Last week in the credit-bearing course I co-teach, we asked students to listen to a podcast episode from MIT Technology Review about how AI is impacting the job application process. Listening to the spirited classroom discussion led me to reflect on my own job search this year and the role AI may have—or may not have?—played.
During my job search (November 2024-March 2025), I experienced a lot of AI-related anxiety, just like my students feel about their own impending job searches. What I remember most was the anxiety triggered by not knowing. Would a human ever look at my CV and cover letter (which I spent hours customizing for each application), or would an AI tool scan and discard it in seconds? What does AI even look for in CVs and cover letters? How could I craft a job application that would stand out to both human and AI reviewers? If I was not offered an interview, was it because my CV didn’t include keywords an AI tool was programmed to look for, or was it because a human had holistically reviewed my application and determined I was not the right fit? How could I revise my materials to have a better chance at future applications? It felt like I was throwing spaghetti at a wall I wasn’t allowed to see.
I applied to positions at 26 academic libraries, and only one—at Gwynedd Mercy University—gave me the ability to opt out of having an AI tool scan my resume and opt in to the guarantee of a human reviewer. This was accompanied by a statement that my selection would not affect my chances of being offered the position. No other universities provided either a) information on if or how AI would be used to evaluate my application or b) the ability to opt out of AI being used to evaluate my application and/or opt in to the guarantee of a human reviewer.
AI’s impact on the job search is far-ranging and ever-increasing. Current AI applications include employers using AI to conduct interviews, research candidates, and write job ads and job seekers using AI to write applications. But today, I’m going to focus on employers using AI for the initial scan of an applicant’s materials before an interview is offered.
I understand why, on the surface, tools like AI resume scanners may seem like positive additions to a library search committee’s toolkit. As job creep and budget cuts worsen year over year, librarians are stretched more thinly than ever, constantly asked to do more with less. Additionally, many library positions are being eliminated or left unfilled, which results in a shrinking job market with larger applicant pools for the few available positions. Using AI to scan resumes and evaluate cover letters could save valuable time, freeing search committee members to spend more hours on essential library work.
However, I suggest that “saving time” is an insufficient positive to counteract the negatives of using AI to evaluate job applications. One significant cause for concern is the influence of algorithmic bias. Some advocate for the use of AI to evaluate job applications because they believe it will eliminate human bias, but unfortunately, that isn’t the case. AI is built by humans and trained on data created by humans. If a hiring algorithm is trained on a biased, incomplete, or nonrepresentative dataset, unfair outcomes may result.
One notorious example of this occurred at Amazon a few years ago. They trained a hiring algorithm on the resumes of people who already worked at Amazon, then deployed it to evaluate incoming applications. The problem? Like much of the tech industry, Amazon is dominated by men. The algorithm accidentally learned that candidates whose resumes included the word “women’s” or mentioned all-girls’ schools were less likely to succeed at Amazon. It penalized those applications in favor of applications from men. This led to the algorithm being discontinued in 2018.
AI tools may overlook qualified candidates for other reasons as well. Without the ability to reason like a human, AI may discard strong candidates due to non-standard resume formatting or unusual (but applicable) work experience. Data privacy is another issue that should be particularly worrisome to librarians, given how our profession values privacy as an essential component of free and fair societies. These AI tools may be operated by third-party vendors that do not adhere to the same privacy standards we do, and resumes, CVs, and cover letters contain very sensitive information. Who is collecting this information? Who can access it? How long is it being retained?
In short, although the job market for academic libraries is getting worse, AI is not the solution. It will only introduce more problems, such as discarding strong candidates, and intensify problems that already exist, such as bias in the profession. As a profession, we need to be proactive about discussing these issues and setting standards. Do we think it is acceptable to utilize AI to evaluate job applications? If so, when and how? For the reasons briefly explained above, I argue that we should not use AI to evaluate job applications at this time.
However, I understand that while our profession leans more AI-critical than some, many librarians are not quite as AI-critical as I am. Therefore, I also suggest that if AI is used in order to evaluate job applications, the following safety measures should be in place:
- A clearly worded, prominently displayed way for the applicant to opt out of AI being used to evaluate their application and opt in to a 100% human review.
- A clearly worded, prominently displayed guarantee that whether or not they opt out of AI review will not impact their chances of being selected for the position.*
- A clearly worded, prominently displayed explanation of applicants’ data privacy rights in the case they select AI review (who will have access to their information, how long will it be retained, can they get it deleted (if so, how), and so on).
- Clearly listed contact information for somebody the applicant may reach out to with questions or concerns about any of the above.
Finally, whether or not AI is used in order to evaluate job applications, a transparency statement must be included on all academic library job ads, explaining the following:
- Whether or not AI will be used to evaluate applications.
- If yes, how AI will be used (in detail: at what stages, which platforms, and so on).
- Why this is the library’s policy.
Additionally:
- This transparency statement must be clearly worded, detailed, and prominently displayed on the job ad.
- There must be clearly listed contact information for somebody the applicant may reach out to with questions or concerns about the transparency statement.
- This transparency statement must be accessible at all times during the application process. If the job ad is taken down before the position is filled, for example, this statement could be attached to email correspondence with applicants under consideration.
*As I’ve explained above, I actually do think that AI evaluation negatively impacts a qualified candidate’s chances of being selected for a position. However, I presume that somebody who chooses to use AI to evaluate applications would disagree with this. What I’m trying to get at with this bullet point is that the applicant should have a tangible record clearly stating that applications will—theoretically—not be preferenced based on mode of review (human or AI) but will be judged based on content, with the same rubric applied regardless of the mode of review.
–
In this post, I have chosen to focus on AI usage in evaluating job applications, but there are many other intersections of AI and academic library HR that should be frankly and transparently discussed. For example, considering the ethics of our profession, what standards should we implement around using AI to interview candidates, monitor employees, or evaluate employees?
I am not an HR professional, nor have I ever sat on a search committee. My perspective is that of a recent job seeker, and my intention is to jumpstart conversation among academic librarians more broadly. To that end, I would deeply appreciate hearing your perspectives in the comments. Here are some questions to prompt discussion, but I would like to hear any thoughts this post has brought up for you.
- For all: What do you think of the policies I’ve proposed in this post? Do you have questions, concerns, or additional or alternate suggestions?
- For those who have recently searched for a position within academic libraries: What AI-related concerns or experiences did (or do) you have throughout your job search? How much transparency did you experience regarding how AI was used to evaluate your application?
- For those who have recently served on a search committee: Have you used AI or considered using AI to evaluate job applications? If so, how and why? I’m also wondering how much choice we have in the digital HR infrastructure we use for academic library job searches. For example, if the university typically uses AI for a first-round scan of applications, would it be possible for the library or other academic departments to opt out (or vice versa)? What would this process look like?