AI, she argues, isn’t something leaders can fully understand from a presentation, a training course or a high-level discussion about strategy. Leaders need enough direct experience with it to develop an instinct for what it can do, where it falls short and how it changes the work itself.

For Jess, that makes AI adoption a question of fluency.

Leaders Have to Get Hands-On

Jess believes leaders cannot delegate their understanding of AI.

“You can’t learn it in a linear way. It’s not about a particular tool or workflow or process. It’s really about internalizing intuition around the art of the possible—understanding what the capability allows you to do differently.”

That intuition comes from using the technology against something specific. Leaders don’t need to become engineers, but Jess believes they do need to get far enough “into the weeds” to understand what they are looking at. Without that grounding, it becomes difficult to do what leaders ultimately need to do: help people supervise AI, gut-check its output, apply judgment and know when to question what it has produced.

At RGP, Jess has seen something else happen when senior leaders learn publicly: it gives other people permission to admit what they don’t know. When leaders visibly experiment and learn, vulnerability stops looking like a weakness and becomes part of how the organization develops fluency.

There Is a Big Difference Between Access and Adoption

Giving people an AI tool is relatively easy. Getting them to use it in ways that genuinely improve their work is another matter.

“There’s a space between access and adoption, right? And it’s highly personal.”

Jess thinks that distinction is particularly important in knowledge work. Employees are not performing identical tasks on an assembly line. They have different responsibilities, clients, expertise and sources of friction. The question therefore becomes very specific: What is the friction in your job, and how could AI help resolve that friction or make you better at the work?

That is why Jess is skeptical of overly generic AI training. People often shut down when confronted with broad questions such as What is AI? or How should you use AI? Give them a concrete example, however, and they begin making connections to their own work.

“Engaging individuals in their own adoption is important because you can’t prescribe it to them.”

That may be one of the biggest departures from traditional enterprise technology adoption. There is no single workflow everyone simply needs to learn.

AI Change Management Has No Stable End State

Traditional change programs often have a destination. A new ERP system is implemented. Employees are trained. Processes stabilize.

AI doesn’t behave that way.

Jess compares it to teaching people how to navigate complex questions. Employees need principles and frameworks they can apply under uncertainty. Many of those principles are not actually new. Organizations already know how to think about data governance, cybersecurity, controls and bias. What is new is the context in which those principles have to be applied.

“You think, ‘Oh, I don’t know.’ But you do actually. Let’s go back and grab all those old tools that you used to use and bring them back.”

Her advice: break the problem into its component parts, understand the work involved and apply the same judgment organizations have always used, while accounting for the new risks introduced by probabilistic technology.

There is another practical reality leaders need to accept.

“Our ability to absorb all of the change is not as high as the rate of change.”

Just because the models improve does not mean every organization needs to adopt every capability immediately. Jessica does not believe companies have to remain perfectly at the cutting edge in every process. Choosing where to absorb change is itself becoming a management discipline.

AI Is Changing What Expertise Looks Like

One of the more interesting questions raised by AI is whether younger employees have an inherent advantage because they are more accustomed to the technology.

Jessica sees some truth in that—but only part of the truth.

Early-career professionals can have stronger raw instincts for leaning on the technology. But knowing how to use AI is not the same as knowing whether the result is good.

“The objective of doing anything well still begins with defining what good looks like.”

That requires understanding the business objective, regulatory environment, organizational constraints and the complex system into which change is being introduced.

“The people that are best at really implementing AI are bringing both that level of judgment and understanding of the larger system and process—the ability not just to execute on a defined path, but to derive the right one from their own understanding of everything that’s going on.”

In that sense, experience may become more valuable, not less—provided experienced people are willing to change how they work.

Jess sees consultants and other “mobile talent” as particularly well suited to this environment because they are accustomed to changing contexts, getting up to speed quickly and synthesizing unfamiliar information. AI can dramatically accelerate that ability.

For managers in particular, AI can provoke a deeper question: If the expertise that helped me reach this level is changing, what is my value now?

Jess calls it an identity question. The answer may require people to separate the tools they have mastered from the more enduring qualities that made them good at their jobs in the first place—judgment, synthesis, problem solving, curiosity and leadership.

We Are Probably Asking Too Little of AI

Jessica’s own experience has led her to another conclusion: many people are still dramatically underestimating what they can ask AI to do.

Her example: she needed a new garbage can and initially asked ChatGPT where she could order one. Instead, the system asked whether she simply wanted it to handle the task.

What followed became a small but meaningful lesson in delegation.

“Just like you would with a really competent individual that’s working for you, you’re never asking enough. You’re never sort of leaning back and saying, ‘How could you help me? And what should I let you do that I’m not actually letting you do?’”

That change in mindset, from asking AI for information to asking what work it can take on is, for Jess, part of developing real fluency.

And it is why she keeps returning to direct experience. Watching someone encounter that possibility for the first time is different from explaining it to them.

“I would die on that hill of go touch it, go feel it, let’s talk about it next.”

Looking Ahead: Three Priorities for Leaders

Build fluency through real work, not training.

Employees need opportunities to experiment with problems that actually matter to them. Specific examples allow people to develop an intuition for AI that abstract training rarely creates.

Teach people how to think when the technology keeps changing.

The goal cannot simply be mastery of today’s tool. Organizations need to help employees develop principles for working under uncertainty—how to judge, supervise, question and safely use capabilities that will continue to evolve.

Don’t mistake perfect conditions for prerequisites.

Jessica is particularly wary of organizations using imperfect data as a reason to defer AI experimentation. As she puts it, clean data is certainly a virtue, but it does not have to be “a gate to start.”

The Bottom Line

People have to develop their own understanding of what AI makes possible. They need room to experiment, permission to learn publicly and enough confidence to reconsider practices they may have spent years mastering. And leaders have to do the same.

The organizations that become truly fluent in AI will be the ones that build the capacity to keep learning as the technology changes, and whose people know enough to ask a much more useful question:

“What am I still doing that I should be asking AI to help me do?”