From Optional to Essential: Three AI Leadership Must-Haves

Forbes

04-04-2025

Gemini-generated image
Last Fall, I began exploring and writing more about our core responsibilities as learning and talent leaders interacting with AI: building AI literacy, fostering experimentation, and leading through change. After covering the first two, I assumed writing about change leadership would be straightforward. It wasn't. This isn't like any technology transition we've led through before.
AI isn't just another technology leaders must adopt; it fundamentally reshapes our relationship with work itself. Culture, managers, and teams have always mattered, but how we must now leverage them has dramatically changed. Successful AI integration demands explicit shifts: organizational culture must deliberately embed practices that build ongoing trust; managers must transition from expert answer-providers to skilled questioners who model curiosity and judgment; and teams must evolve from isolated task performers to adaptive collaborators who fluidly engage internally and across functions. These aren't optional refinements—they're essential shifts for organizations to fully realize AI's potential.
My own journey illustrates this unprecedented challenge. Just as I developed a workflow with Claude 3.5, the 3.7 Sonnet version emerged with capabilities that fundamentally changed how I leveraged it. ChatGPT 4.0's release forced me to rethink my customized prompts, while Deep Research transformed the research part of my writing process in ways I hadn't imagined possible. Each evolution didn't just add features—it shifted my entire collaboration system with AI.
I joined the "Women Defining AI" community, learned to build a chatbot based on my book ReCulturing, pivoted toward an app through Vercel, and then a colleague passionately advocated for Lovable—each tool compelling yet disruptive enough to reset my approach. Even my writing process shifted dramatically after enrolling in Every's "Writing with AI" course, demoting AI from editor-in-chief back to my trusted assistant. Now, with agentic AI leading the conversation, I'm experimenting again with AI-assisted hotel reservations for our trip across the Middle East.
This isn't about keeping up with new AI features—it's about fundamentally reshaping how we lead and work. Unlike previous technology transitions, which followed predictable roadmaps and stable best practices, AI integration demands continuous adaptation. We aren't simply teaching teams to use new tools; we're guiding them to form working relationships with technologies that think, learn, and evolve alongside them. This introduces unprecedented psychological and practical challenges for change leadership.
Past technology shifts—from desktops and mobile devices to cloud computing—allowed gradual implementation over stable timelines. AI disrupts this rhythm dramatically, changing weekly or even daily. It isn't a static system we learn once and upgrade occasionally; AI continually surprises us, adapts, and pushes us to rethink our roles and interactions. For leaders, this creates an essential paradox: How can we effectively lead others through changes we ourselves are still learning to understand?
Traditional change models, such as the familiar "unfreeze-change-refreeze" approach highlighted by Michael Mankins and Patrick Litre in their recent HBR article, 'Transformations that Work,' simply don't apply. Bain's research confirms the urgency of shifting toward continuous transformation: while over one-third of large organizations constantly launch change initiatives, only 12% achieve lasting success. Organizations that intentionally embed trust-building practices into their culture significantly improve their odds of sustainable AI integration.
The stakes for leaders have never been higher. Unlike a typical CRM rollout, mismanaged AI adoption can quickly erode trust, amplify biases, or permanently disrupt team dynamics. Effective change leadership requires promoting active experimentation while establishing clear ethical guardrails—a balancing act unique to AI's complexity.
Ultimately, successful AI adoption depends far less on mastering the latest tools or algorithms and far more on intentionally strengthening the foundational human elements essential to change: culture, managers, and teams.
The fundamental challenge of AI adoption isn't technical—it's emotional. Employees aren't just learning new tools; they're grappling with existential questions: Will AI make my expertise irrelevant? Can I trust its outputs? Who's accountable when AI makes mistakes? These concerns aren't theoretical. McKinsey found that although 78% of organizations adopt AI, employee resistance remains a primary barrier. Similarly, multiple research reports over the last ten years show that 70% of digital transformation failures result not from technology but from organizations' inability to shift habits and behaviors. Without trust, AI strategies merely automate inefficiencies and amplify existing fears.
At Udemy, we experienced this firsthand. Initially focused primarily on AI training and experimentation, we quickly discovered that real success required rebuilding the psychological contract between leaders and employees. Instead of pretending AI would seamlessly enhance everyone's work, we openly acknowledged the messy reality: AI would disrupt familiar practices, create new anxieties, and force us to redefine successful collaboration.
As I detail in ReCulturing, sustainable organizational change—especially involving AI—requires aligning strategy explicitly with behaviors and practices. For instance, our value of "Courageously Experimental" translated directly into trust-building behaviors:
The result wasn't just increased AI adoption—it was a deeper level of organizational trust. IBM's research supports our experience: Their 2024 CEO Survey found that 64% of leaders now recognize that AI success "depends more on people's adoption than the technology itself." Trust isn't built through grand AI initiatives but through daily practices that demonstrate we value human judgment as much as artificial intelligence. This foundation becomes essential because AI isn't a one-time change—it's a continuous evolution that requires a sustained commitment to experimentation, transparency, and, above all, trust.
While culture creates the context for trust, managers are the ones who make it real. They must navigate paradoxes that didn't exist with previous technologies: How do we maintain leadership authority when AI tools might know more than we do? How do we drive automation while ensuring your team continues to develop critical skills? When should we trust AI's recommendations, and when should we lean more heavily on human judgment?
McKinsey's research also highlights this tension: "The biggest barrier to scaling AI is not employees—but leaders who aren't steering fast enough." Managers need to actively model curiosity and questioning to encourage employees to ask better questions about themselves, each other, and the AI itself. Active questioning isn't new, but organizations vary greatly in how deeply they embed it as a managerial practice and develop it as a core skill. I still work with many managers who think their value comes from being the expert and being able to answer questions rather than knowing when and how to ask the right question. This skill of asking questions is important for managers' teams as they re-think the collaborative system of their teams working with AI. Managers need to model asking questions so that employees can ask better questions of themselves, their colleagues, and the AI.
Throughout my twenty-five-year career developing leaders, I've consistently found that strong managers are the key to successful change initiatives. At Udemy, this proved especially true with AI adoption. When we initially rolled out AI tools, the teams that successfully integrated AI weren't those with the most technical expertise. They openly shared their AI learning curves, including mistakes and uncertainties, protected their teams' time for experimentation while maintaining clear performance standards, and deliberately balanced AI and human contributions. Additionally, they engaged actively in team discussions around redefining roles and emphasized the importance of human skills such as strategic thinking, relationship building, and complex decision-making.
Conversely, AI adoption stalled when managers avoided or pushed the technology too aggressively without addressing team concerns. While the manager's role has gradually evolved for years—from being the primary source of answers to becoming a skilled curator of insightful questions—AI has dramatically accelerated this shift. The evolution from expert answer-provider to strategic questioner is no longer optional; it's now essential. Managers who continue to rely solely on their own expertise will increasingly struggle, whereas those who consistently ask critical questions—Which AI applications genuinely advance our goals? How can we preserve and enhance our unique human strengths? What new capabilities must we build next?—will position their teams and organizations for successful AI integration.
The fundamental challenge teams face with AI isn't just learning new tools—it's the unprecedented speed of change. Traditionally, teams adapt through structured learning and development sessions and gradual implementation. However, AI's rapid evolution disrupts this familiar approach. Processes documented today might become obsolete tomorrow, and best practices often become outdated before teams can fully adopt them. While organizations have discussed upskilling extensively over recent years, new MIT research highlights that traditional training methods are still insufficient for AI adoption, highlighting that 55% of organizations report workforce skills becoming outdated within months, not years.
Collaborative and social learning have long been cornerstones of effective development. Over the past two decades, formal training combined with "on-the-job" experiences has evolved into blended learning strategies. Yet, in the AI era, cohort-based collaborative learning isn't merely beneficial—it's essential. Traditional methods, such as slide decks and workshops, simply can't match AI's relentless pace, often becoming outdated before implementation.
To address this challenge, our team created learning networks where members could explore, share discoveries, challenge assumptions, and build on each other's experiments in real time. What began as informal conversations quickly evolved into a dynamic community capable of adapting as rapidly as AI itself. Similarly, external communities such as Women Defining AI provided fresh perspectives and inspiration, demonstrating firsthand that knowing what and how to ask is as critical as knowing the answers.
Yet effective collaboration alone isn't enough. Teams must also cultivate what computer scientist Jürgen Schmidhuber first termed adaptive confidence—the ability to learn, adjust, and innovate within uncertain environments continuously. Introduced in Schmidhuber's 1991 technical report, Adaptive Confidence and Adaptive Curiosity, the concept originally described how adaptive systems reliably navigate uncertainty. Today, adaptive confidence extends beyond skill-building; it requires fundamentally resetting team practices. Rather than merely mastering new tools or skills, teams must embrace ongoing experimentation, regularly challenge assumptions, actively engage in iterative learning, and collectively apply human judgment to understand and leverage AI's evolving capabilities.
Embedding these strategic shifts into our culture, managerial approach, and team dynamics positions us to actively adopt, shape, and successfully integrate AI. Potential actions you can take today in these three areas are:
Effective change leadership in the AI era requires intentional trust-building, managers who actively model strategic questioning, and teams capable of dynamic collaboration. How are you preparing your culture, managers, and teams for these new realities?