As AI makes résumés easy to polish, does sitting across from a candidate actually prove their credentials are real?
Do in-person interviews reduce concerns about fake credentials?
This explores whether meeting candidates face-to-face helps employers trust that skills and credentials are real, at a time when AI makes résumés, applications and even evaluators easier to game.
This explores whether in-person interviews can restore trust in credentials now that AI is flooding hiring with polished, hard-to-verify claims. The short answer: the corpus has no study that tests in-person interviews directly. What it does show is why the question keeps coming up, and why a face-to-face meeting may fix less than people expect.
Start with the credential problem itself. In a large experiment with recruiters, listing AI skills raised a candidate's chance of an interview by 8 to 15 percentage points, but a formal certificate added only a little over simply claiming the skill Do AI skills help candidates get more job interviews?. Recruiters were rewarding the claim without checking it. Scale that up and you get what Greenhouse's CEO calls a hiring 'doom loop': 41% of job seekers say they've used hidden AI prompts to slip past screening filters, and 34% of recruiters spend half their week weeding out spam Are job applicants and employers locked in an escalating AI arms race?. In-person interviews look attractive because they step outside this loop entirely.
The less obvious reason to want a human in the room is that the automated gatekeepers are also easy to fool. LLM judges reliably score answers higher when they include fake references or heavy formatting, regardless of quality, and no access to the model is needed to exploit this Can LLM judges be fooled by fake credentials and formatting? Can LLM judges be tricked without accessing their internals?. When LLMs screen résumés, eight of nine models preferred résumés they had rewritten themselves over matched human-written versions. The preference came from writing style, not content Do language models favor resumes they rewrote themselves?. So a 'fake credential' isn't only a false claim. It can also be a real claim written in the style a machine screener happens to like.
This is where the corpus warns against treating face-to-face contact as the cure. Research on ChatGPT found that people trust a conversation partner for being responsive, fast and fluent, not for being accurate Does conversational style actually make AI more trustworthy?. Humans in interviews rely on the same shortcuts, so a confident, smooth candidate can win trust they haven't earned. An interview mainly proves something narrower: that a real person exists and can perform live. Personhood credentials try to verify exactly that, digitally and without revealing identity Can people prove they are human without revealing who they are?. That suggests the in-person interview is really a way of proving you're human. It is not a way of proving competence.
The takeaway: in-person interviews can rule out bot applicants and AI-written applications, but they replace one set of easily gamed signals (formatting, keywords, authority cues) with another (fluency, confidence, rapport). If you want to go further, the open question is how to verify competence itself, through live work samples or credentials that are actually checked, rather than just moving the trust problem into a different room.