Are Your Leadership Expectations Keeping Pace?

Across organizations, AI is rapidly changing the division of work between people and machines.

Much of the early attention centered on productivity. The implications now reach further, changing where analysis, synthesis, recommendations, and execution occur—and how work gets done.

The work-design challenge is already visible. Deloitte's 2026 Global Human Capital Trends found that 66% of leaders see the intentional design of human-AI interaction as important to organizational success, but only 6% say their organizations are making great progress. Deloitte argues that effective human-machine interaction will not happen by default; it has to be intentionally designed.1

Leadership is part of that design challenge. Leaders remain accountable for setting direction, delivering results, and building the people and teams needed to succeed—even as AI becomes part of how that work gets done.

That creates a new test for leadership expectations—and for the models that make those expectations explicit. Most leadership models were built before AI began taking on meaningful portions of the analysis, synthesis, recommendations, and execution traditionally performed by people. As the division of work changes, do those models still describe what leaders need to do particularly well?

Our research suggests that many leadership fundamentals will endure. The harder question is whether the model is specific enough about how those fundamentals apply in AI-enabled work, how related expectations should differ by level, and what leaders in this organization will need to do particularly well for the future business to succeed.

More capacity does not automatically create capability

AI can increase capacity by expanding the speed, scope, and volume of work individuals and organizations can accomplish. But greater human-AI capacity does not automatically produce better work.

A 2024 meta-analysis of 106 experimental studies found that AI generally helped humans perform better than they did alone, but human-AI teams still did not outperform whichever—human or AI—was the stronger performer on its own. The researchers describe this as human augmentation without consistent human-AI synergy, with the gap particularly evident in decision tasks.2

These were not leadership studies, but the finding complicates a common assumption: adding AI to human effort does not necessarily produce the strongest result. More capacity at our fingertips does not automatically translate into better outcomes. And capacity is not the same as capability.

At RBL, we define capability with an outside-in perspective. It is not simply what an organization can do effectively, but what it is good at doing in ways that create value for customers and other stakeholders. Dave Ulrich recently described the relationship as “AI for Capacity—Organization for Capability.” AI can create substantial new capacity. The larger challenge is turning that capacity into capability and stakeholder value.

That requires judgment.

AI shifts where judgment matters

As AI takes on more analytic groundwork—searching information, spotting patterns, summarizing options, or performing well-defined analysis—the risk is handing off too much of the thinking along with it.

More of the leader's work may shift from producing analysis to exercising judgment about it. Some judgment comes before AI is ever used: framing the problem and deciding what criteria matter. Some occurs while leaders work with AI, as they test evidence, assumptions, and alternatives. And some comes afterward, when a leader must still make the tradeoff, choose a course, and remain accountable for what follows.

Our 2024 Leadership Code research identified “navigate complexity” as one of nine timely leadership practices that have become increasingly important. High-performing leaders think clearly, critically, and systemically; see the bigger picture; and derive implications that lead to better direction and decisions.

Organizations do not need humans in every decision loop. They do need greater precision about where human judgment improves the work and what quality of judgment the work requires.

For leaders, the implication is not “trust AI less.” It is to scrutinize both sides of the interaction: the quality of the machine’s output and the assumptions, confidence, and mental models leaders themselves bring to interpreting it.

Timeless leadership, timely application

None of this necessarily requires an entirely new definition of leadership.

Across more than 2.3 million Leadership Code ratings and more than two decades of research, the five core domains of effective leadership have remained remarkably durable. At the same time, the behaviors leaders need to emphasize—and how they apply those behaviors—change as the context changes. We describe this as the difference between timeless principles and timely practices.

AI provides a current test of that distinction. Strategists may spend less time personally finding trends and running analysis and more time framing the question, testing assumptions, and deciding what AI-generated findings mean. Execution increasingly includes deciding how work should move among people and technology, where guardrails belong, and who remains accountable. Leaders developing talent may need to become more deliberate about which human capabilities and experiences must be retained as AI takes on work that once helped people build expertise. Personal proficiency may place greater weight on the self-awareness required to challenge both AI output and one’s own assumptions.

The underlying leadership domains have not disappeared, but their expression in the work is changing. Simply adding “AI fluency” as another competency is too narrow a response. The larger question is whether the leadership model is explicit enough about how enduring leadership capabilities need to be applied differently as AI changes the work—and how those expectations should change as the scope, complexity, and consequence of leadership increase.

Judgment is not the same at every level

Judgment makes the need for that level-specific precision especially visible. What good judgment requires from a first-line leader is not the same as what it requires from a mid-level leader or executive.

First-line leaders work closest to employees, customers, exceptions, and day-to-day priorities. Their judgment often turns on local context: when to follow the pattern, when the situation does not fit, and when to escalate.

Mid-level leaders integrate across teams and systems. Their judgment requires seeing interdependencies, reconciling competing priorities, and translating broader strategy into coordinated action.

Executives make choices about future direction, risk, resources, and organizational capability with broader and longer-term consequences.

AI may change different parts of the work at each level, but it does not erase the different requirements of judgment. More information does not remove the mid-level leader's need to integrate across competing demands. Better scenario analysis does not relieve an executive of choosing which future to pursue. Faster recommendations do not remove a first-line leader's responsibility to understand the person or situation in front of them.

A leadership model should therefore clarify what the same fundamental capability requires as scope, complexity, and consequence increase—and what good judgment looks like at each level.

Three lenses for clarifying leadership expectations as AI changes the work

Together, these lenses help clarify how AI is reshaping leadership expectations: where enduring capabilities need to be applied differently, how those expectations vary by level, and which organization-specific requirements deserve greater emphasis.