Nepotism in the Age of AI: Same Old Game, New Tools
Everyone needs at least one break. Most careers start with someone taking a chance on us: a teacher who made an introduction, a manager who overlooked a thin résumé, a friend who passed along a posting. Sociologist Mark Granovetter showed decades ago that many jobs come through personal contacts rather than open applications.
So the problem with nepotism isn’t that people get breaks. It’s who gets them, and who never sees the door.
It shows up where the stakes are highest
Look at police services, fire departments, unions, and medicine. Entry is gated, the work is high-trust or high-risk, and much of the knowledge is tacit. Someone raised around the work knows how the exams, interviews, and culture operate. Nobody has to bend a rule for that to be an advantage. Seats, postings, and apprenticeships are scarce, and the gatekeepers are often insiders.
Two things get lumped together here. Children following a parent into a vocation is often admirable. A relative getting an interview, a posting, or a pass on a standard because of the surname is favoritism. The test is whether the same documented criteria and independent review apply to everyone.
Why it persists: the four layers
I keep coming back to the idea that people don’t change, because the world they operate in doesn’t change. Outcomes follow four layers:
- Physics. Time, attention, and information are limited. Nobody can evaluate every candidate, so we lean on shortcuts, and trust is the oldest shortcut.
- Human incentives. The person with a slot to give gains from favoring family and allies: loyalty, lower risk, repaid favors, and the deep instinct to look after your own. The costs fall on strangers who never see the decision.
- Power dynamics. Incentives become outcomes where someone has discretion and faces little accountability. Outsiders lack the information, standing, or leverage to push back, and pushing back can cost them the very opportunity they wanted.
- Institutions. Rules are how a society tries to change the payoff: open competitions, documented criteria, conflict-of-interest declarations, independent review, public data. They work only when the people being constrained can’t easily rewrite them.
Nepotism is what you get when individual incentives meet concentrated discretion. A new technology doesn’t repeal any of that.
How AI can hurt
- It devalues the old signals. When anyone can generate a polished résumé and cover letter in seconds, employers drown in applications that all look good. Credentials and writing samples carry less weight, so decision-makers fall back on what AI can’t fake: a trusted person vouching for someone. Referrals become worth more.
- It can launder favoritism. A model trained on past hiring data can learn that people from certain schools, networks, or families succeed, then reproduce that pattern with a veneer of objectivity. “The system picked them” is harder to challenge than “the boss picked them.”
- It guards the front door only. Humans configure, deploy, and override screening tools. The real decisions can still happen through a side door: a direct appointment, a “special project,” a contract that never goes to tender.
- It creates a new inherited advantage. Better tools, better coaching, and more time to learn them flow through the same family and network lines as everything else.
How AI can help
- Auditing. This is the biggest opportunity. AI makes it cheap to analyze hiring, promotion, overtime, and contracting data for patterns: do family ties or shared addresses predict outcomes beyond what stated criteria explain? A pattern that once took a team of auditors months to find can be surfaced by one motivated person.
- Testing skills directly. Work samples, structured assessments, and standardized interviews can reduce reliance on pedigree and gut feel.
- Widening the pool. Used well, AI can surface candidates who would never have had a connection and help people outside the network prepare for processes insiders already understand.
The tool amplifies whoever holds it
AI is not on the side of fairness or favoritism. It amplifies whoever controls it. In the hands of people with discretion, it adds a polished surface to existing preferences. In the hands of journalists, auditors, applicants, and citizens, it makes patterns visible that used to be easy to deny.
So the useful questions are the same ones we’ve always needed to ask:
- Who decides?
- What do they gain?
- Who can check them?
- What happens when someone objects?
If the answers are “insiders, a lot, no one, and nothing,” family networks will keep showing up no matter what tools are in use. If we want fewer closed doors, the work is institutional: posted openings, documented criteria, conflict-of-interest disclosure, public data, and independent review. AI can make that work cheaper. It can’t make anyone do it.
That should make us less fearful and more active. The technology isn’t going to fix this or ruin it. People will, as usual, and we get a say in which.

No comments:
Post a Comment