Reading this report made me pause: when did ‘entry‑level’ start requiring ten years of experience?
I just finished reading PwC’s 2026 Global AI Jobs Barometer, and one finding has stayed with me: AI isn’t eliminating roles, it’s rebalancing them.
That sounds like a positive at first, but as I read further it got more complicated. Skills in the most AI-exposed jobs are increasingly human-intensive capabilities (empathy, creativity, judgment, presence) increasing the value of human capabilities. However, many of the jobs where AI is being implemented are entry and junior-level roles and the report shows that many of the skills being pushed into the remaining entry‑level roles are the very skills we traditionally develop over years of experience such as communication, empathy, innovation, judgment, presence, stakeholder management, and the ability to navigate ambiguity with confidence.
In other words: We’re asking early‑career employees to operate like seasoned professionals without giving them the time, practice, and guidance seasoned professionals had.
And that’s the leadership challenge I fear many are not seeing or addressing.
If organizations don’t redesign early‑career pathways now, through intentional mentoring, structured partnerships, and modern apprenticeship‑style development, they will quietly erode their future leadership pipeline. Not because people aren’t talented, but because we’ve removed the scaffolding of people, processes, and practice that used to help them grow.
I see this playing out in real teams with real work now:
New hires expected to “own” decisions without ever seeing how good decisions are made.
Analysts asked to challenge assumptions without having witnessed healthy debate.
Early‑career staff told to “be strategic” without ever being shown what strategy looks like in practice.
AI didn’t cause this gap but those responsible for future growth and business continuity own what has been exposed.
What you can do today (small actions, big impact)
These are simple, human moves you can make in the flow of work to broaden and build the capabilities of the people close to you by intentionally:
Narrate your thinking. When you make a decision, take 60 seconds to explain how you got there. This is how judgment is taught.
Invite a junior colleague into a conversation one level above their role. Let them observe how strategy, disagreement, or prioritization actually happens. Mentor with meaning.
Assign “micro‑ownership.” Give early‑career teammates a small but meaningful decision to make and coach them through the tradeoffs rather than correcting the outcome. Key to coaching is to ask questions, get them to GROW (tips in link).
Pair AI fluency with human fluency. If someone is learning to use AI for analysis, pair it with a conversation about how to communicate insights, challenge assumptions, or frame recommendations. Partner people with different experience levels together.
These small practices compound. They rebuild the apprenticeship learning layer that AI‑accelerated work has quietly stripped away.
Why this matters
If we don’t redesign early‑career development now, we’ll feel the consequences in three to five years when retirements accelerate, mid‑career leaders thin out, and organizations realize they’ve unintentionally hollowed out the beginning and middle segments of their talent pipeline.
But the good news is this: The skills AI elevates are deeply human and we learn best from each other. We can fix this. We just have to be intentional.
I’d love to hear from you
How are you seeing early‑career roles shift? Are you seeing fewer entry level people and how are you preparing for the future? Share what’s working, or not, in your organization so we keep learning together.
And if you’re exploring how to redesign early‑career pathways, build human‑centered AI adoption strategies, and use coaching to develop leaders at all levels, I’m happy to help.
#FutureOfWork #LeadershipDevelopment #AIAndWork #EarlyCareerTalent #LeadershipStrategy
The four practices are the right response to a real gap, but they treat judgement as something a manager can narrate into existence a few months early. It isn't. Judgement is what accumulates through years of experience turning into knowledge, and knowledge turning into wisdom — the ability to spot something because you've seen its shape before, or read the context a case doesn't state out loud. A machine does exactly what it's told. Hand a junior a confident answer and take away the years that would let them tell when it's wrong, and there's a specific accountability gap sitting underneath: if nobody in the room has the experience to challenge the output, who is answerable when it turns out to be wrong?
None of the four practices ask where the tools actually go first. Most rollouts start at the bottom, with the people who have the least standing to push back on what a model produces, and the practices on offer are a set of manager habits designed to compensate for that after the fact. Reverse the order instead: put the tools at the top of the pyramid first, where the experience to challenge a confident wrong answer already exists, and let use work down from there.
That sequencing is also what makes self-organising structures and servant leadership survivable rather than aspirational. Both depend on people lower down having real authority to decide and challenge, not just cover for a machine's answer. An executive team that's already using the tool, and already knows where to override it, can hand that authority down deliberately. A junior team handed the tool first has no equivalent authority to hand back up — which is usually where the aspiration to self-organise quietly stalls.