Do AI skills help candidates get more job interviews?

The paper's central claim is that AI skills work as a hiring signal in their own right. Its authors ran "an experimental survey with 1,725 recruiters" who compared pairs of synthetically designed résumés and chose which candidate to invite. Across graphic design, office assistance and software engineering, AI skills "significantly increase interview invitation probabilities by approximately 8 to 15 percentage points" compared with candidates without them. Credentials such as a university or company-backed certificate "only lead to a moderate increase in invitation probabilities compared with self-declaration of AI skills." The authors call AI skills "a powerful hiring signal". The estimate is theirs, from their own survey, and it measures recruiter choices, not hiring outcomes.

The reasoning runs through signaling theory. Frankel and Kartik's "Muddled Information" framework argues that as signals get cheaper to manufacture, they tell employers less about natural ability and more about "gaming ability". The paper's expectation follows from that: "If employers view AI skills with uncertainty, costly and verifiable credentials should provide stronger signals than self-reported claims." The reported result is consistent with that expectation, though the credential gain is described as moderate. The design is what supports the causal reading. Each pair differs only on the manipulated attributes, and the forced choice is meant to remove scale-use bias and reduce social desirability bias. The authors contrast this with observational studies of job postings, résumés and wages, where AI proficiency may stand in for motivation, elite access or multiskilling.

The nearest notes sharpen where this sits. Can self-ratings replace objective performance scores for AI competence? argues that self-reports are a weak basis for judging performance. This paper's recruiters reward self-declared AI skills without any check on them, so the two results make a useful contrast, though this paper does not measure candidate competence. Is AI creating common skills across jobs or deepening divisions? tracks demand in vacancy data, while this paper finds the premium varies by occupation: strongest for office assistants and weaker for graphic designers, where the excerpt reports "more skeptical recruiter attitudes toward AI in creative work." These are different measures and should not be read as one trend. The finding that recruiters who use generative AI assign a higher premium fits What makes accountable judgment scarce when AI cognition is cheap?: the value of AI skills is set by gatekeepers, not by the skills alone.

The excerpt does not establish several things. The outcomes are stated preferences in a hypothetical scenario, a limit the authors name themselves. The excerpt gives no confidence intervals or standard errors for the 8 to 15 point range, and the Annex C tables it cites are not included. The sample is described inconsistently: the abstract lists recruiters from the UK, the US and Germany, while the introduction and conclusion say UK and US. Nothing in the excerpt measures wages, hires or productivity, which the authors list as future work. At the strength the evidence allows, AI skills raise the chance that a résumé is invited to interview in a hypothetical screen, across three occupations. Whether that carries through to pay or performance is open.