- AI is eliminating entry-level roles faster than organizations realize, cutting off the foundational work experiences where early-career professionals build real skills and judgment.
- Junior roles were always about more than output. They were structured learning environments where repetition, mistakes, and feedback turned knowledge into genuine expertise.
- A looming senior talent shortage, driven by mass boomer retirements and weak succession planning, will compound the skills gap if organizations fail to act soon.
- Leaders must identify which tasks are also learning experiences, and invest in deliberate knowledge transfer before the expertise gap becomes impossible to close.
By Derek Bruce, Chief Learning and Knowledge Officer at Easygenerator
I had a conversation recently with two Senior HR professionals whom I'd never met before. We were at a conference in Abu Dhabi and somehow ended up talking about AI, families, and the future of work.
One of them had a son with a Finance degree, struggling to get into the Finance industry, as he couldn't get a foot in the door anywhere. The other worked in the tech industry and said something that has become a common observation recently in my network. He said the entry-level roles aren't disappearing because companies don't need the work done. They're disappearing because AI is doing it.
That observation is starting to get some serious academic backing. A 2025 Stanford University study tracked millions of U.S. workers using ADP payroll data and found a 13% relative decline in employment for workers aged 22 to 25 in the most AI-exposed occupations since late 2022. Workers in the same roles who were older were unaffected. Separately, Revelio Labs reported that U.S. entry-level job postings had fallen roughly 35% since January 2023. That's over 100,000 fewer new monthly job postings.
What junior roles give all of us
Most organizations are looking at this the wrong way. The question they're asking is whether AI can do the work that junior employees used to do. In most cases, the answer is yes. But that's the wrong question.
Junior work wasn't just production. It was a learning environment and experience. A place where we build judgment, pattern recognition, and professional instinct through repetition, mistakes, and feedback. A junior doctor who reads scans manually, forms a view, and gets challenged on it builds something over time that AI assistance can never produce: the ability to catch what isn't obvious.
Cognitive science has been making this argument for decades. Sustained practice with feedback is what gradually reorganizes knowledge into something more than knowledge. That process takes time, and it requires actually doing the thing.
Recent studies are now showing what happens when you shortcut it. An Anthropic study in 2026 tracked junior engineers learning a new codebase. The AI-assisted group finished tasks faster but scored 50% on a follow-up assessment compared to 67% for the group that worked without AI. The largest gap was in debugging, which is exactly the skill that requires understanding why something goes wrong, not just producing something that works. MIT Media Lab researchers coined the term "cognitive debt" after EEG monitoring showed that habitual AI users had measurably lower neural connectivity over time.
The compounding problem
This would be a serious issue on its own. But it's running into something else at the same time.
Ten thousand baby boomers in the United States turn 65 every single day, according to Pew Research. And according to a Robert Walters survey of 250 UK organizations published in 2026, only 14% of companies have a succession plan in place. 70% of leaders said they already had a shortage of senior talent. Cornerstone's Skills Economy Report 2026 puts numbers on why: organizations could already automate 30% of entry-level work hours, the same tasks that once helped early-career professionals build practical experience.
So senior knowledge is walking out the door faster than it can be transferred, and the cohort that was supposed to absorb it and develop over the next decade is getting less exposure and less access to the foundational work that builds genuine expertise. If you run that forward ten or fifteen years, organizations will end up short of the kind of people who know when AI is wrong or can identify nuances based on having built a foundation of skills.
What organizations should actually do about it
The answer isn't to resist AI or pretend the efficiency gains aren't real. They are, and organizations that ignore them will fall behind. But efficiency and expertise development are different problems, and they need different answers.
The question every leader, and this isn’t a HR or L&D concern, should be asking is not which tasks AI can absorb, but which tasks are also learning experiences. Those two things often overlap, and when they do, automating the task also automates away the development.
The other side of this is making the transfer of expertise deliberate rather than incidental. For most of the past century, expertise moved through organizations the way it always had: through proximity, mentorship, and years of working alongside people who knew more than you did. Organizations that invest in structured ways to capture what their most experienced people know, and to create genuine practice opportunities for people earlier in their careers, will have something others don't in ten years. They'll have people with real judgment, not just access to good tools.
AI is one of the most powerful things available to organizations today. And we do have a need to also ensure we build human-centric skills alongside skills in how to use AI. But it produces outputs, not expertise. And if we keep treating those as the same thing, the consequences won't show up in a quarterly report. They'll show up a decade from now, when the people who were supposed to know how to do the hard things never got the chance to learn.
A final thought for you… I saw a reel on Instagram which has the caption ‘Your future Doctor is currently using ChatGPT to pass their exams. So, start exercising and eating healthy.’
While this was a humorous sentence, consider how close we may be to this future.
About Derek Bruce
Derek Bruce is Chief Learning and Knowledge Officer at Easygenerator, where the focus has been on exactly this problem: how organisations can capture the expertise that lives inside their most experienced people and make it accessible to everyone else. Before joining Easygenerator, Derek worked with L&D teams at ABN AMRO, dsm-firmenich, Signify, and Tesco, where he saw firsthand how much institutional knowledge never gets passed on. He writes and speaks on how organizations can build genuine expertise at scale in a world where AI is changing what work looks like.
About Easygenerator
Easygenerator is an AI-powered e-learning suite that helps organizations create company-tailored training at scale. Built for internal experts and L&D teams alike, its author-first AI captures the depth of organizational knowledge and transforms insights, documents, and ideas into didactically-sound courses. Teams can create professional training videos with AI avatars and multilingual voiceovers, and help employees practice real workplace conversations to build skills with confidence.
