Part 3 of The AI Skills Paradox Series
Chief Learning Officer "June" — Follow Her Story
June's team was getting better at distinguishing between learning that needed to be easy and learning that needed to require effort. They were using AI to remove unnecessary friction while creating more opportunities for employees to practice, make decisions, receive feedback, and try again. The result was a learning strategy increasingly focused on capability rather than simply delivering information.
Then a business leader came to June with a familiar request: managers across the organization needed leadership training.
Not long ago, fulfilling that request would have required weeks of discovery, design, development, and review. Now, AI could generate a polished leadership curriculum almost immediately. Within minutes, it could propose learning objectives, create modules on coaching and feedback, develop realistic scenarios, write knowledge checks, recommend reinforcement activities, and produce manager discussion guides.
The output was impressive. It was also almost entirely disconnected from the problem the business leader was trying to solve.
June realized that AI had made it remarkably easy to answer the request without understanding the need. The organization could become exceptionally efficient at creating learning while becoming less disciplined about determining why learning was necessary in the first place.
That introduced the next challenge in the AI Skills Paradox: when content becomes abundant, context becomes scarce and increasingly valuable.
AI Is Very Good at Answering the Question You Ask
Generative AI has dramatically expanded what learning teams can create. Give it a topic and it can produce a course outline, facilitator guide, assessment, simulation, case study, video script, coaching framework, or reinforcement plan. Give it more information and the outputs become increasingly sophisticated.
That capability creates enormous value for L&D. Work that once consumed significant development time can be accelerated, allowing teams to focus their attention elsewhere and respond more quickly to changing business needs.
The challenge is that AI generally begins with the problem it has been given.
If the request is "create leadership training," AI can create leadership training. If the request is "build a course on consultative selling," it can produce a course on consultative selling. If someone asks for a coaching program, AI can develop a thoughtful coaching program.
What it cannot safely assume is that any of those interventions address the actual performance problem.
June's leadership request illustrated the distinction. The organization didn't really need "leadership training." It needed managers to demonstrate different behaviors in specific situations. Before her team could determine what development experience would help, they needed to understand which behaviors mattered, why managers weren't demonstrating them today, and what conditions surrounded the problem.
AI could accelerate the solution. L&D still needed to diagnose the problem.
Stop Taking Training Orders
For decades, L&D teams have often operated through a familiar service model. A business stakeholder identifies a problem, requests training, and the learning organization responds by designing an appropriate program.
That model made some sense when creating learning required specialized expertise and substantial production resources. L&D was one of the few functions capable of turning business knowledge into scalable training, so much of its value naturally became associated with producing and delivering learning.
AI changes that relationship.
When a business leader can generate a reasonable training program independently, being exceptionally good at fulfilling training requests becomes less differentiating. L&D creates greater value by helping the business determine whether training is the appropriate response in the first place.
In his memo to Unboxed employees, President Dave Romero frames this distinction through a series of questions. Before building anything, he argues that learning professionals should understand what people are doing today, what needs to be different tomorrow, and why the desired behavior isn't already happening. The cause might be knowledge, skill, process, motivation, leadership, environment, or some combination of those factors. The Work We Are Really In
Those questions fundamentally change the role of L&D. Instead of functioning primarily as an order taker for training, the learning organization becomes a performance partner capable of diagnosing what the business actually needs to change.
The Same Performance Problem Can Have Very Different Causes
June returned to the leadership request and started asking questions.
The business leader believed managers weren't coaching employees frequently enough. The obvious solution appeared to be coaching training, but further investigation revealed a more complicated picture.
Some managers genuinely didn't know how to conduct effective coaching conversations. They needed skill development.
Others understood coaching well but struggled to identify which behaviors employees needed to improve. They needed better performance insight.
Some managers wanted to coach but spent most of their time in operational meetings and administrative work. Their problem was capacity.
Others believed their role was to solve problems for employees rather than develop them. Their challenge involved expectations and leadership culture.
A single statement—"our managers need coaching training"—actually represented several different performance problems.
A generic program might help one group while doing relatively little for the others.
This is why diagnosis matters. Two employees can demonstrate the same performance gap for entirely different reasons, and the appropriate intervention depends on understanding what is preventing the desired behavior.
L&D's job isn't simply to ask, "What should we teach?"
It is to determine, "What needs to change, and what will actually help change it?"
Context Determines Whether Learning Belongs in the Business
Once June's team understood the underlying problem, another challenge emerged. Even when skill development was appropriate, generic best practices weren't enough.
A manager might need to learn how to provide feedback, but effective feedback looks different depending on the organization's culture, operating model, workforce, and expectations. A salesperson might need to improve discovery, but a useful practice scenario must reflect the customers, products, objections, competitive environment, and buying process that seller actually encounters.
Context determines whether an experience feels like training or feels like work.
Dave makes this distinction directly in his memo. AI can generate a scenario, but determining which scenario matters requires understanding the learner and the business. AI can generate feedback, but someone must first understand which behavior the organization is trying to change. AI can generate an assessment, but the organization still needs to determine whether answering questions or completing a course provides meaningful evidence that someone can perform.
That context includes factors such as:
- The organization's business strategy and priorities
- The employee's role and working environment
- The customers or colleagues they interact with
- The situations in which performance matters most
- The behaviors associated with successful performance
- The organizational constraints influencing those behaviors
- The consequences of getting the behavior right or wrong
- The evidence that would demonstrate meaningful improvement
The more context L&D understands, the more effectively AI can be used. Better context produces better prompts, better scenarios, better feedback, and ultimately more relevant development experiences.
The future isn't human judgment competing with artificial intelligence. It is human judgment giving AI enough context to become genuinely useful.
- Business acumen to understand what matters to the organization.
- Performance consulting to identify the real cause of a business problem.
- Behavioral analysis to define what people need to do differently.
- Learning science to determine what experiences support development.
- AI fluency to use technology effectively and responsibly.
- Practice design to create opportunities for application and feedback.
- Measurement to determine whether capability actually improved.
- Human judgment to connect all of those elements in the right way.
Content Abundance Changes What Expertise Looks Like
For much of L&D's history, creating high-quality content required specialized expertise. Instructional designers knew how to structure information, write learning objectives, develop assessments, design interactions, and translate subject matter expertise into experiences employees could understand.
Those capabilities still matter, but AI is changing where the greatest value resides.
As Dave observes in his memo, producing high-quality learning was historically difficult and expensive enough that production itself carried substantial value. AI is creating a world in which organizations can produce enormous volumes of courses, videos, scenarios, assessments, job aids, and communications far faster than before. The risk is becoming "incredibly content rich" while becoming "context poor." The Work We Are Really In
In that environment, expertise increasingly moves upstream.
The most valuable learning professionals will not simply be those who can produce the best course. They will be those who understand the organization deeply enough to know whether a course should be produced at all.
They will understand how to investigate performance, identify behavioral gaps, distinguish symptoms from causes, translate business priorities into capabilities, and design experiences that reflect the reality employees encounter at work.
AI lowers the barrier to content creation. It raises the value of knowing what deserves to be created.
Human Judgment Becomes More Important as AI Gets Better
This creates an interesting paradox.
The more capable AI becomes, the easier it is to imagine that human judgment becomes less important. Yet within L&D, increasingly capable technology may produce exactly the opposite effect.
When generating a scenario required hours of work, simply producing a good scenario represented meaningful value. When thousands of scenarios can be generated almost instantly, the valuable question becomes which scenario an employee actually needs.
When assessment questions were expensive to develop, creating a reliable assessment represented meaningful work. When AI can produce hundreds of questions in seconds, the important question becomes what evidence would genuinely demonstrate capability.
When content was scarce, creating content solved a problem.
When content becomes abundant, choosing what matters becomes the problem.
That choice requires judgment.
Learning professionals must determine when an employee needs information and when they need practice. They must decide when AI should provide an answer and when the learner should struggle with the problem first. They must distinguish between a knowledge gap and a motivation problem, between a skill problem and a broken process, and between a development need and an organizational environment that prevents the desired behavior.
AI can inform those decisions.
It should not eliminate the responsibility to make them.
The Future Instructional Designer Looks More Like a Performance Consultant
As June thought about the future of her own team, she realized that AI wasn't eliminating instructional design. It was changing the center of gravity of the profession.
Production would remain important, particularly for complex experiences requiring sophisticated design. But the capabilities surrounding production would become increasingly valuable.
Tomorrow's instructional designers may need to be equally comfortable discussing business performance and learning science. They will need to conduct discovery with leaders, analyze workflows, understand performance data, identify critical behaviors, design meaningful practice, structure coaching, and determine how capability should be measured.
Their toolkit becomes broader:
This is a more strategic role than simply producing learning content. It positions L&D closer to the performance challenges business leaders actually care about.
Dave reaches a similar conclusion in his memo: instructional design may become more important, not less, because as production becomes easier, "the thinking behind it becomes even more important." The Work We Are Really In
AI Should Expand L&D's Capacity for Curiosity
June also began thinking differently about the productivity gains AI was creating.
The obvious opportunity was to use saved time to produce more. If AI reduced development time by 30 percent, the organization could theoretically increase output by 30 percent. That logic was straightforward, measurable, and tempting.
But it wasn't necessarily the best use of the capacity AI created.
What if the time saved during production was reinvested in understanding the business?
Learning teams could spend more time observing employees, interviewing managers, analyzing performance, talking with customers, studying workflows, testing interventions, and evaluating whether behavior actually changed. Instead of converting every efficiency gain into additional output, L&D could reinvest some of that capacity into deeper understanding.
Dave raises a similar concern in his memo about AI-driven productivity. Individual tasks may take less time, but organizations often fill the resulting capacity with more meetings, messages, research, and work. His argument is that the real opportunity may be deciding more intentionally where attention and craftsmanship matter most. The Work We Are Really In
For L&D, that could mean using AI to spend less time manufacturing content and more time being curious about people.
That may ultimately be one of the most valuable productivity gains AI creates.
June's Definition of Great Learning Began to Change
By this point in her journey, June had stopped asking whether AI could create good learning. The answer was clearly yes, and the technology would only continue improving.
She had started asking a more difficult question.
What makes learning meaningful when anyone can create it?
The answer increasingly came back to context.
Meaningful learning reflects the situations employees actually encounter. It asks people to practice decisions they genuinely need to make. Feedback is connected to behaviors the organization values. Coaching reinforces those same behaviors. Measurement determines whether people have become more capable rather than simply whether they completed an experience.
None of those qualities depend primarily on the polish of the content.
They depend on understanding the person, the work, and the business well enough to create something that belongs there.
Dave uses the analogy of a text message and a handwritten card. Both can contain exactly the same words, but their meaning differs because of the context, intention, relationship, and experience surrounding them. The Work We Are Really In
June began to believe the same principle applied to enterprise learning. In a world where AI can create nearly unlimited content, what distinguishes meaningful development isn't simply what the content says.
It's how deeply the experience understands the person it is trying to develop.
Key Takeaways
Artificial intelligence is making high-quality learning content dramatically easier to create. Courses, scenarios, assessments, feedback, job aids, and reinforcement materials that once required significant time and specialized expertise can increasingly be produced quickly and at scale. That represents an enormous opportunity for enterprise L&D, but it also changes where learning organizations create the greatest value.
As content becomes abundant, context becomes increasingly important. Understanding the business problem, identifying the behavior that needs to change, diagnosing why that behavior isn't occurring, and determining what someone needs to experience to become more capable cannot be reduced to content production. Those decisions require business understanding, curiosity, empathy, performance expertise, and human judgment.
The most effective L&D organizations will therefore use AI aggressively without allowing faster production to become the objective. They will use technology to create capacity for deeper diagnosis, better practice, more relevant experiences, stronger coaching, and meaningful measurement. Instead of competing with AI on how quickly they can create content, learning professionals can focus on the work that makes that content matter.
As production becomes easier, understanding what should be created, for whom, in what context, and toward what performance outcome becomes more valuable. That shift doesn't diminish the importance of L&D. It moves L&D closer to the business problems it was always meant to solve.
Coming Next
In Part 4, June brings the AI Skills Paradox together and confronts the most fundamental question of the series: What business is L&D actually in? We'll explore The Work We're Really In and why the future of enterprise learning isn't defined by courses, content, software, or even AI. It's defined by an organization's ability to continuously develop human capability.
