If you are a learning and development (L&D) leader being told that you need train the business on how to use artificial intelligence (AI), I have news for you. They are already way ahead of you; they just aren’t talking about it. AI adoption exploded in 2025, and much of it is happening in the shadows.

According to new research from Cornerstone, 80% of workers already use AI, but the majority don’t disclose it to colleagues or managers. Not out of fear or embarrassment, but because they aren’t sure what “good” AI use looks like, where the guardrails are or what their company expects of them.

With the rapid evolution of AI in recent years, many companies find themselves somewhere between encouragement (“Go experiment with AI!”) and enablement (“Here’s how to use it responsibly and effectively”).

For learning leaders, this is an opportunity to lead, but getting it right is hard for myriad reasons. AI literacy means something different to every team, every role and every individual in the organization. Old L&D playbooks do not apply.

I spoke with three leaders sitting at the center of this shift — one driving internal AI adoption at a tech company, one researching how organizational learning must adapt as AI accelerates, and one leading global analysis on AI behavior in modern workplaces. Their perspectives indicate that the path forward for L&D will require clarity, curiosity and transparency — not static training.

Why Employees Aren’t Talking About Their AI Use — And Why That Matters

Cornerstone’s research revealed a surprising paradox: younger workers use AI the most but are the least likely to disclose it. Mike Bollinger, global vice president of strategic initiatives at Cornerstone, said this signals both cultural and clarity gaps.

“People are being encouraged to use AI, but without proper training or clear expectations, they’re reluctant to say how they used it,” said Bollinger.

The result is what Bollinger calls “shadow AI.” Employees are using AI quietly, without guardrails, feedback loops or shared learning.

If people learn in private, L&D can’t measure behavior change, understand risks or scale what’s working. To fix this, transparency needs to come from the top, said Bollinger.

“The best way to create transparency is for leaders to talk about their own AI use. That makes it a safer place for others to share,” he said.

Why AI Training Requires a Different Approach for L&D

Traditional training is built on defined outcomes, structured curricula and predictable skill progression. AI breaks every part of that model.

“This isn’t the cycle we’re used to in L&D. It can’t be. And if we pretend it is, we’re going to have less and less relevance in this space,” said Melissa Brown, learning and development manager for Holland & Hart, LLP.

She describes AI training as a paradigm shift similar to the move from in-person training to virtual delivery — a shift that also broke L&D models until teams rebuilt their practices.

But the AI shift is even faster and more unpredictable. Tools evolve weekly. Roles differ widely in how AI applies. And employees are teaching themselves faster than L&D can build modules.

“We could build a whole program today and it would be outdated next week. So we can either keep building programs people ignore, or we can become better at facilitating self-directed learning,” said Brown.

What AI Literacy Should Look Like in Your Workforce

Erin Goldman, senior manager of people development at ZipRecruiter, who is rolling out AI enablement across her organization, distinguishes between AI literacy and AI fluency like so:

“Literacy is knowing how to use the tool responsibly, when to use it and how to partner with it in daily work. Fluency is innovating with AI and using it for competitive advantage,” said Goldman.

But these definitions are highly fluid, Goldman notes. A pilot group experimenting with AI tools and workflows could quickly start to move toward fluency. On the other hand, “things are changing in this space all the time. Something that you are fluent in today could be a moot point next week,” she said.

Rather than try to pin down a definition of AI literacy, L&D’s goals should focus on:

• Proficiency, not mastery. Can employees find the tool? Use it effectively? Fact-check it? Apply it to real tasks?
• Safe experimentation. Employees need explicit permission — and expectations — to try things, fail and share what they learn. They also need to know where they can go for help if they are stuck or want to brainstorm, says Brown. “They are in the driver’s seat, not us. It’s very unlikely that L&D will be a better subject matter expert on their specific AI use case than they are. Give up control; focus on being a great partner,” said Brown.
• Role-specific application. Find out what AI literacy means for different parts of the business. Be a part of the conversation as different teams start to figure out what AI can do for them, advised Goldman.
• Workforce differences. As Cornerstone’s research reveals, AI usage looks different across age groups, and stigma and misconceptions persist. L&D should account for these behavioral patterns and psychological barriers.
• Behavior signals, not ROI. Brown said, “You can’t measure early AI training with ROI. Curiosity is the metric.” Goldman’s team looks at an increase in the number of conversations with Gemini happening throughout the business as a success metric.

The bottom line is that literacy is not a course or even a clearly-defined end state. It’s a progression that L&D shapes in partnership with the business.

Where L&D Should Begin

When L&D receives the directive to “increase AI literacy,” all three experts said the first move is to push back — strategically.

Bollinger said, “Your return volley should be: ‘What part of the business do we expect it to move the needle on?’”

Brown echoed this: “If we’re told to increase literacy without the autonomy to define what literacy means for whom, at what level and how fast, we’re just shooting in the dark. You can’t be successful if you don’t know what the business is really trying to accomplish.”

AI training only makes sense when it ladders up to use cases that matter — whether that’s foundational awareness, responsible use guidelines, role-specific enablement or deeper fluency for selected teams.

In the absence of defined business goals, Goldman recommends pointing AI training toward efficiency gains.

“Aren’t we all trying to do more with less? It’s a universal business need,” said Goldman. Rather than stalling on AI training while waiting for a clear directive, start to explore these tools’ ability to help you do more with less, she said.

Importantly, L&D should not do this alone.

“You’re going to need help. Start with pilot groups. Work with experts. Partner with tech, legal and business leaders. Don’t try to prescribe everything yourself,” said Bollinger.

Goldman emphasized the importance of cross-functional alignment early: “AI raises questions around data privacy, security and governance. Legal was one of our earliest partners. They helped define our rules of the road. Tech helped us understand what was possible. L&D helped with enablement. None of this happened in isolation,” she said.

“We can’t tell people what to do with AI without understanding what their roles require,” said Brown. “So rather than prescribe, we guide. We ask questions. We help them brainstorm. We point people to each other’s experiments.”

The Opportunity — and the Risk — for L&D

AI creates one of the biggest enablement challenges L&D has ever faced and one of the biggest openings to lead.

“Things that come with huge opportunity also come with huge risk. If L&D doesn’t rethink how people learn, AI will expose that quickly,” said Bollinger.

Employees are already learning AI on their own. They’re experimenting in quiet, fragmented ways. They’re eager for direction.

This is L&D’s moment to step in — not to be the AI experts, but to help everyone learn together. You can bring clarity where there’s confusion, build psychological safety and help people share what’s working. That’s how you accelerate real adoption across the organization.

Explore the AI Adoption and Workforce Readiness Certificate today.