When AI Interviews AI: The Emerging Reality of Bot-to-Bot Recruitment

Artificial intelligence has steadily infiltrated nearly every stage of the hiring process for both job seekers and employers, yet the interview phase has lagged behind. Voice-based AI remains notoriously difficult to execute, as automated agents frequently stumble over accents, verbal pauses, and conversational nuances. Behaviors that come instinctively to humans—such as refraining from interrupting a candidate—represent what Ophir Samson, head of voice AI at the recruitment platform Greenhouse, describes as a “very, very difficult engineering problem.” Despite these challenges, recruiters burdened by unprecedented volumes of applications have increasingly adopted voice AI screening tools. According to Greenhouse data, 63 percent of job seekers report encountering an AI-led interview during their search.

As employers embraced AI, applicants followed suit. The practice of candidates using artificial intelligence during interviews has become so widespread that recruitment startups like Ribbon now market tools designed to detect “overly scripted, AI-assisted, or coached” responses—precisely the type seen in one candidate’s recent experiment.

Some observers suggest that bot-on-bot interviews represent not deception but a glimpse into the future of hiring. Mark Monaghan, vice president of organizational development at the call center company IQor, characterizes this evolution as the “next logical stage” of recruitment technology.

This trajectory does not necessarily bring him comfort. In the case of one applicant who used AI to conduct his interviews, Monaghan believes the response was entirely justified.

“If you’re going to send a bot to me,” he explains, “I’ll send a bot to you.”

Following five consecutive interviews that led nowhere, the applicant—who uses the pseudonym Christopher—decided to test the system one final time with a company called Everforth Apex. Convinced that the issue might not lie with AI itself, he reasoned that perhaps a more qualified candidate would yield different results. He therefore constructed an entirely fictitious applicant named Don Dickner, crafting a résumé overflowing with credentials drawn directly from the job listing.

Once Christopher submitted this fabricated application, the virtual recruiter known as Riley responded immediately. On this occasion, Riley conducted a 23-minute conversation with ChatGPT, which had been programmed to impersonate Dickner. Their exchange touched on topics including “sustainable operational improvement,” preserving “customer experience under volume pressure,” and preventing “tribal knowledge drift.” ChatGPT delivered a prepared response for every question and offered a relevant personal anecdote to match each qualification. As the discussion drew to a close, ChatGPT indicated that it had “a few more questions I’d like answered.”

“I appreciate it,” Riley replied. “Upon carefully reviewing your application, we will assess your qualifications for the position. Should you meet our criteria, a recruiter from Everforth Apex will reach out to schedule a brief conversation.”

Christopher never received another message from Riley.

Source link