Everyone Looks Qualified Now. That's the Problem

For years hiring has run on the same few signals: the CV, a good interview answer, a tidy list of credentials. That system is starting to fail. Around 90% of hiring managers say that working out who is actually any good from applications and interviews has got harder over the last two years, and most of them know why. AI has made every application read well. Three in four job seekers now use it to write their applications, and the volume shows it: applications per recruiter are up 412%, and the average posting now draws 254 candidates (Greenhouse, AI in Hiring 2025). Cover letters are sharper, written answers are more polished, and candidates now turn up sounding more or less equally strong on paper. So the top of the funnel fills with people who all look like a great hire, and the old filters that used to sort the strong from the average just don't do it any more. Several managers described the same thing: a candidate who looks impeccable in writing, then cannot explain their own claimed achievements when you ask them to. The pattern is familiar in the UK too: BBC reporting on AI in hiring has employers describing an explosion of applications and a growing difficulty telling one polished CV from another, while candidates say they rarely get past automated screens to a human.

Look at what actually predicts performance and it gets uncomfortable. The signals we lean on most, meaning CVs, degrees, references and the interview itself, are the ones hiring managers rate as the worst at predicting who will do the job well. Close to 60% have watched someone interview brilliantly and then struggle once they start. A Robert Half survey found most HR leaders saying AI-generated applications are slowing hiring, and most hiring managers saying AI-enhanced resumes make skills harder to verify. And it is no longer just the interview that can be gamed: attempts to cheat assessments have doubled in a year, from 16% to 35% of candidates, and closer to 40% at entry level (CodeSignal 2025). In one vendor analysis of 19,000 real interviews, 38.5% showed signals of AI-assisted cheating, and 61% of the flagged candidates passed anyway (Fabric, State of Hiring Fraud 2026). Candidates feel the opacity too: Gartner found just 26% of applicants trust AI to evaluate them fairly. When a hire does not work out, it is slow and expensive to put right. It often takes one to three months just to realise there is a problem, and for around 40% of managers a single mis-hire costs somewhere between 15 and 50 thousand pounds once you add up salary, lost time and rehiring. We are making some of our most expensive, hardest to reverse decisions on evidence we admit we do not trust.

The part that stands out most is what hiring managers say does work. More than 80% say that seeing a candidate do a real piece of the actual work would be valuable, and they rate it above any credential or conversation as the best indicator of how someone will perform. Yet it is the one thing almost none of them feel able to use right now. That gap, between what we know predicts performance and what our hiring actually measures, is the real problem. In a world where anyone can sound qualified, the most useful question left is also the simplest. AI can write their CV. It can't do the job. So the only thing worth asking is: can you show me?

(Research carried out July 2026)

---

Sources

1. Greenhouse recruiting benchmarks — applications per recruiter (+412%) and candidates per posting (254).
2. Greenhouse, AI in Hiring / “AI trust crisis” — AI use in applications and hiring volume (2025).
3. BBC News, “Computer says no. Are AI interviews making it harder to get a job?” — UK reporting on AI screens and polished applications (18 Mar 2026).
4. Robert Half survey on AI-generated applications — slowed hiring and harder skill verification (Mar 2026).
5. CodeSignal on assessment fraud — cheating attempts roughly doubled (16% → 35%) in 2025.
6. Gartner on applicant trust in AI evaluation — 26% of applicants trust AI to evaluate them fairly (2025).

Fabric, State of Hiring Fraud 2026, is cited above as vendor analysis of interview AI-assistance signals; no public primary URL was available at publication.