The conversation around artificial intelligence (AI) and jobs often focuses on whether companies are replacing people. Headlines tend to frame the issue as a choice between automation and employment, with organizations either reducing headcount or maintaining it. While that question matters, it overlooks a more significant shift underneath the surface.

Many organizations are not eliminating entry-level positions altogether. Instead, they are changing the work those employees do. As AI systems become capable of handling research, drafting, analysis, summarization and other foundational tasks, the nature of early-career work is starting to evolve.

The question organizations need to ask is not whether AI can perform these tasks. The more important question is what happens when employees no longer learn from doing them.

Entry-Level Work Was Never Just About Getting Work Done

Companies often think of entry-level work primarily in terms of productivity. Junior employees conduct research, prepare presentations, draft documents, gather information and support more experienced colleagues. But those activities have always served a second purpose because they are the mechanism through which expertise develops.

People rarely build judgment through formal training alone. They build it through repetition, encountering problems, making mistakes, receiving feedback and gradually learning to recognize patterns. Over time, they develop a better understanding of what matters, what does not and how experienced professionals approach decisions. The work itself becomes the training ground.

That is why foundational work matters. Research, drafting and analysis are not merely tasks that need to be completed. They are how employees learn to evaluate information, organize their thinking, navigate ambiguity and develop professional judgment. These activities may not look like leadership development, but they are often where professional instincts begin to form. The tasks themselves may not always be exciting, but they provide something that is difficult to replicate elsewhere: experience.

Those experiences build judgment, pattern recognition and professional instincts that compound over time, preparing employees for increasingly complex responsibilities.

AI Is Beginning to Absorb That Experience-Building Work

According to our research at D2L, 56% of HR leaders report that generative AI is reducing the amount of foundational work delegated to junior employees, while nearly one-third say they are shifting toward fewer entry-level workers and more experienced employees supported by AI.

Those decisions are understandable. Organizations face constant pressure to improve efficiency, and AI can often complete routine work faster than a new employee. If a senior employee can use AI to perform tasks that previously required several junior contributors, the short-term business case is easy to understand.

The long-term implications are less clear because the same work that organizations are automating often served as the mechanism through which employees learned how the business operated. Research, drafting, analysis and problem-solving were never simply tasks on a checklist. They were opportunities for employees to develop expertise, build judgement, and gain experience.

As those opportunities become less common, organizations may discover that they have made work more efficient without fully considering how expertise will continue to develop. That concern is reflected in this research, which suggests that organizations may be changing the nature of entry-level work faster than they are changing their approach to employee development.

Expertise Still Requires Practice

One of the most important questions raised by AI adoption is whether reviewing work creates the same learning experience as producing it. Many organizations appear to assume that it does. If AI generates the first draft, employees can review the output, make edits and move on to higher-value work.

In some situations, that may be true. However, expertise has always depended on active participation. People develop judgment by working through uncertainty, defending conclusions and discovering where their reasoning falls short.

As an example, consider a new learning and development specialist tasked with identifying why a training program is receiving low participation rates. If AI analyzes the survey data, summarizes feedback and recommends solutions, the specialist’s role may be limited to reviewing the recommendations. But if the specialist has to interview stakeholders, analyze participation trends, identify root causes and develop their own recommendations, they learn how to diagnose organizational challenges, navigate competing perspectives and connect learning solutions to business needs. The most valuable development comes from wrestling with the ambiguity of the problem and determining what action to take — not simply evaluating an AI-generated answer.

An employee who spends years generating analysis develops different instincts than one who spends years reviewing generated analysis. Those differences may not be obvious immediately. They often become visible later when individuals are asked to solve unfamiliar problems, make decisions without guidance or lead others through uncertainty.

D2L’s research again found that employers are already reporting growing challenges around problem-solving, communication and interpersonal skills among recent hires. While AI is certainly not the only factor driving those trends, organizations should consider whether reducing opportunities for hands-on learning could make those gaps more difficult to address.

Organizations May Need to Become Much More Intentional About Development

For decades, most companies did not need a formal strategy for developing expertise because learning happened alongside execution. Employees started with smaller responsibilities, gained experience through the work itself and gradually took on more complex challenges. The system was imperfect, but it generally produced future managers, specialists and leaders.

AI changes some of those assumptions. If foundational work increasingly shifts to automation, organizations may need to think much more deliberately about how expertise develops.

This creates an opportunity to rethink the way we have designed traditional programs such as mentorships, apprenticeships, rotational experiences, shadowing opportunities and structured development programs by building intentional moments for “AI-free” reflection, synthesis and decision making. The goal is not to preserve manual work for its own sake. The goal is to ensure that employees still have opportunities to build the judgment and experience that organizations will need in the future.

That challenge becomes more urgent when paired with another finding from the study: 74% of organizations report having no active plan to replace learning opportunities that AI may remove. Many companies appear to recognize that work is changing. Far fewer appear to have a strategy for how development should change alongside it.

The Real Question Is Not About Jobs

Discussions about AI and employment often focus on how many positions will exist in the future. That is an important question, but it may not be the most important one.

Organizations have always needed a way to transform inexperienced employees into experienced ones. Every manager, executive, specialist and leader began somewhere. They developed expertise because they spent years doing the work, learning from it and building judgment through experience.

As AI takes over more of that work, companies will need to determine what replaces those learning opportunities. The challenge is not simply preserving entry-level jobs. It is preserving the pathway that turns entry-level employees into experts.