Artificial intelligence (AI) has quickly become one of the most talked-about topics in learning and development (L&D). As organizations experiment with AI, L&D teams have understandably focused much of their attention on content creation. They are using AI to generate eLearning, videos, job aids, scripts, assessments and other learning materials faster than ever before.

But speed and volume isn’t the same as impact.

Based on conversations with more than 100 learning leaders across industries, including interviews for my Learning Leaders Spotlight podcast and small-group benchmarking discussions, I believe a more significant shift is emerging. AI isn’t simply changing how quickly we in L&D create learning. It is changing how our learners perform their jobs.

And that creates a much bigger opportunity for L&D professionals.

The Misplaced Focus on Content Creation

In many organizations, AI is simply helping L&D do more of what it has always done. More courses are being created. More content is being produced. The underlying approach to learning remains largely unchanged.

That raises an important question: If training wasn’t improving performance before, why would more training improve performance now?

As AI lowers the barrier to content creation, organizations also face another challenge: an explosion of content that can be redundant, inconsistent or disconnected from the work employees need to perform.

The more important shift may therefore be from creation to curation and from content to context.

Three Stages of AI Maturity

From my conversations with learning leaders, I see organizations moving through three stages in how they use AI in training: adoption, integration and transformation.

1. Adoption: Doing What We Already Do, Faster

Most of the learning leaders I’ve spoken with described AI uses that fall into the adoption stage.

Here, AI is primarily being used to improve efficiency within existing processes. An instructional designer uses AI to develop a first draft. A facilitator uses it to create activities. A team generates assessments or job aids more quickly.

These are worthwhile efficiencies. But the work itself hasn’t fundamentally changed.

In a recent podcast conversation, learning technology strategist Lan Tran described a useful way to think about technology decisions at this stage: Don’t begin with a mandate to use AI and then search for a problem it can solve. Begin with the problem.

Tran shared an example from a global learning environment where traditional voiceover was expensive to produce and difficult to update, while subtitles alone did not adequately support all learners. AI-generated voiceover presented a potential solution but only if the technology could produce sufficiently natural voices across languages.

The important part of the example isn’t the AI voice technology. It’s that the technology followed the learning and business need rather than the other way around.

That mindset becomes increasingly important as organizations move beyond adoption.

2. Integration: Moving Learning Into the Work

The second stage is where things become much more interesting for L&D.

Instead of simply using AI to produce learning more efficiently, organizations begin embedding knowledge and support directly into employees’ workflows.

In another podcast conversation, learning leader Yusuf Sadrud-Din pointed to retail environments as an example. Employees on the floor can use technology to ask a question and receive a step-by-step action guide that helps them respond to a customer or complete a task in the moment.

Traditionally, we have tried to anticipate what employees will need to know and teach it to them in advance. But if employees can reliably access certain information at the moment they need it, we need to become much more discerning about what deserves formal training.

That means asking different questions:

• What does an employee truly need to know and retain?
• What must they practice until they can perform it proficiently?
• What information simply needs to be accessible at the moment of need?
• Where is human judgment critical?
• And what can technology now do for them?

Sadrud-Din makes another important point: training represents only a small part of an employee’s working life. Time spent in training is also time spent away from the job. For employees in roles such as sales, performance support may sometimes be more valuable than taking time away from work to complete another learning event.

That doesn’t mean formal learning disappears.

Another learning leader I spoke with emphasized that organizations will continue to need both learning in the flow of work along with intentional skill building over time. People cannot develop every capability while performing the job.

The opportunity for L&D is therefore not to replace training with performance support. It is to make better decisions about which is appropriate.

That is a very different starting point for learning design.

3. Transformation: Redesigning the Work Itself

Integration changes where support happens.

Transformation changes the work.

At this stage, organizations stop asking only how AI can help employees perform existing tasks and begin asking whether those tasks, workflows and even roles should exist in their current form.

In a recent podcast conversation, AI and workforce transformation advisor Dr. Markus Bernhardt described a common barrier to making that shift. When employees are first given AI tools, they tend to use them within their existing workflows doing the same work more efficiently rather than reconsidering how the work itself could be done. As employees develop greater AI literacy and fluency, they become better equipped to recognize fundamentally different ways of working.

As Bernhardt explains, organizations can show plenty of AI activity without producing meaningful business change. If cycle times, cost structures and workflows remain the same, greater individual efficiency doesn’t necessarily mean the organization itself has changed.

Redesigning work requires understanding how people perform it today. Tran cautions that much of an organization’s real workflow may never have been formally documented. Processes often include workarounds, judgment calls and knowledge that exists primarily in employees’ heads.

Before organizations automate or redesign a workflow, they need to understand that reality.

Otherwise, they risk using AI to automate a process they never fully understood in the first place.

What Learning in the Flow of Work Means for L&D

This progression from adoption to integration to transformation creates a fundamental challenge for learning teams.

When employees can receive contextual prompts, retrieve organizational knowledge, get step-by-step guidance or receive feedback while performing a task, the boundary between learning and working begins to disappear. That changes the role of L&D.

Our expertise becomes less about owning courses and more about understanding the relationship between work, performance and business outcomes.

We need to determine where knowledge should live, when employees need it, how they will access it, which skills require practice and feedback and where technology can reduce unnecessary cognitive load.

It also means L&D needs to become involved earlier.

If we wait until a business leader comes to us requesting training, the workflow and solution may already have been defined. Learning leaders have an opportunity to participate while the work itself is being designed to help determine what should be learned, what should be supported, what should be automated and what should remain dependent on human judgment.

That is a much more strategic role than responding to training requests.

What Learning Leaders Should Do Now

Moving from adoption to integration and ultimately transformation does not require L&D teams to abandon everything they are doing today. It does require them to broaden the questions they ask.

• Measure business impact, not production. Faster development and more learning content may demonstrate efficiency, but they do not demonstrate impact. Start with the business metric the initiative is intended to influence — sales, productivity, quality, customer satisfaction, safety or another meaningful outcome — and determine whether it improved.
• Examine the workflow before designing the learning. Understand how employees perform the job, including the informal workarounds and knowledge that may never appear in a process document.
• Separate what must be learned from what must be accessible. Not every piece of information deserves a place in a course. Determine what employees need to retain, what they need to practice and what can be delivered at the moment of need.
• Look for opportunities to remove unnecessary training. Sometimes the best solution may be better performance support, a redesigned process or technology that eliminates the need for formal training altogether.
• Partner with the business earlier. Don’t wait for the training request. Engage when roles, workflows and processes are being reconsidered so learning expertise can help shape how work will be performed.

A Defining Moment for Learning and Development

Many traditional approaches to learning were designed around separating learning from work. Technology is giving us an opportunity to reconsider that assumption. For learning leaders, that creates an enormous opportunity.

We can continue using AI to make learning more efficient. Or we can help organizations rethink the relationship between learning, technology, work, performance and business outcomes. That is where transformation begins.