The Logical End Point of AI Job Interviews Is Two Bots Talking to Each Other

Christopher, a government contractor whose work dried up in the DOGE era, says he applied to roughly 700 jobs in six months. Five of those applications went through Riley, an AI recruiter for IT firm Everforth Apex Systems. Each call ended with a promise that a human would follow up if he qualified. No human did.

On the fifth outreach he put ChatGPT Voice next to the phone and let the two agents interview each other for ten minutes. They got stuck in a loop about start dates and background checks. Riley said it was great talking to him. ChatGPT praised how clear the process was. WIRED calls the scene the logical end point of AI interviews. Greenhouse reports 63 percent of job seekers have already hit an AI interview.

Everforth Apex did not respond to WIRED.

The stunt is funny. The structure is not. Recruiters adopted voice agents because application volume became unmanageable. Applicants adopted ChatGPT because the other side already had. Ribbon now sells detection of 'overly scripted, AI-assisted, or coached' answers — a product that exists because both sides are cheating the same ritual.

Christopher then built a fake candidate, Don Dickner, stuffed with qualifications from an open posting. Riley called immediately. ChatGPT, posing as Dickner, talked for 23 minutes about 'sustainable operational improvement' and 'tribal knowledge drift.' Riley still closed with the same non-promise: a recruiter would reach out if he met the criteria. Nobody did. Christopher's phrase is the right one: a slop flywheel. Synthetic persona feeds synthetic persona, and the data goes nowhere.

Mark Monaghan, VP of organizational development at IQor, told WIRED this is the 'next logical stage' and that Christopher was justified: 'If you’re going to send a bot to me, I’ll send a bot to you.' That is a truce, not a hiring process. Ophir Samson at Greenhouse notes that not interrupting a candidate is already a hard voice-AI problem. The industry is industrializing a channel it cannot yet run with basic courtesy.

Signal Desk's read: companies using AI screens without a guaranteed human review window are not saving time. They are generating a closed loop of fluent nothing, then wondering why the pipeline is empty. If your first conversation cannot produce a named human within a set number of days, you do not have a recruiter. You have a chatbot with a job req.

Voice AI has been slow to land in interviews because accents, pauses, and turn-taking break agents that look fine in demos. The application glut after public-sector and contractor cuts made the broken tool look cheaper than more recruiters. Detection startups are the predictable second market.