Why the problem is no longer skills, but system design
The debate about early-career talent and AI is becoming increasingly polarized. Graduates are portrayed as unprepared or unrealistic, while organisations fear that AI will simultaneously replace entry-level work and demand skills that barely exist.
What troubles me is not the pace of change itself, but how consistently we misinterpret what is actually happening.
Over the past weeks, two very different sources landed on my desk. A Future Talent Council webinar where employers listened directly to students across the globe, and a sharp analysis by Toby Culshaw in Recruiter on the “ talent intelligence horizon ” for 2026. Different angles, same conclusion: we are designing talent systems for assumptions, not for reality .
AI dominates the narrative, not the labour market
One of the most grounding moments during the Future Talent Council session came from Bill Boorman , who reminded us how distorted the AI conversation has become.
Based on live labour-market data, only around 3 percent of entry-level roles currently require explicit AI skills . Yet ask most leaders and they will confidently estimate ten or twenty times that number.
This insight aligns almost perfectly with the data cited in Recruiter . Lightcast figures show that while HR functions are experiencing rapid growth in AI-related demand, the absolute numbers remain low . Even for talent acquisition and HR leadership roles, AI skills appear in only a small fraction of postings.
The conclusion is uncomfortable but clear: AI is strategically important, but operationally still peripheral for most early careers .
We are preparing young professionals for a future that has not yet arrived, while neglecting the foundations they actually need today.
Efficiency pressure without learning infrastructure
Toby Culshaw describes 2026 as a year defined by paradox. Organisations demand efficiency while downsizing teams. Venture capital pours billions into generative AI, yet 69 percent of TA leaders report their GenAI stack is not delivering expected productivity gains . MIT research suggests that up to 95 percent of enterprise AI initiatives fail to create measurable business value .
This matters deeply for early careers.
Historically, entry-level roles served a hidden but critical function: they were learning systems. Repetitive tasks, shadowing, coordination work and exposure to “how things really work” allowed people to develop judgement, context and confidence.
Many of those tasks are now automated or removed in the name of efficiency.
As Bill Boorman put it during the FTC discussion: “How are people going to learn work if the tasks that used to teach them are disappearing?”
This is the question too few organisations are answering.
Students are not unprepared, they are under-informed
When students were asked how ready they feel to enter the workforce, the results showed cautious confidence, not panic. The real gaps appeared elsewhere: communication norms, workplace culture, financial security, psychological safety and clarity about expectations.
Claudia Tattanelli summarised it sharply: the issue is not resilience, but transparency. Students are asking whether they are allowed to ask questions, make mistakes and grow without fear.
What struck me is how closely this mirrors Culshaw’s description of the broader workforce. Suppressed attrition, frozen promotions, rising living costs and platform fatigue are creating what he calls “geological pressure” inside organisations.
Early-career anxiety is not generational. It is structural.
Talent intelligence is escaping its silo
One of the strongest insights from the Recruiter piece is that flat labour markets are forcing a shift from “buy” to “build” strategies. When people are not leaving, organisations cannot hire their way out of capability gaps. They must understand, develop and redeploy what they already have.
This is where talent intelligence is finally maturing. Not as a recruitment tool, but as connective tissue across workforce planning, learning and performance.
The Future Talent Council conversation reinforces this. Students are open to hybrid and frontline roles, but lack visibility into how those paths actually lead somewhere. Learning does not fail because people resist it, but because pathways remain implicit and fragmented.
AI-ready people, unclear rules
Another paradox surfaced clearly. Students are widely using AI tools and feel comfortable experimenting with them. At the same time, organisations send mixed signals, encouraging AI use in theory while penalising it in applications and assessments.
This governance gap is not trivial. You cannot decentralise intelligence, as Culshaw argues we must, without clear data foundations, consistent taxonomies and explicit rules of play .
The Lightcast data makes this even more concerning. While non-IT generative AI roles have surged dramatically, AI ethics and governance skills appear in less than 1 percent of job postings .
We are decentralising power faster than we are building guardrails.
The real gap sits in the middle
What connects these sources is not fear of technology or generational conflict. It is misalignment.
Employers overestimate how clearly expectations are communicated. Students overestimate the stability of traditional career paths. And both navigate a BANI* world without shared reference points.
*We are no longer operating in a VUCA world that can be managed with better forecasts and faster decisions. We are moving into a BANI reality: brittle systems, anxious workforces, nonlinear change and growing incomprehensibility.
In such a context, early-career uncertainty is not a generational issue. It is a systemic one.
Belonging, learning and trust are no longer outcomes of good intentions. They are design challenges.
My takeaway
The most important shift ahead is not adopting more AI, but designing better systems of learning, exposure and progression . Early-career work used to teach people how organisations function. We are dismantling that system faster than we are replacing it.
Listening is no longer a soft exercise. It is a strategic one. And talent intelligence, if grounded in reality rather than hype, can become the mechanism that reconnects efficiency, development and human experience within organisations.
If this resonates with your organisation’s reality, I’m always open to continue the conversation. Whether it’s a sparring session, an internal discussion or an inspiration session on early careers, AI and talent intelligence, feel free to reach out.