MIT Economist: Sacking Junior Staff for AI Is the Wrong Move — Here's the Smarter Framework

Companies racing to cut entry-level roles with AI are making a mistake they'll pay for in five years, according to MIT economist Frank Nagle. His three-bucket model for thinking about AI and work is the most useful thing you'll read this week — and it's directly applicable to a team of three or thirty.

Frank Nagle, economist at MIT's Sloan School of Management, has a blunt message for business owners eyeing their junior headcount as an AI cost-saving opportunity: stop. Nagle argues that gutting entry-level positions in favour of AI automation is a "critical strategic mistake" — not for sentimental reasons, but for structural ones. The people doing those jobs today are the people running your business in ten years. Remove that pipeline and you hollow out your own future capability.

His framework divides all work into three broad buckets. First: tasks that AI can handle fully and reliably right now — data formatting, basic drafting, scheduling, routine correspondence. Second: tasks that are fundamentally human and will stay that way — relationship management, complex judgement calls, ethical decisions, anything requiring lived context. Third: the large middle ground where AI assists but a human directs, reviews, and takes responsibility. Nagle's point is that most businesses are deploying AI aggressively in bucket one while mistakenly treating bucket three as bucket one — and the consequences of that confusion are significant.

The research backing this concern is mounting. A 2024 report from the Resolution Foundation found that young workers aged 18–24 are disproportionately exposed to AI displacement in administrative and service roles — the same roles that have historically served as the training ground for every layer of management above them. The McKinsey Global Institute estimates that between 60 and 70 per cent of time currently spent on work tasks could be automated with existing technology, but that figure says nothing about whether it should be — or what gets lost in the process.

For Scottish SMEs, this matters in a specific way. Scotland's labour market is tight. According to the Scottish Government's Labour Market Statistics, unemployment in Scotland remains below the UK average, meaning the competition for experienced mid-level talent is fierce and expensive. If you use AI to eliminate the entry-level roles that develop that talent, you're not saving money — you're transferring the cost to your future hiring budget. A junior account manager who learns your clients, your systems, and your culture over three years is worth considerably more than whatever you save on their salary by running a chatbot instead.

The smarter model — and the one Nagle advocates — is augmentation over replacement. Deploy AI in bucket one aggressively: free your junior people from the drudgery of repetitive admin so they can spend more time in bucket three, developing the human skills that actually compound over time. A small Edinburgh firm that does this well ends up with a team that moves faster, learns faster, and stays longer. That is a genuine competitive advantage. The firm that replaces its graduate intake with a subscription to an AI tool is, as Nagle puts it, making a bet it cannot afford to lose.

What this means for you: Before you restructure any role around AI, map it to Nagle's three buckets: full automation, human-only, or assisted. Automate the first category hard — and use the time it frees up to develop your people in the third category, not to reduce headcount. If you're a small Scottish business where every person counts double, your junior staff are your bench strength. Invest in tools that make them faster, not decisions that make them redundant.