Replacing graduates with AI will hurt businesses in the long run
Replacing graduate roles with AI does not build a smarter business, it builds a shallow one, writes Paul Armstrong
Youth used to be the bet businesses made when they were serious about building future resilience. Early-career hires, valued for their adaptability, capacity for growth and affordability formed the bedrock of institutional knowledge, creating the conditions for future leadership and depth. Now that engine is being systematically dismantled, not due to a lack of work, but because businesses are prioritising automation, AI and offshoring over long-term capability.
The problem is huge, has been coming for a while and just got a shot in the arm. Graduate job listings in the UK have dropped by over 26 per cent in the past year, according to Adzuna data. In accountancy alone, the fall has been even steeper, with roles dropping by 44 per cent. The Big Four are at the sharp end of this shift. KPMG slashed graduate hiring from 1,399 to 942. Deloitte reduced intake by 18 per cent. EY and PwC followed suit with 11 and six per cent cuts respectively. These numbers are not temporary contractions, they signal a fundamental change in how firms perceive junior labour, and AI is only just getting into the bloodstream.
Youth unemployment is on the rise
Rather than reimagine entry-level work, most firms are removing it altogether. Tasks that used to introduce young employees to business operations like compiling research, preparing documents, writing reports and processing data are now being absorbed by AI systems. Large language models handle most of the documentation. Scheduling, transcription and initial analysis are increasingly automated. Anything that doesn’t require direct client interaction is either routed through an AI layer or sent to lower-cost teams offshore.
Esther O’Callaghan OBE, founder of Hundo, and Sam Hyams, CEO of Springpod, the virtual work experience provider, are speaking at the House of Commons on 21 July about what this means for the UK, and the world. Their message is stark: “For the past decade there has been an increasing decline in graduate programmes – the natural off-ramp for universities into employment. With AI now entirely disrupting workforce skills requirements and the number of entry level roles reducing in tandem has created a perfect storm for early job seekers. We now have over 1.5bn young people worldwide facing a level of job loss at a pace and scale we have never seen in history. We’re beyond discussion… rapid action is needed.“
Graduate salaries are rising modestly, now averaging £28,486, up 4.5 per cent over the last year. That is not a sign of opportunity; it reflects scarcity and a narrowing funnel. Fewer roles, higher expectations and minimal training infrastructure mean businesses are attempting to hire fully-formed junior staff who require no onboarding and deliver immediate value. Such an approach might reduce short-term overhead, but it’s going to remove the runway for talent development.
Organisations need to stop treating entry-level roles as administrative placeholders. These positions are formative experiences, not just deliverables
Youth unemployment continues to grow, currently at 12.4 per cent, nearly triple the national average. Many graduates, particularly outside London and the Southeast, are falling out of the pipeline before they ever get started. Faced with fewer opportunities and growing debt burdens, more young people are turning to freelance work, short-term gigs or leaving their field entirely. You’re not going to solve that with more career advice or softer skills training, if the entry points are vanishing.
Firms that once relied on deep internal training grounds are now pushing young people toward an unstructured labour market. What remains are narrow paths to progression, often dependent on elite credentials, personal networks or access to the few firms still investing in talent development. Businesses that cut back on junior hiring often report difficulty with succession planning, middle-management capacity and culture continuity. The causes are internal, and so are the consequences.
The jobs AI can’t do
Some of the same firms that are scaling down junior roles are also investing heavily in AI audit and governance functions. Deloitte, EY and PwC are all developing AI assurance offerings to help clients validate and trust their machine learning models. These service lines are growing, yet they still require humans with technical, legal and ethical judgement to deliver them. Without a strong internal pipeline, that talent has to be bought, and often at a premium. Businesses are paying more to compensate for the fact that they no longer train their own.
Removing entry-level roles without replacing the development pathway creates problems down the line, namely with organisational fragility. AI can perform many tasks, but it cannot replace the situational awareness, ethical reasoning or interpersonal intelligence that leaders rely on. Those traits are not innate, they are learned over time, through exposure to imperfect situations, difficult clients and messy collaboration. A team built without that experience layer may perform well under routine conditions but will struggle under pressure, change or crisis.
Forward-looking organisations are beginning to question this current approach. Capgemini has launched an AI fluency programme specifically aimed at early-career hires, focusing on interpretation, oversight and cross-functional thinking. IBM is testing skills-based hiring models to widen access and avoid over-reliance on legacy credentials. These moves reflect an understanding that while tasks can be automated, trust and context cannot. Are they perfect? No. Is it a start? Absolutely.
Most companies, however, remain in reactive mode. Automation is still being deployed as a blunt cost-saving tool rather than a strategic enhancement. Offshoring continues to replace local training. Procurement decisions still prioritise efficiency over resilience. The risk isn’t simply short-term instability, but the creation of an organisation that cannot learn or adapt without external intervention. Beyond this, we’re effectively saying that we’re okay building a generation where many, who are often already poorer than their parents were, could have even fewer prospects.
Hiring processes also need serious reform. Many graduates are being excluded from consideration by automated systems that filter based on keyword matches, previous employer prestige or academic pedigree. Talent that could thrive with support is being missed entirely. AI-powered HR systems are replicating the same inequities they were supposed to disrupt and, in doing so, are cutting off access to future potential.
The many benefits of junior talent
Organisations need to stop treating entry-level roles as administrative placeholders. These positions are formative experiences, not just deliverables. Without them, businesses risk losing important perspectives, cultural depth and bottom-up innovation. Simply replacing junior execution with AI does not build a smarter business, but it will build a shallow one.
Training for the AI era must be re-engineered to reflect how knowledge is now distributed. Early-career hires should be embedded in hybrid roles that combine human mentoring with machine support. Rather than removing the work, businesses need to redesign the learning. Judgement gets taught, and isn’t assumed. Complexity should be introduced gradually, not avoided. The longer businesses avoid confronting the erosion of the graduate pipeline, the more they risk creating a two-tier workforce: one layer of decision-makers who remember how things used to work, and another with no path to learn how to think critically at all. Bridging that divide will become harder each year that hiring and development are treated as optional.
Cost control remains important in the trump-certain world we live in currently. So does innovation. Neither will be possible without a workforce that knows how to navigate ambiguity, build relationships and interpret nuance. Those skills don’t arrive fully formed or through a Claude or Gemini dialogue box. Skills like that emerge through practice, feedback and time.
No AI model is going to fix that for you. Talk disruption all you want, but without junior talent, you’re just rearranging risk. Rebuilding the early-career system will take deliberate effort and long-term thinking. For companies serious about surviving disruption rather than simply talking about it, or worse, leaning on bad quick fixes, investing in young talent is no longer optional. Ignore this, and you’re optimising for future collapse of more than just your own business.
Paul Armstrong is founder of TBD Group and author of Disruptive Technologies