Leadership and change management in the AI era: from automation to augmentation

By Richard Steele (McKinsey Quarterly) · July 1, 2026.

The article, written by Richard Steele, a partner at McKinsey in New York, starts from a clear premise: traditional change management was already under pressure before AI arrived, and the AI boom has now added another layer of disruption that calls for a complete rethink of the role of leadership.

Steele explains that organizations used to manage discrete, time-bound changes, such as implementing a new system, restructuring a function or rolling out a defined process, with relatively stable end states. Today, by contrast, the environment is one of continuous disruption. The number of change initiatives that executives at large corporations have taken on has soared over the past decade, to the point where many organizations are juggling dozens or even hundreds of initiatives deemed top priority at the same time.

This overload has a direct effect on employees: many are choosing to opt out of discretionary efforts tied to these initiatives, even when taking part could benefit their careers. The author cites research from the McKinsey Health Institute finding that mental health and well-being problems are widespread in today's workforce, with significant shares of employees reporting symptoms of anxiety or burnout. According to Steele, "something fundamental isn't working."

The article notes that the AI boom has had three additional surface-level effects on change management. First, it has further increased the number of initiatives companies are pursuing or feel they must pursue. Second, because AI itself is advancing so quickly, executives feel pressure to speed up not only AI experimentation and adoption but other change initiatives as well. For example, AI works best with modern technology systems and high-quality enterprise data, and it can also accelerate software development, so many companies are speeding up efforts to modernize legacy technology platforms. That in turn raises expectations for the pace of transformation in general. Third, because AI could automate parts of many jobs in the future, it has pushed employee anxiety and labor-market uncertainty, which were already too high, even higher.

Steele particularly stresses that these problems are made worse because AI use has not spread far enough into the executive ranks. Executives tend to adopt AI tools less often, partly because of the nature of their roles and their own time pressures, and partly because they are not in a position to experiment as much as more junior employees are.

The author offers direct, personal advice: leaders should step back and reimagine their own roles as augmented by AI, with much of their routine work automated. Doing so, he argues, will help them realize that what they most need to accelerate is their own learning.

Steele describes how what used to be a widening digital divide has now become an "AI divide." It goes beyond simply using tools to include understanding what new AI tools can do and how long tasks should now take. As a result, leaders lose credibility, and employees notice. Deep down, many leaders know this too, and some have lost confidence in their ability to lead on the basis of accumulated experience.

To address this, the article argues that leaders need to reimagine their own roles as reshaped by AI. AI can help them ask illuminating questions about the future of the business and how value will be created: how to think differently about risk, innovation and new business models; where the organization should accelerate adoption and where it should be more selective; and which activities no longer matter as much and can be stopped to free up capacity and give employees some breathing room.

Steele insists that stepping back to reimagine leadership roles, with much of the routine work automated, puts leaders in a better position to help others experiment with new tools, because they will have accelerated their own learning first.

The author uses a particularly telling analogy, comparing current progress in AI adoption to being, say, three years into the development of steam power. At that point, the potential of a new technology is not yet fully understood because its impact is still hard to measure. Seen this way, it is clear that leaders can do more to build learning cultures. They should see leadership itself as an exercise in helping others learn. That includes moving away from a directive stance toward a question-asking role that generates energy rather than adding to collective exhaustion.

However, the author also warns that leaders need to think more strategically about the gains AI could deliver. Many organizations are choosing obvious applications focused on automating tasks when they could be planning bigger, more imaginative bets. He gives a concrete example: instead of automating tasks to save a fraction of a researcher's time, organizations could empower those researchers to use AI to run thousands of trials at once.

The article's central conclusion is that the opportunity for leaders lies in viewing AI from the standpoint of human augmentation, not just workforce automation. These important mindset shifts can help leaders manage disruption in new ways. Change management, Steele argues, is now about reinvention. It requires leaders to move beyond discrete programs and initiatives and instead lead continuous adaptation across their organizations. Their own roles can evolve as well, from directing change and providing answers to creating the conditions for learning and experimentation. Rather than simply implementing new processes or structures, they can rethink how a company creates value at its core while helping employees navigate ongoing uncertainty and disruption.

The article was edited by Barbara Tierney, a senior editor in New York, and notes that Richard Steele is a partner in McKinsey's New York office. The email also links to other related pieces by the author. One is 'Change is changing: How to meet the challenge of radical reinvention,' on how leaders should rethink their traditional change-management tools. The other is 'Bias Busters: Escaping the echo chamber at the top,' on biases such as egocentric anchoring and authority bias, which can amplify leaders' views and silence those of others.