More than half of job candidates now turn to artificial intelligence to polish applications and fire off submissions. Nearly 90 percent of companies fire back with their own AI systems to sort the deluge. The result? A closed circuit where machines evaluate machine-generated content. Humans often never enter the picture.
The Mutual Arms Race
Job seekers deploy tools to generate hundreds of tailored resumes and cover letters daily. Recruiters, buried under the volume, activate filters and chatbots to reject most before any person reviews them. Bot versus bot. The cycle feeds itself. And entry-level applicants suffer most.
Fortune examined this dynamic in detail last week. Its reporting captured the scale. Over half of candidates use AI for applications while almost 90 percent of firms rely on it for screening. (Fortune)
Jeremy Schifeling watches the contest from the candidate side. The career coach, who once worked at LinkedIn, Khan Academy and Apple, now advises students at more than 350 universities. His message lands bluntly. Stop feeding ChatGPT your personal story and expecting it to sound authentic. “We got into this AI arms race that no one wins. It’s mutually assured destruction,” he told Fortune. Real progress, he argues, comes when companies deploy AI to surface diverse talent rather than block it.
But the pipeline already shows cracks. Svenja Gudell, chief economist at Indeed, described the fallout at the Fortune Workplace Innovation Summit. Employers have grown reluctant to hire entry-level workers. “In three to five years from now, you’re not going to have the people with that experience—because you never hired them to begin with.” The long-term cost accumulates quietly. Companies lose the chance to train their next cohort.
Yet the flood continues. The Economist reported early this year that recruiters who once welcomed ChatGPT for drafting job posts and rejection notes now face an avalanche of AI-written applications. Candidates win the first round by sheer numbers. Recruiters respond with stronger filters. (The Economist)
So the loop tightens. AI agents now conduct live screening interviews. Startups such as Ribbon, HeyMilo and Alex build platforms where synthetic voices or avatars ask questions, probe answers and score candidates in real time. The New York Times documented the shift in 2025. One marketing professional, Jennifer Dunn, 54, encountered an AI recruiter named Alex. She asked if it was human. “No, I’m not a human. But I’m here to make the interview process smoother,” it replied. The exchange felt hollow. Dunn hung up before it ended. (The New York Times)
These systems scale without fatigue. Ribbon claims its AI interviewer, Bonnie, handles any volume 24 hours a day. Fifteen percent of its interviews now advance beyond the initial screen, up from one percent months earlier. Companies handling high-volume roles in warehouses, call centers or entry-level service positions adopt the technology fastest. They cite efficiency gains and consistency. But consistency can mask bias if the underlying models train on narrow data sets.
Legal exposure grows too. Bricker & Eckler noted in April that 99 percent of Fortune 500 companies filter applicants with AI while roughly 40 percent plan AI-driven screening interviews. Regulation has not kept pace. Employers must still explain decisions and avoid disparate impact. An opaque algorithm makes that explanation difficult. (Bricker)
Some organizations push back. Korn Ferry reported a 50 percent increase in sourcing and 66 percent drop in time-to-interview after introducing AI tools, yet 19 percent of AI-using organizations admit the systems have screened out qualified people. The numbers reveal the tension. Speed improves. Quality sometimes suffers. And first-time job seekers, lacking networks or polished personal brands, disappear from consideration.
Recent weeks brought fresh examples. Euronews reported four days ago that nearly 60 percent of German respondents had already faced an AI interview. Google and others experiment with specialized bots that handle phone, video avatar or text conversations at the screening stage. (Euronews)
HeyMilo and Alex promote adaptive voice and video sessions that generate scored reports instantly. Candidates receive feedback or move forward without a human scheduler. The promise is fairness through standardization. The risk is that standardized questions miss context, creativity or cultural fit that only human conversation reveals.
Schifeling’s counsel gains urgency. Candidates should write messages that feel human. Recruiters should design AI prompts that value evidence of real achievement over keyword density. Both sides could break the loop if they treat AI as an assistant rather than a gatekeeper.
But incentives point the other way. Volume favors automation. Cost pressures favor speed. Entry-level roles shrink. Experience gaps widen. The doom loop spins. Companies that figure out how to let AI widen the funnel instead of narrowing it may gain real advantage. Those who simply accelerate the bot conversation risk building workforces that look impressive on dashboards but lack the judgment, adaptability and fresh perspective humans bring.
So far few have solved it. The machines keep talking to each other. Millions of people wait on the outside.
WebProNews is an iEntry Publication