Listing 'AI skills' on a résumé lifts interview odds almost as much as a certificate, so are recruiters trusting proof or the claim?

This explores how AI skills act as a hiring signal next to traditional credentials like degrees, and whether employers can trust that signal. The clearest evidence is a large experiment with recruiters. Listing AI skills raised a candidate's chance of getting an interview by 8 to 15 percentage points across occupations. The surprising part is that a formal certificate added only a little over simply saying 'I have AI skills' Do AI skills help candidates get more job interviews?. Recruiters are rewarding the claim more than the proof. In that sense AI skills currently work less like a degree, which an institution vouches for, and more like a keyword that opens doors.

That keyword is strong enough to compete with credentials. In the same line of research, AI skills partly or fully offset the hiring penalties that older candidates and candidates with associate rather than bachelor's degrees normally face Can AI skills help older or less-educated job candidates?. That looks like a real opening for people traditional credentials leave out. The effect was much stronger for office assistant roles than for graphic designers, though. The signal is worth most where the job is least specialized. Job-vacancy data from ten countries points the same way. Demand for AI skills is splitting into a technical core (Python, SQL, machine learning) inside STEM jobs, and non-technical jobs are moving away from that core rather than toward it Is AI creating common skills across jobs or deepening divisions?. So 'AI skills' may mean very different things depending on whose résumé it appears on.

The problem is that a self-declared AI skill tells you almost nothing about actual ability. Pooled across three studies, people's self-ratings of their AI competence correlated with their measured performance at just .055, which is statistically close to zero Can self-ratings replace objective performance scores for AI competence?. One possible reason is that working with AI feels like competence. When a model produces polished output, users read that smoothness as evidence of their own skill, even though they didn't produce the work Does processing ease mislead users about their own competence?. Candidates may honestly believe they're proficient, and recruiters take them at their word.

There is a deeper reason this signal is hard to read. Traditional credentials assume that a polished essay or a finished project shows the thinking behind it. AI breaks that link: it can produce the outward form of intellectual work without the reasoning that normally creates it Does AI separate intellectual form from the thinking behind it?. Research on AI models shows the same split. Smaller models trained to imitate larger ones copy their surface style well but not their reasoning Do all AI skills improve equally as models scale?. Form is cheap to copy and substance isn't, for people and for models.

The corpus has no studies that track whether AI-skilled hires actually perform better on the job, so it can't say whether the signal is earned. What it does show is a gap. Employers are pricing AI skills like a credential, but nothing yet does the checking that makes a credential worth anything. Whoever builds reliable performance-based tests of AI competence will probably shape this part of the labor market.