For most of the last half-century, organizational change followed a predictable arc. Leaders defined a future state, guided the business through a structured transition, and waited for new ways of working to take hold. The goal was always stability.
That approach was built on a quiet assumption: that change had an end point, a moment when the work would be done and stability would return. Lewin called it “refreezing.” Kotter built his change models’ eighth step around it: anchor change in culture, declare victory, let the organization settle. The premise embedded in both was the same – change is a journey with a destination.
In the AI era, that promise is increasingly difficult to keep. Change arrives continuously, reshaping tasks, roles, and workflows in real time so that by the time an organization reaches its intended destination, the ground has already shifted, and the next wave of transformation is already underway.
The strain on traditional change models isn’t theoretical. Five structural patterns have emerged across organizations navigating AI transformation, and they are proving remarkably consistent:
These are structural symptoms of a model built for a different era of change that assumed disruption was temporary and stability would return. Three fundamental breaks explain why that assumption no longer holds.
Break 1 — You Can’t Refreeze Stabilization is now a temporary state measured in weeks, and new ways of working become old ways quickly.
Break 2 — The Bottom Knows Before the Top AI’s impact shows up first in workflows and frontline teams sense the change first; leaders must interpret these shifts.
Break 3 — Success Is Adaptation Capacity Competitive advantage now comes from learning velocity, and the organizations that succeed are building the capacity to change.
The World Economic Forum reinforces the scale of this: 95% of organizations surveyed in 2024 went through more than two major transformations in the past three years, and 61% went through more than four. Sequential frameworks were not built for this pace.
These three breaks trigger a compounding crisis. Governance models built for annual planning cycles can be strained under AI’s velocity. Trust fractures when communication can’t keep up, when early adopters surge ahead while others lag, and when uncertainty goes unacknowledged. As Council Advisors’ co-founder and CEO, Dave Niles noted in the Financial Times, successful AI diffusion depends on trust embedded in leadership behaviors, cultural norms, and daily operating systems, not technology deployment alone. Harvard Business Review research makes the stakes concrete: employees in high-trust workplaces experience 74% less stress, 50% higher productivity, and 40% less burnout.
The evidence from organizations navigating this well points to a consistent pattern that is about a different relationship with uncertainty itself. In fact, it requires a more adaptive leadership approach.
Harvard’s Ronald Heifetz, whose work on adaptive leadership is featured the Charter x Council Advisors AI playbook, draws a precise distinction that cuts to the heart of it. Technical problems have known solutions. Adaptive challenges do not. AI is unambiguously the latter, which means the instinct to reach for the change management playbook, declare a future state, and drive toward it is exactly the wrong response. The very act of trying to refreeze is the trap.
What the evidence shows instead: the leaders navigating this moment most effectively have stopped treating AI deployment as a transformation with an end point and started treating it as a permanent operating condition. They run small experiments rather than making large bets, push decision-making to the people closest to the work, and communicate direction without promising certainty. Harvey’s chief people officer Katie Burke captured it plainly: the goal is not to calm the seas, but to develop better sailors. Duolingo’s “FrAIdays” – two hours every Friday dedicated to AI experimentation across the organization – reflects the same logic: building adaptive muscle is an ongoing operating practice.
None of this replaces Kotter or Lewin. Where change is a technical problem with a known solution, the classics still apply. But AI has made adaptive challenges the rule rather than the exception, and the leaders who treat every deployment as a journey toward a stable new state are measuring themselves against a finish line that keeps moving.
The question is no longer How do we refreeze? It is How do we lead when refreezing is no longer an option?
We use data collected by cookies to analyze traffic on this website. By clicking “Yes,” you agree to our use of cookies as described in our Privacy Policy.