AI in hiring traps both sides in a loop nobody wins

Jodi Beggs, a data scientist hunting for a role, discovered her résumé was rejected for running two pages long. Adding a middle initial to one version and dropping it from another lowered her score. Swapping the word "percent" for the % symbol boosted it. The feedback came from a tool that compares a résumé against a job description and ranks the match — the best-known being Jobscan. "If my job is to please the robots," she says, "then even if I don't like AI-generated content, maybe the machines do."

The ATS myth versus the reality

The premise behind these tools is that hiring teams use AI to auto-rank candidates inside an applicant tracking system, or ATS, and that only the top decile or quintile ever gets human eyes. The reality is messier. Daniel Chait, chief executive of Greenhouse — itself an ATS vendor — says there is a huge amount of folklore around these systems, mostly from job seekers. No two systems are alike, the technology moves fast, and whether an AI component exists at all depends on the specific product and the features an organization has bought and switched on.

The optimization business model

Handing your job search to a tool that promises ATS optimization is a risky bet. One recruiter proved that even with poor Jobscan scores she landed a dozen interviews and an offer. Jobscan runs $30 to $50 a month (roughly 110 to 185 shekels), and the company is not exactly incentivized to move you out of the job-seeker pool. Add an economy of few layoffs and little hiring, ghost postings, and scams that legislatures are now weighing laws against, and you get a two-sided trust crisis: candidates pour in hours and hear silence, while employers receive hundreds of résumés that look nearly identical.

The AI doom loop

Chait calls it an "AI doom loop." Both sides have a problem; each uses AI to solve its own problem in a way that makes the problem worse, so more AI begets more AI with no benefit to anyone. The more it happens, the worse it gets. Employers drowning in identical files feed them into auto-ranking systems trying to manufacture a quick distinction, which only ratchets up the pressure on candidates to keyword-stuff and format-tweak.

Philosophy, not organization size

I spoke with dozens of recruiters, HR leaders, and small-business owners who hire directly. Some admitted to auto-ranking; others insisted they don't. The split doesn't track organization size or application volume — it comes down to philosophy and culture. Kim Jones, vice-president of human resources at Toshiba, says humans review every application there. She doesn't object to candidates using AI to polish materials, but she is clear that it "won't really help get past the ATS." Their screen is based on role requirements and salary.