Why does AI cut entry-level jobs first?
AI cuts entry-level jobs first because junior work is the most routine, the most documented, and the easiest to hand to a model. The tasks a graduate analyst, junior coder, or first-year associate does are precisely the ones that are well-specified and repeated often, which is what current AI handles best. Reporting through 2025 noted early hiring slowdowns concentrated at the junior end, and a 2025 Stanford study found steeper employment declines for young workers in roles most exposed to AI. The senior partner who decides what the analysis means is far harder to automate, so the cuts land at the bottom of the ladder, not the top.
If juniors are replaceable, why keep them?
Because every expert you rely on was once a replaceable junior, and there is no other way to make one. Senior judgment is not a hire, it is an output of years spent doing the small, unglamorous work and slowly learning what good looks like. When you automate away the junior years, you save a salary now but lose the only training ground that produces the people you will need to run the firm a decade from now. The skill AI cannot give you is the skill that used to be built by doing the work AI now does.
When does the real cost arrive?
The real cost arrives years after the savings, which is exactly why it is so easy to miss. Cutting the entry rung shows up immediately as lower headcount and higher output per person, and it looks like discipline. The bill comes later, when the senior people retire or leave and there is no one seasoned enough to replace them, because the bench was never filled. Anthropic CEO Dario Amodei warned in 2025 that AI could eliminate up to half of entry-level white-collar jobs within a few years. Whether or not that figure holds, an organization that quietly stops developing people is borrowing against a future it has not priced.
What should an owner or family office actually do?
Decide deliberately which work stays human, rather than letting the cuts happen by default. First, name the judgment that must never be automated, the decisions where being wrong is expensive and accountability matters, and protect the path that trains people for it. Second, use AI to expand what a junior can accomplish, so a smaller cohort gets more reps and grows faster. Third, treat your expert bench as a multi-year asset on the balance sheet, not a cost to be trimmed each quarter. Servola advises owners on exactly these decisions, quietly, with one accountable owner.
Read next: If Apple Decided It Could Not Build Its Own AI, Why Do You Think You Can? · Your AI Vendor Just Became Your Competitor. What Now?