills for the future that training budgets have not yet caught up to.