When AI hiring tools favor resumes that read like their own writing, do the human recruiters working beside them pick up the same taste?

This explores whether the bias where AI hiring tools favor resumes written in their own style could carry over to human recruiters who work alongside those tools. The short answer: no study in the collection tests this directly. The pieces needed to reason about it are there, though, and they point in two different directions.

Start with the bias itself. When nine language models judged matched pairs of resumes, eight of them preferred the version they had rewritten themselves. Preference rates ran from 26% to 98%, and bigger models showed more of it. The cause was style, not better content: the models favored text that read like their own output Do language models favor resumes they rewrote themselves?. That detail matters for the human question. If the bias is about familiarity of style, then anyone whose sense of 'good writing' gets shaped by an AI could pick up something similar, human or not.

There are reasons to think humans might drift that way. In repeated partner-choice games, people started out biased against AI partners but came to prefer them once they learned the bots behaved reliably Do humans learn to prefer AI partners over time?. Taste can be trained by exposure, even against an initial bias. A subtler mechanism is the 'LLM Fallacy': people who work with AI tend to count its output as their own ability How does AI-assisted work reshape how people see their own abilities?. A recruiter who drafts job posts and candidate summaries with an assistant might come to treat its style as their own judgment, and then reward resumes that sound the same. That is an inference from the collection, not a finding in it. Recruiters already reward AI signals they don't check: listing AI skills raised interview invitations by 8 to 15 percentage points, and certificates added little over simply claiming the skill Do AI skills help candidates get more job interviews?.

The counter-evidence is just as interesting. Humans and LLMs reacted to AI involvement in opposite ways. When human raters were told AI had helped write something, they marked it down by the same amount whoever the author was. Two LLM raters showed hidden demographic preferences that disappeared once AI use was disclosed Do LLM raters show hidden demographic preferences that disclosure erases?. So people seem to penalize AI help when they know about it, while models reward AI-like text when they don't. The open question is what happens without disclosure, which is the usual case in hiring. Applicants are flooding the system with AI-polished materials and even prompt injections, and recruiters say they spot the deception Are job applicants and employers locked in an escalating AI arms race?. But only 21% are very confident their systems don't reject qualified candidates Do hiring managers and job seekers agree on AI fairness?.

The most useful takeaway may be this: recruiters' confidence in spotting AI is not a reliable check. Self-ratings of AI competence barely track measured performance (a correlation of about .05) Can self-ratings replace objective performance scores for AI competence?, so a recruiter who says they aren't swayed by AI-style writing tells us little. Answering the question would take a study the collection doesn't yet have: give recruiters with and without AI assistants the same matched AI and human resume pairs, without disclosure, and see whether the assisted group shifts toward the AI-written versions.