The Ghost and the Machine: Dispatches from Academic Reference Services in 2026

Mac Murray

MA in English, New York University

MSLIS, Long Island University

The Ghost

        A specter is haunting the university library – the specter of student protest.^[1]

In the wake of the 2024 student protests against the war in Palestine, universities across the country responded by instating new, ostensibly temporary security measures that profoundly reconfigured their campuses. New York University added over 28,000 hours to its security guard staffing during that time, and reported benefiting from an additional 6,000 hours of NYPD surveillance on campus.^[2] It is unclear to what degree these staffing levels have been maintained, but everyone who has spent time on campus before and after the protests can recognize that there have been seismic shifts. Now, to enter Bobst Library (the sole and central library for the Washington Square campus), students must pass three separate checkpoints: they must navigate around an orange security barricade outside the doors, present their student ID cards to security guards stationed outside these doors, and then tap their student ID cards on turnstiles to finally gain entry to the library itself. Security guards staff desks on either side of these turnstiles, as well as a booth by the express elevator that takes administrative staff and authorized visitors to their offices on the 11th and 12th floors.

Coined in 2003 by computer security expert Bruce Schneier to describe the tactics of the Transportation Security Administration (TSA), the term “security theater” refers to high-visibility measures designed to result in an increased sense of public safety, without any necessary correlation to actual safety.^[3] Like all good theater, security theater works best by obscuring its behind-the-scenes machinations and immersing audience members in the performance. It is a Foucauldian tactic to “control behavior by virtue of being conspicuous.”^[4] The lens of security theater seems apt to apply to NYU’s present-day security apparatus, since it is both highly conspicuous and apparently illogical (if the purpose is to prevent violence, why does it consist of ID checkpoints and not, say, metal detectors? Moreover, from whence does the sudden urgency to prevent violence spring, given that NYU’s 2024 Bobst sit-in was entirely peaceful,^[5] and the student pro-Palestinian movement has been overwhelmingly peaceful as a whole?^[6]).

Remarkably, though, Bobst Library does not have the beleaguered feeling of a TSA screening line. More survey research is needed to determine how exactly this surveillance makes students feel about entering the library – it is easy to imagine some students feeling comforted, others (especially students who are vulnerable to police violence or deportation) uneasy, and still others merely annoyed. From anecdotal observation, it seems they have mostly learned to take these security measures in stride, and the interior of the library still bustles with activity: students working independently, co-working, socializing, and approaching the Circulation and Reference desks for assistance. For the most part, the heightened surveillance has been folded into everyday academic life. It is sometimes thrown into sharp relief, though, when a non-NYU student attempts to access the library, such as an alum or a student at a consortial partner school – both groups were formerly allowed walk-in access to Bobst but now must register days in advance, and have limited resources available to them.^[7] Thus, the effects of tightened security on library use are at least twofold: both affective (characterizing the users’ experience of the library space before they even walk through the doors) and material (fewer people are able to access library resources, including the reference desk itself).

I passed through this security theater every time I arrived for a shift shadowing the reference desk – so did the library workers whom I was shadowing, and so did all the students who approached the desk for assistance. In fact, during those interactions, that was one of the few things we all definitively shared in common. What did it mean for all of us to be bound together like that – not only in intellectual inquiry, but also as members of a community literally gatekept from the city surrounding us, under constant surveillance and scrutiny? How does that impact the kinds of questions we could imagine discussing with each other? The university library is architected to encourage learning, collaboration, and critical inquiry – but only to the extent that such inquiry does not prompt criticism or protest of the university itself.

A search for the term “security” in the EBSCO Library & Information Science Source database overwhelmingly returns sources focused on cybersecurity, suggesting that this is an undertheorized space in the field.^[8] And since it seems likely that more libraries – both academic and public – will be circumscribed by security for some time to come, I encourage soon-to-be information professionals like myself and my peers to continue to think and write critically about this issue, beyond the scope of this essay. It will continue to define both our professional experiences and the experiences of the patrons we seek to serve.

The Machine

        There is another powerful force at play in the university library, one less visible but even more ubiquitous. It is estimated that 86% of undergraduate students use generative AI, such as ChatGPT, to help with schoolwork.^[9] Many education researchers have sounded the alarm, cautioning that it could reduce critical, creative, and reflective thinking capacities.^[10] But many have also encouraged a cautiously optimistic outlook. A 2024 qualitative case study in the Journal of Web Librarianship, for example, found that “ChatGPT offers considerable benefits in enhancing knowledge management and user interaction,” though “it requires careful integration and ongoing management to realize its full potential.”^[11]

        From a standpoint purely focused on information retrieval, ChatGPT is mostly a terrific tool. It synthesizes vast quantities of data almost instantaneously, presenting users with ready-made answers to complex questions. However, its helpfulness wanes as inquiries become more specific and localized. It also has a tendency to fabricate sources (among other attribution problems).^[12] Human library workers, on the other hand, never fabricate information, and they reliably exceed chatbots in terms of accuracy, relevance, and friendliness.^[13] Working with a reference librarian still offers many benefits over chatbots for students.

In addition to citation woes, there are many potential downsides to AI use, including environmental impact and the ethical concerns of its data scraping. Here, I am most interested in the psychic and moral injury that could result from habitually turning to sycophantic bots as sources of knowledge, rather than engaging in difficult conversations with fellow humans. Anna Mae Duane, a professor of English at the University of Connecticut, writes about the dangers she foresees in students’ use of AI companion chatbots: “I fear how humans will be damaged by the moral vacuum created when their primary social contacts are designed solely to serve the emotional needs of the ‘user.’”^[14] This anxiety can be productively extended to the information/reference space, as well – workers at the Bobst Reference Desk recounted interactions where students simply emailed them questions formatted like a prompt for a large language model (LLM), without salutations or additional context. These interactions devalue not only the expertise of the library workers, but also the entire enterprise of human co-creation and sharing of knowledge. “Human connection is supposed to require something of us,” Duane writes, and “it is worth the effort.”^[15]

Toward a More Human Reference Praxis

For current reference professionals, as well as those training to enter the field, how can we work together to preserve the humanity of information-seeking, amidst these increasingly inhuman trappings? Further, how can we continue to demonstrate the value of this practice to students who may be skeptical?

The academic reference professional stands in a unique position today, at the intersection of the security theater and generative AI. We are at a moment, to borrow a formulation from literary critic Northrop Frye, when the introduction of a new technology brings about “a corresponding increase in the amount of ritual habit necessary to life.”^[16] Before that “ritual habit” becomes broadly accepted and unconscious, there is a moment when it still feels strange and constraining: when “ritual habit can still be seen as an imminent possibility, as something we may or may not acquire.”^[17] I argue that we are at such a moment with regard to both university security theater and universal adoption of generative AI. In interactions with students, the reference professional is empowered to call attention to this moment and prompt students to think about what “ritual habits” they are being encouraged to adopt.

        NYU Libraries are already modeling these values, with their dedication to staffing the virtual Ask-a-Librarian chat almost 24/7 with human library workers. In reference queries and interviews, they also strive to embody the Reference and User Services Association (RUSA) Guidelines for Behavioral Performance of Reference and Information Service Providers,^[18] which emphasize intellectual empathy, courtesy, and inclusion – all values human library workers can uniquely employ.

        One further practice I suggest reference providers adopt is calling attention to the political, physical constraints imposed on academic freedom, where relevant. This is especially important in the context of a four-year university, where institutional memory (such as that held by library workers) often far outstrips students’ experience of the institution. Put another way, today’s undergraduate students simply might not know that such drastic security measures in a university library are not (historically) normal. If they are aware that these measures are new, they still may not be aware of the direct link to 2024 student protests. Where possible, library workers should openly discuss with students the reasons for any limitations they may encounter in their research. Library workers should also adopt this same candor and curiosity when discussing students’ AI use, drawing on RUSA’s guidance to practice intellectual empathy with patrons. Whether they choose to use LLMs or not, students should feel warmly welcomed and supported in taking the next step with their research at the reference desk.

Security and generative AI technologies are both poised to shape the information landscape irrevocably – and, left unscrutinized, they will do so. It is in these everyday interactions between library workers and patrons that we have a chance to reclaim and reshape the terms of our engagement with each other.

Acknowledgements: I would like to acknowledge NYU library workers Sam Mandani, Roxane Pickens, Alex Provo, Danny Mendelson, Phyllis Heitjan, and Amanda Watson, conversations with all of whom helped inspire and refine my thinking on these subjects. Any views expressed in this piece do not necessarily reflect any of their views. Special thanks to Sam for always taking my ideas seriously, and to Danny and Phyllis for their time working with me on the reference desk and Ask-a-Librarian chat. And finally, thank you to Gracie Anderson for the endless co-brainstorming over the years, and for the final draft review.

Author Bio:

Mac Murray is a writer, dog walker, and graduate student at NYU and LIU pursuing degrees in English and Library & Information Science.