AI Takes Over Entry-Level Grunt Work, Reshaping Graduate Hiring

At London ad agency Catalyst, one junior employee now manages five client accounts — a workload that previously required several people. AI tools handle the research, draft the first pass of copy, and gather the data, according to founder Tobias Green.

"Then you need someone to come in and fix it up," Green told CNBC. He describes the arrangement as "a reverse Mechanical Turk": AI works behind the scenes while the human is the face presented to the client.

The shift is forcing Catalyst and other employers to rethink entry-level roles entirely. "If AI removes the grunt work, then the grunt work is no longer relevant," Green said. "They don't need to learn those skills … They need to learn how to manipulate and utilise AI."

The stakes extend well beyond advertising. AI is now adept at research, drafting, coding and administrative work — the tasks traditionally assigned to junior employees across industries — and the effects are showing up in hiring data.

Employers are already cutting junior roles

Open University's 2026 Business Barometer report, which surveyed 1,500 U.K. business leaders, found that 51% of respondents said AI was changing how they hire. Some 19% said they had reduced entry-level recruitment, and 42% of that group cited AI adoption as a reason.

Lucy Beaumont, global SVP of product at Surrey-based talent assessment company SHL, said some major graduate recruiters are hiring fewer people while they assess whether they need those roles at all.

"If we've had an organisation that's a pyramid shape, we're now seeing what we call a diamond, with more people at that mid-level," Beaumont told CNBC. She warns the approach is a short-term fix that can create a crisis for employers further down the line.

"Those entry-level roles, they're your training ground — that's where you get your training wheels, you get your badges, get your war wounds, make mistakes and learn how to operate in a corporate environment," Beaumont said. "If you take that away, it has serious implications for every rung of the ladder."

Junior roles are changing, not vanishing

The picture is not uniformly negative. In a survey of 500 U.K. and Irish mid-market leaders by technology consultancy Klarus, 45% said AI was helping junior employees work better and faster, and nearly a quarter said it was creating new roles and opportunities.

Software development shows the trend most clearly. Sheila Flavell, COO of the business and technology consultancy at London-based FDM Group, said tools including Claude Code and Codex mean developers increasingly oversee projects completed by both AI agents and people.

"The skillsets that will be incredibly important are understanding how to work with others," Flavell told CNBC. AI could push junior workers toward supervisory work much earlier in their careers.

Hiring for this model is its own challenge. Beaumont said some employers have stopped recruiting graduates into predefined roles and instead hire for broad capabilities, deciding where to deploy people afterward.

"It is hiring for potential," she said. "Have you got re-skilling potential? Could we re-skill you onto different things? It doesn't really matter where you've come from."

She pointed to one large global retail bank she works with that now recruits graduates in subjects like psychology and law, on the reasoning that technical skills can be taught while critical thinking is harder to build from scratch.

The judgement problem

Cheney Hamilton, director at research firm Bloor, where she runs research into workplace transformation, frames the core issue as whether juniors can supervise work they have never learned to do themselves. Grunt work, she argues, was never just work — it was the mechanism through which people accumulated context and judgement.

"AI raises the floor of what a junior can produce, but it doesn't give them the judgement to know when the output is wrong, and that judgement historically came from repetition and consequence," Hamilton told CNBC.

The answer is not preserving menial tasks for their own sake. Hamilton said companies can treat entry-level work as "exposure" rather than "output," with juniors checking and correcting AI-generated work while senior colleagues deliberately teach the judgement behind those corrections.

FDM Group is running a similar experiment. Flavell said the firm now trains people on real business problems and asks them to find solutions using agentic engineering, rather than simply teaching how the AI tools work — an approach she suggests could otherwise leave companies short of experienced workers in the coming years.

"This is why I argue the real issue isn't 'early careers' time served, but 'early knowledge,'" Hamilton added. "Are we deliberately building the contextual knowledge and judgement that used to accrue by osmosis? If not, you risk a generation with AI-enabled breadth and no depth."