Most AI initiatives do not underperform because organizations selected the wrong technology. They underperform because technology changes what people are capable of doing much faster than organizations change the conditions under which people work.
Where Many AI Transformations Quietly Lose Momentum
Conversations about AI transformation almost always begin with technology. Leadership teams discuss vendors, security, licensing, governance and deployment plans because these are the visible elements of the programme. They come with budgets, timelines and clearly defined responsibilities. Much less attention is devoted to the quieter questions that emerge once the technology is in use. How will managers lead teams whose work has fundamentally changed? Which decisions should remain with people, and which can now be delegated to AI? What happens to performance expectations when employees suddenly have access to significantly greater cognitive capacity than before?
McKinsey has repeatedly argued that successful AI transformation is largely a business transformation rather than a technology project, a conclusion that reflects what we observe in organizations ourselves. The difficulty is not that leadership teams disagree with this principle. The difficulty is that organizational redesign is inherently more ambiguous than technology deployment. There is no implementation guide for redefining judgement, accountability or collaboration, even though these questions often determine whether AI becomes a source of lasting competitive advantage or simply another productivity tool.
It is this imbalance that explains why so many organizations report widespread experimentation with AI while struggling to demonstrate equally meaningful business impact. The technology has arrived. The organization surrounding it has often changed much more slowly.
AI Changes More Than Workflows
One consequence of AI receives surprisingly little attention. Most discussions focus on which activities become faster or more efficient, yet the more profound shift concerns the changing nature of human contribution itself.
As AI assumes a growing proportion of routine cognitive work, the value people create increasingly lies elsewhere. Employees spend less time gathering information and more time interpreting it. They coordinate fewer routine processes and exercise more judgement in situations where context, ethics and experience still matter. Work becomes less procedural and more discretionary. The question therefore ceases to be whether AI improves productivity. The more important question becomes whether organizations are preparing people for a different kind of work altogether.
This is where many transformation programmes become unnecessarily narrow. Process redesign undoubtedly matters, but processes do not redesign themselves. Employees still need to understand how success will be evaluated, managers need to lead work they may no longer perform personally, and leadership teams need to decide which human capabilities will become more valuable rather than less. Those conversations determine whether AI changes the quality of work or merely the speed at which existing work is completed.
The Organizations Creating the Greatest Value Are Redesigning Expectations
The organizations making the greatest progress with AI rarely distinguish themselves through access to fundamentally different technology. More often, they distinguish themselves through the questions they begin asking once the technology is available.
Instead of concentrating exclusively on automation opportunities, they examine how decision-making is changing across the organization. They explore whether existing management structures still reflect where expertise now resides, whether employees have sufficient autonomy to use newly available cognitive capacity well, and whether performance systems continue rewarding behaviours that no longer create the greatest value. In other words, they recognise that introducing AI inevitably changes the expectations placed on both employees and managers, and they treat those changing expectations as a strategic issue rather than an operational consequence of deployment.
This shift is less visible than purchasing new software, yet it has considerably greater implications for organizational performance. Employees rarely struggle because they cannot access AI. They struggle because the assumptions governing their role have changed while the surrounding management system continues behaving as though nothing has.
Leadership Becomes More Important as Technology Becomes More Capable
Perhaps the most interesting paradox of AI transformation is that the increasing capability of technology appears to increase, rather than reduce, the importance of leadership.
When routine work becomes easier, questions of judgement become more prominent. When information becomes abundant, interpretation becomes more valuable. As AI expands what individuals can accomplish independently, managers spend proportionally less time supervising execution and considerably more time helping people exercise judgement, navigate ambiguity and make sound decisions in situations where no algorithm can yet determine the appropriate course of action.
For many organizations, this represents a much larger adjustment than adopting the technology itself. Leadership practices developed for environments characterised by procedural work and predictable decision-making do not automatically translate into environments where employees are expected to exercise greater discretion. Managers therefore need support that extends well beyond learning how AI functions. They need to understand how leadership itself changes when the nature of human work changes.
Looking Beyond Adoption
This is one of the reasons we encourage leadership teams to evaluate AI transformation through a broader lens than adoption alone. Usage statistics, training completion rates and productivity improvements provide valuable information, but they reveal relatively little about whether the organization is genuinely adapting to the new capabilities AI creates.
A more informative conversation examines whether managers are redefining performance expectations in response to changing work, whether employees understand where human judgement has become more valuable rather than less, how confidently teams are making decisions alongside AI, and whether the additional cognitive capacity technology provides is being invested in innovation, customer value and better decision-making rather than simply increasing the volume of work completed. These questions are often more difficult to answer than adoption metrics, but they also explain far more about why some organizations continue generating value long after deployment while others plateau once the initial enthusiasm fades.
Over time, these differences become cumulative. Organizations that redesign leadership, managerial practice and human capability alongside technology gradually establish ways of working that competitors find difficult to replicate. Those that limit their ambition to technology deployment often discover that productivity improves while the broader organization changes remarkably little.
The CEO’s Role Is Changing
The most effective CEOs increasingly approach AI as an organizational transformation that happens to involve technology rather than as a technology programme that requires organizational support. That distinction shapes where they invest their attention. Alongside questions about platforms, governance and implementation, they devote equal energy to understanding how work is changing, what managers now require from their teams, which capabilities deserve further development and how leadership itself must evolve in an environment where human contribution is becoming progressively more centred on judgement, creativity and adaptation.
Technology will undoubtedly continue advancing. Whether organizations convert those advances into sustained business performance depends far less on the pace of technological innovation than on the pace at which leaders help their organizations adapt alongside it. In our experience, this is where the greatest opportunities now exist, not because technology has become less important, but because its value is increasingly determined by the quality of the human system surrounding it.