User Perceptions and Labor Market Outcomes of AI-Generated Resumes
This mixed-methods study investigates how recruiters perceive AI-generated resumes and whether their use improves hiring outcomes, aiming to determine if the structural advantages of AI tools are offset by an "authenticity penalty" when authorship is suspected.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
The modern job hunt has become a game of two distinct players, both increasingly aided by machines. On one side, companies use automated systems to scan thousands of applications, looking for specific words and patterns that match their needs. On the other side, job seekers now have access to powerful computer programs that can write and polish their resumes, turning raw life experience into professional-sounding documents. These tools promise to level the playing field, helping people translate unusual backgrounds into standard qualifications. But a critical question remains unanswered: do these computer-written documents actually help people get hired, or do they trigger a hidden bias among the human recruiters who make the final decisions?
A new study by Elijah White at the University of Florida investigates this exact tension. The research team wanted to see how professional recruiters react when they see a resume written by a human, one written entirely by an artificial intelligence, and one that starts with AI but is then edited by a human. They combined three different methods to get a complete picture: they watched recruiters' eyes as they read resumes, they tracked the real-world job search results of hundreds of people over six months, and they asked recruiters directly what they thought about AI-written applications. The goal was to move beyond technical tests of the software and understand the human side of the hiring process.
The researchers began by placing sixty professional recruiters in front of a screen that tracked where their eyes looked and for how long. These recruiters reviewed sets of resumes for jobs like project manager and data analyst. Some resumes were written by human career counselors, some were generated by an AI tool, and some were a mix where a human edited the AI's work. The eye-tracking data revealed a clear difference in how the documents were processed. Recruiters spent about fifteen percent less time looking at AI-generated resumes compared to those written by humans. Their eyes moved in a straight, linear line down the page, suggesting they were scanning for quick facts rather than reading a story. In contrast, when looking at human-written resumes, their eyes jumped around more, indicating they were actively searching for specific details. Despite this faster, more superficial reading, the AI resumes were not rated as lower quality overall. However, they did receive significantly lower scores for authenticity. Recruiters felt these documents lacked a genuine voice, a problem that was especially noticeable for creative roles like marketing, where personality matters more than for technical roles like data analysis.
The study then moved from the lab to the real world to see if these perceptions translated into actual job offers. Two hundred job seekers were split into two groups. One group used an AI resume generator to help them apply for jobs, while the other group used traditional professional resume services. Over six months, the group using AI sent out thousands of applications and received more interview invitations than the comparison group. Specifically, fourteen point two percent of their applications resulted in a callback, compared to eleven point eight percent for the traditional group. This advantage was not spread evenly; it was most powerful for people making difficult transitions, such as academics moving into industry or workers returning to the job market after a long break. For these individuals, the AI tool seemed to successfully translate their unique experiences into the standard language that employers understand.
However, the benefits came with a catch. The study found that the most successful applicants were not those who simply accepted the AI's output without change. Instead, the best results came from a "centaur" approach, where the job seeker used the AI to generate a draft but then heavily edited and personalized it. These participants saw the largest increase in interview callbacks. This suggests that the most effective strategy is to use the machine for structure and keyword optimization while keeping the human touch for voice and specific details. The researchers also discovered that the AI tool appeared to act as an equalizer. Women and people from racial minority groups saw a much larger boost in callback rates when using the AI tool compared to their white male counterparts, whose results did not change significantly. This implies that the tool may help candidates overcome the subtle biases that often affect how their qualifications are perceived.
The final piece of the puzzle came from interviews with the recruiters themselves. The researchers found that recruiters are wary of AI-generated content, often looking for subtle clues like overly uniform bullet points or generic language. If a recruiter suspects a resume was written by a computer, they tend to rate it lower, even if the content is excellent. This "authenticity penalty" means that a perfectly qualified candidate could be rejected simply because their document feels too polished or impersonal. One recruiter noted that the best resumes tell them something about who the person is, and if a document could have been written by anyone, it raises suspicion. Yet, a minority of recruiters expressed a more pragmatic view, accepting that AI is now a standard part of the process and focusing solely on whether the candidate can do the job.
The study concludes that AI resume generators are neither a magic solution nor a trap. They are a tool that works best when used as a starting point for human refinement. For job seekers, the path to success involves using the technology to organize their experience and hit the right keywords, but then taking the time to inject their own voice and specific stories back into the document. For employers, the findings suggest a need to be careful about how they judge applications, as their instinctive suspicion of AI writing might cause them to miss out on talented candidates who simply used a modern tool to prepare. As artificial intelligence continues to reshape the labor market, the most effective approach appears to be a partnership between human judgment and machine efficiency, ensuring that the final product is both optimized and authentically human.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.