AI Recruiters Deploy AI To Catch Applicants Using AI, Creating Perfectly Circular Hiring Market
Recruiters are fighting artificial intelligence with more artificial intelligence, leaving human judgement trapped somewhere between two loading screens.
LONDON, An annual survey by the Institute of Student Employers found UK reports of misconduct in job applications had reached a record high. The most common complaint was unauthorised artificial intelligence during virtual interviews and assessments.
Recruiters say they can identify the cheats. Their detection software watches applicants with the fixed suspicion of a customs officer examining a calculator. In one imaginary office, it flagged a candidate for saying “I enjoy problem-solving”, a phrase the same software had taught them.
The National Office for Digital Suspicion, which operates from a cupboard marked PLEASE DO NOT AUTOMATE, has introduced a three-stage screening process. An applicant receives a question. An AI answers it. Another AI then decides whether the applicant looked confused enough while the first one was speaking.
“We preserve the human element by sending it a push notification.”
“We are committed to preserving the human element,” said Priya Nettle, deputy director of Human Oversight and Decorative Plants. “Every automated decision is therefore reviewed by a human who gets the result as a push notification and approves it immediately.”
The system also looks for suspicious pauses, polished sentences and applicants who seem prepared. A spokesperson for the Bureau of Authentic Vibes said those signs caused concern because they suggested a candidate had “either used technology or taken the process seriously”, both of which remain under investigation.
At a demonstration witnessed by a man standing nearby, an interviewing robot asked an applicant where they saw themselves in five years. The applicant’s AI produced a detailed career plan. The recruiter’s AI rejected it for sounding too detailed. The recruiter then asked the applicant why they lacked ambition.
Officials say the contest can run indefinitely. Each side adds another layer of software until the interview becomes a silent exchange between encrypted dashboards. The successful candidate will be whoever has a machine most capable of proving they are not a machine, probably by producing an error designed specifically to look human.
The survey’s actual finding is that unauthorised AI use in applications and assessments is rising. Everything else follows naturally: an employment system restoring trust by placing two mutually suspicious computers between a person and a question.