At a recent alumni event at my business school with representatives across many sectors, I asked participants to pinpoint where they see the most immediate benefits of AI adoption in their organisations. I heard typical comments about AI pilots that work well in isolation but don’t yet scale, early experimentation with AI agents and the individual use of generative AI for menial and admin tasks.
But another point reverberated through the room and spilled over into the post-session cocktail: how the people I spoke to felt the need to show some return for the substantial investment in deploying AI, and that an easy way to do that was by reducing entry-level roles. Compressing existing career ladders is an understandable approach because entry-level positions sit at the fringes of the organisation; it is possible to replace them without having to touch more complex organisational processes and systems themselves.
That this is not just anecdotal evidence is reflected in a King’s College London study of millions of job postings and LinkedIn profiles from 2021 to 2025. It found that firms highly exposed to AI reduced employment, with the decline concentrated almost entirely in junior roles.
However, the real risk is not only that AI removes junior work. It is that AI removes the low-risk, repetitive, feedback-rich tasks through which people learn to become good at more complex work. In other words, without carefully designed early-career experiences, organisations may save on junior staff today while creating a shortage of experienced talent tomorrow.
 |
Get our essential weekday HR news and updates.
|
Rethinking career development in the AI era
The traditional approach to talent development is to buy talent externally, build talent internally, or borrow talent for more temporary needs. However, if critical, advanced skills become scarcer, the acquisition of senior talent is going to become considerably more expensive.
Using traditional ways to develop talent internally will be similarly difficult if AI erodes the very tasks that serve as a training ground. After all, how will junior talent be able to build necessary skills such as verifying AI-assisted work output in work domains if they themselves haven’t mastered those skills yet, especially if AI prevents them from being able to do so? And there is no compelling reason for why the most sought-after skills should be easily accessible through flexible work arrangements.
So how can organisations enable future talent to develop the right skills and capabilities that allow them to progress into more senior roles where AI becomes a necessary input to decision-making?
There are several ways how companies can reimagine career development more deliberately. First, companies should redesign junior roles around judgement rather than routine execution. Early-career employees may no longer need to produce first drafts of analyses, reports or presentations from scratch, but they still need to learn how to evaluate AI-generated output, spot weak assumptions, ask better questions and communicate recommendations. In this sense, “verification work” can become a new developmental stepping stone, provided it is treated as a learning-rich activity rather than mechanical checking.
Second, organisations could create AI-based apprenticeship systems. Junior employees need structured opportunities to work alongside experienced professionals and observe how they frame problems, challenge AI outputs, manage ambiguity and make decisions. Some deliberately inefficient learning work may also need to be preserved, even when AI could complete the task faster, because such work builds the foundations of future expertise.
Third, companies can make more deliberate use of rotations, internal project marketplaces, and cross-functional assignments to expose employees to different kinds of work, including both AI-heavy and human-intensive domains. This helps early-career talent understand not only what AI can do, but also where AI is a necessary input to judgement and decision-making.
Fourth, managers must be held accountable for talent development, not only for productivity gains. If AI allows teams to do more with fewer junior employees, managers may be tempted to coach less. Organisations should therefore track feedback quality, skill progression, internal mobility and readiness for future roles.
The rise of the career ecosystem
Finally, firms should build broader career ecosystems, both by expanding the talent pool from which they hire and by conceiving sequences of career development that transcend organizational boundaries. This might include partnerships with universities and bootcamps, hiring from start-ups and scale-ups, project-based external learning opportunities, alumni networks and boomerang pathways. Career development, then, will become less like climbing a ladder and more like building a portfolio of experiences within and outside the organisation.
Existing career development in organisations was built for a world in which junior talent learned by doing the work that senior employees no longer wanted to do. However, AI is eroding this bargain. The organisations that thrive will not be those that simply remove the bottom rungs of the ladder, but those that rebuild career development as an ecosystem of projects, rotations, mentors, external experiences, and deliberate human judgement.
Sebastian Reiche
Sebastian Reiche is a Professor of People Management at IESE Business School. His research focuses on how leaders, professionals and organisations cultivate connection and navigate distance in today’s workplace. He is author of the book Proximity: How to balance belonging and difference in today’s workplace (Bloomsbury, 2026).
