As artificial intelligence transforms entry-level roles, organizations must protect something less visible than jobs: the experiences through which people learn to lead.
For generations, future leaders learned partly by doing work that was not especially glamorous.
They conducted research, prepared initial analyses, drafted proposals, checked data, attended meetings and watched more experienced colleagues navigate difficult decisions. The output mattered, but so did the exposure. Through repetition, feedback and observation, young professionals gradually learned how their organizations operated and how good judgment was exercised.
Artificial intelligence is beginning to perform much of this work faster and more efficiently.
That may be good for productivity. But it raises a question few organizations were asking until recently:
If AI performs the work through which junior employees traditionally learned, where will the next generation develop the judgment required to lead?
The Hidden Purpose of Entry-Level Work
Organizations tend to evaluate work according to its visible output. Did the analysis get completed? Was the presentation prepared? Was the client request answered accurately and on time?
But early-career work has always produced a second, less visible output: capability.
The junior analyst does not simply create a spreadsheet. She begins to understand which numbers matter and which anomalies require investigation. The young consultant does not merely build a presentation. He learns how an argument is structured, how evidence is selected and how senior clients respond when recommendations challenge their assumptions.
The first draft, the basic analysis and the background research are not only tasks. They are part of an informal apprenticeship.
This is why the discussion about AI and entry-level work cannot be reduced to the number of jobs that may disappear.
In September 2025, Amy Edmondson and Tomas Chamorro-Premuzic argued in Harvard Business Review that eliminating entry-level roles is short-sighted because these positions help organizations develop future leaders, generate innovation and renew their cultures.
More recently, McKinsey Quarterly described research, documentation, data preparation, basic coding and preliminary analysis as precisely the activities through which young professionals have traditionally built instincts, developed judgment and prepared for greater responsibility.
When these activities migrate to AI, the scaffolding supporting early-career development can begin to disappear.
The Productivity–Learning Confusion
One of the risks of AI is that it can make a person appear more capable before that person has developed the underlying capability.
A junior employee may produce an impressive analysis with AI assistance. But can they identify when the analysis is incomplete? Can they explain why one conclusion is more credible than another? Can they recognize when a recommendation is technically correct but inappropriate for the organizational context?
And what happens when the tool is unavailable, confidently wrong or confronted with a situation not represented in its data?
McKinsey cites evidence suggesting that when people use generative AI to complete tasks they cannot perform independently, the apparent capability may disappear once access to the technology is removed. Producing a stronger output is not necessarily the same as developing stronger judgment.
This distinction will become increasingly important.
Organizations may believe they are accelerating development because junior employees are producing higher-quality work earlier. In reality, they may be accelerating output without accelerating learning.
AI can provide an answer. It does not automatically teach the user how that answer was reached, when it should be questioned or how the underlying reasoning can be transferred to a different situation.
The Seniorisation Paradox
At the same time that AI is removing some foundational tasks, entry-level roles are becoming more demanding.
The Financial Times has described this as the “senior-isation” of junior work. Employers increasingly expect new professionals to demonstrate judgment, initiative, communication, decision-making and interpersonal awareness—capabilities previously developed over a longer period.
This creates a development paradox.
Organizations expect judgment earlier while potentially eliminating the experiences through which judgment was traditionally built.
A recent World Economic Forum report found that more than one in three young workers globally are employed in occupations with medium-to-high exposure to AI-driven task change. The report identifies job design and talent pipelines as two of the critical issues organizations must now address.
The challenge is therefore not simply to preserve yesterday’s entry-level jobs. Some of those jobs were overly repetitive, and AI offers a genuine opportunity to improve them.
The challenge is to redesign early-career work so that employees can contribute at a higher level without losing the experiences required to develop depth, context and judgment.
AI Can Accelerate Learning, but Only by Design
The relationship between AI and human development does not have to be negative.
AI could give early-career employees access to expertise that previously took years to acquire. It can provide immediate feedback, create simulations, expose employees to alternative approaches and allow them to practise difficult decisions in lower-risk environments.
But learning will not result from access alone.
The relevant distinction is between using AI as a substitute for thinking and using it as an instrument for improving thinking.
In the first model, the employee asks the tool for an answer, accepts the output and moves to the next task.
In the second, the employee forms an initial view, compares it with the AI-generated response, investigates the differences, seeks feedback and decides what to retain or revise.
The first model creates efficiency. The second can create learning.
McKinsey describes an “answer-key” approach in which employees first attempt a task independently, compare their work with an AI-generated result and then discuss the differences with an experienced manager. The AI provides speed and breadth, but human coaching supplies context and judgment.
This approach is consistent with an argument advanced by Molly Kinder at the Brookings Institution. She proposes that professional organizations learn from medical residencies, where less experienced practitioners make real decisions under supervision and gradually receive greater autonomy. In this model, learning is not an additional benefit attached to the job. Learning is deliberately designed into the job.
I believe this is the right direction. The choice is not between protecting old-fashioned junior work and embracing AI. It is between allowing development to erode accidentally and redesigning it intentionally.
Where Learning Agility Enters the Equation
At Burke Assessments, we define Learning Agility as a behavioral capability: how individuals learn from experience and apply that learning when facing new or unfamiliar situations.
This makes Learning Agility particularly relevant to AI-enabled work.
The critical human contribution will not be producing information faster than a machine. It will be knowing how to engage with information when the answer is incomplete, ambiguous or potentially wrong.
Several Learning Agility behaviors become especially important:
Information gathering
A learning-agile employee does not treat an AI-generated answer as the end of the inquiry. They examine additional evidence, explore the context and determine what information may be missing.
Flexibility
AI often generates a plausible response, but plausibility is not certainty. Flexible thinkers consider competing interpretations and remain willing to revise their conclusions.
Experimenting
Instead of asking AI for one definitive solution, employees can test different assumptions, prompts and approaches, observing how the result changes and what that reveals.
Feedback seeking
AI feedback cannot replace insight from people who understand the organization, its stakeholders and the consequences of a decision. Learning-agile employees actively seek that perspective.
Interpersonal risk-taking
Sometimes learning requires challenging an answer that colleagues—or senior leaders—have already accepted. It also requires admitting uncertainty or acknowledging that an initial conclusion was wrong.
Reflecting
Without reflection, AI can accelerate the movement from one task to the next without creating durable learning. Employees need to pause and ask what they understood differently as a result of the experience and how that lesson applies elsewhere.
These are not abstract qualities. They are observable behaviors that can be measured, discussed, practised and developed.
AI may increase access to knowledge, but Learning Agility determines how actively and intelligently a person engages with that knowledge.
Five Principles for AI-Enabled Leadership Development
Organizations can begin protecting their future leadership pipelines by incorporating several principles into the design of work.
- Let the human attempt come first
When the developmental value of a task is important, employees should form an initial view before consulting AI. The objective is not to compete with the technology but to make their current reasoning visible.
Without an independent attempt, it becomes difficult to distinguish what the employee understands from what the tool has supplied.
- Require comparison, not acceptance
Employees should examine where their reasoning and the AI output differ. Which approach identified a risk the other missed? Which assumptions produced the difference? What additional evidence would resolve the uncertainty?
The gap between the two answers can become the learning opportunity.
- Make people explain the reasoning
A polished answer can conceal shallow understanding. Employees should be able to explain why they trust an output, what limitations it contains and under which conditions they would make a different decision.
The ability to explain reasoning is a stronger sign of development than the ability to produce an impressive document.
- Redefine the manager as a coach of judgment
When AI handles more task execution, managers may need to spend less time correcting mechanics and more time explaining context.
Why did an experienced leader reject the apparently logical option? What organizational history influenced the decision? Which stakeholder concern was not visible in the data? When should the employee escalate rather than act?
This knowledge is rarely found in a manual. It is transmitted through conversation, observation and feedback.
- Measure capability, not output alone
If organizations measure only speed and quality, AI-assisted employees may appear to develop faster than they actually do.
Development measures should also examine whether employees can apply learning without the tool, transfer it to unfamiliar situations, recognize exceptions and revise their thinking when circumstances change.
A strong output demonstrates that the work was completed. Adaptive performance demonstrates that learning occurred.
Preserving the Human Pathway to Leadership
The future leadership pipeline will not be protected simply by continuing to hire the same number of graduates. Nor will it be protected by providing every new employee with an AI assistant.
Organizations need to identify which experiences build judgment and ensure that those experiences remain present—even when the underlying tasks change.
Some development may happen through simulations. Some may involve employees working independently before comparing their reasoning with AI. Some will require closer apprenticeship, cross-functional assignments, supervised decisions and deliberate reflection.
What matters is that learning no longer be treated as an accidental by-product of junior work.
This is also an opportunity to improve leadership development. Traditional apprenticeship was not equally available to everyone. Access often depended on proximity, informal networks and whether a senior colleague chose to invest time in a particular employee.
A more deliberate model could make high-quality developmental experiences visible, structured and accessible to more people.
AI Will Not Build the Leadership Pipeline by Itself
Artificial intelligence can prepare the analysis, summarize the meeting, draft the recommendation and identify patterns that a human being may overlook.
But it cannot, by itself, ensure that a person develops the judgment to know when the analysis is misleading, when the summary misses what mattered, or when the recommended action will fail in the human reality of an organization.
That development still requires experience.
It requires the willingness to attempt something unfamiliar, seek feedback, examine mistakes, consider different perspectives and apply the resulting learning to a new situation.
In other words, it requires Learning Agility.
The defining talent question of the AI era may therefore be larger than which jobs technology will replace.
It may be this:
How will we use AI to improve performance today without removing the experiences that create the leaders we will need tomorrow?
AI can do more of the work.
Organizations must ensure that people continue to do the learning.
