Toward More Critical Futures of AI in Academic Libraries

Finding a middle path between AI refusal and uncritical adoption

Posted on January 12, 2026 in Blog Posts

Clarivate’s 2025 Pulse of the Library report found that 72 percent of academic libraries are exploring or implementing AI tools and technologies. In particular, many academic libraries are evaluating or deploying AI research assistants in library systems, like Clarivate’s academic AI research assistants or JSTOR’s AI research tool. 

Patrons can use AI research assistants in library systems to perform natural language search queries to discover better sources more easily than they often can by performing keyword search queries. At the same time, these tools also produce for patrons AI overviews of the sources they discover, or AI answers to questions about library materials (“what’s this text about?”).

I recommend that academic libraries should carefully implement AI technologies for discovery, like natural language search, but they should be deeply skeptical about deploying AI technologies for summarization, like AI research assistants that produce AI overviews or answers about library materials. In this blog post, I analyze AI research assistants in library systems as a case study for ultimately imagining different futures of AI in academic libraries.

I propose that we in academic libraries should spend less time debating AI refusal, and more time discerning when, where, and how AI may and may not actually support patrons. In this way, I believe that we can build together more critical futures of AI in academic libraries—beyond the uncritical adoption of AI in higher education. 

---

🔥 Stay up-to-date with LibTech Insights by signing up for our free newsletter. Just one weekly email with our new blog posts, top tech news stories, and other bonus content. Check out some posts from our archive:

• Outstanding Academic Titles 2025: Technology, AI, and Misinformation
• Invisible Sources: Transparency and Attribution as Foundations of AI Literacy Instruction
• Beyond the Job Title: Senior Digital Asset Management Specialist

---

Uncritical Futures of AI in Academic Libraries 

AI research assistants in library systems represent one kind of future of AI in academic libraries, where patrons’ primary interactions with scholarly sources will be mediated by generative AI. In this scenario, patrons will regularly encounter and produce genAI summaries of library materials that strip citations and other information from their contexts in actual sources. 

Advocates for “conversational discovery” expect that researchers will be able to use genAI tools to produce genAI answers to their research queries, instead of needing to interact with scholarly sources. Clarivate expects that implementing academic AI technologies into library systems will improve what it calls “glanceability” (a term for which the Oxford English Dictionary returns zero results). 

Researchers are more skeptical about the use of genAI technologies to summarize scholarly literature. Claire Baytas and Dylan Ruediger at ITHAKA S+R found “vastly contrasting opinions on the utility of using AI-generated summaries” for literature reviews on account of “concerns about inaccuracies.” Some researchers in STEM disciplines, for example, find genAI summaries useful for mitigating information overload. But many others are concerned about the negative impact on research quality, especially in situations where they and their peers may not be able to easily recognize inaccuracies in genAI summaries. 

Librarians are similarly concerned about the negative impact on research quality. Frauke Birkhoff writes, “Should a user accept the answer generated and quickly move on, working from the assumption that it is correct, or should that user invest considerable amounts of time and effort into making sure that the answer the tool has generated is correct? With how fast-paced academia is and how convenient these tools feel, we need to ask about the impact the tools we offer have on research quality.” 

I’m also worried that genAI overviews in library systems will habituate students to approach scholarship as a database of citations and answers, instead of a conversation among communities of researchers. When AI research assistants distill the most relevant information from the sources they retrieve to produce genAI overviews of search results, they communicate to patrons that research is abstract information-seeking, instead of a real process of collaborative inquiry toward new knowledge and ongoing scholarly conversation to socially negotiate meaning. 

Teaching students how to approach research as an open-ended, curiosity-led exploration and engagement with information (often ambiguous) from multiple perspectives is hard work; I’m worried that genAI overviews and answers in library systems will make this work harder. This isn’t the kind of future of AI in academic libraries that I want for my patrons—or that most of my patrons truly want for themselves. 

From Uncritical Adoption toward Critical Discernment 

At the same time, I don’t believe that advocating against employing AI research assistants in library systems means refusing or resisting AI. I propose that it’s simply refusing or resisting specific AI tools because they are often unhelpful, inefficient, and essentially antithetical to how we say we understand and try to teach information, research, and scholarship in academic libraries. 

Ultimately, we in academic libraires need to move from uncritical adoption toward critical discernment of AI, especially genAI. Like how advocating against employing specific AI tools doesn’t mean refusing or resisting AI, advocating for employing other AI tools doesn’t mean uncritically adopting AI. We in academic libraries need to carefully discern when, where, and how AI may or may not actually support patrons.

Natural language or semantic search tools, like at Clarivate’s Primo NDE UI or JSTOR, use gen AI technologies to convert patrons’ natural language search queries into genAI Boolean search queries to help patrons find better sources more easily than they often can by performing keyword search queries. When patrons can also use these tools to review and edit genAI Boolean search queries, like they can at Clarivate’s natural language search tool, then these tools can also help librarians teach students and researchers about effective search strategies. 

We in academic libraries need to spend less time thinking in terms of “to adopt or refuse AI?” and more time thinking critically in terms of “to adopt or refuse AI in this context and in this way?” I’m deeply concerned about the uncritical adoption of AI in higher education, but I don’t believe AI refusal represents a productive alternative. At heart, I’m both a humanist and a digital technologist (a “digital humanist,” if you will). My preference is for carefully discerning and building together more critical futures of AI in academic libraries. 

Related Posts