AI in Library Search: How Harvard Is Testing the Incorporation of AI into Search Functionalities

Chen, Carmen

Source: Dickie, T. (2025, March 11). Harvard Library search tool will understand intent behind the terms. The Harvard Gazette. https://news.harvard.edu/gazette/story/2025/03/at-harvard-library-building-a-tool-that-understands/  

Summary:

A Harvard Library team has developed a Collections Explorer tool using generative AI and semantic search technologies to help library search functions determine the intent of a search. Users will now be able to search using natural language instead of using keywords and more complex, unfamiliar search techniques. Search results include explanations of why these results were returned and are a good match to answer the initial question. It also suggests other research questions that may spark new discovery for the user, one they may have not considered before.

Thoughts and Comments:

I think this is an expected integration of AI into the modern library environment. Being able to search for items in a collection by using natural language would be a powerful ability, one that could lower barriers for search since not all library users have the ability to perform searches using complex search techniques (like Boolean search), and since not all keyword searches return relevant items due to the ambiguity of language. I wonder how this Collections Explorer tool takes into consideration these aspects of language then. It would be fantastic if even further down the line, the tool was able to accommodate for different languages. 

AI and Libraries - A Brief Discussion

Lund, B.D. and Wang, T. (2023), "Chatting about ChatGPT: how may AI and GPT impact 

academia and libraries?", Library Hi Tech News, Vol. 40 No. 3, pp. 26-29. 

https://doi.org/10.1108/LHTN-01-2023-0009

Summary:

ChatGPT has considerable and worrying advantages for those in the library profession to be able to use in the coming years. It is important to note the ways in which to use ChatGPT and other AI technologies responsibly and understand the ethical implications that may occur when abusing the new and updating technology. Utilizing GPT as well as other AI programs can save on a librarian's time throughout the day, but it should never be the deciding factor on any decisions for a library. It is a great research tool and can be used to improve academic libraries, but there are drawbacks like any tool that librarians should be aware of.

Review:

I found the article rather insightful. I don't usually utilize AI tools in my day to day just because of the risk that they pose on potentially being wrong. I do think the article was insightful on both the dangers and the positives for AI usage, specifically that of ChatGPT which tends to be the most common AI tool so far. There is always going to be bias, whether it be human or AI standards, but again that is when we don't just focus on one person or tool to be the deciding factor when it comes to decision making for a whole library population.

-Smothers, Kelsey

An early or somewhat late ChatGPT guide for librarians

 Posted by: Joe Mullins

APA Citation:

Hall, B., & McKee, J. (2024). An early or somewhat late ChatGPT guide for librarians. Journal of Business & Finance Librarianship, 29(1), 58–69. https://doi.org/10.1080/08963568.2024.2303944 

Summary:

    This article presents an up-to-date guide (this summary was written in December of 2024) of ChatGPT in respect to librarians and their work. Acknowledging the current immense usership of ChatGPT by society, Hall and McKee recognize the need for better education, workshop, and general discussion on the stalwart AI tool, particularly for information professionals. However, in this article not only do the authors advocate for research on AI and education of information professionals on ChatGPT but also highlight the importance of caution in using these tools due to limitations. Ultimately, they assert that libraries and librarians alike must stay current and share information with peers on the usages of AI as well as alert their colleagues and patrons to the softwares pitfalls and how to overcome challenges when using the tool.

Evaluation

    This article does a great job in exploring the current design of ChatGPT and advocating and calling on information professionals to become educated in the AI platform to better understand the applications uses and pitfalls. After providing a cursory overview of the popularity and usership of ChatGPT, the article then provides a great literature review of trending and noteworthy articles, workshops, and videos created by university libraries on the software – please see below the evaluation for an attached version of this reading & watching list. Following this literature review, the article then begins to provide a well written piece surveying and demonstrating the uses of ChatGPT as well as acknowledging the pitfalls of the software. They first describe some of the uses of the AI software in detail, which includes assistance in developing curricula, summarizing content, investigating websites, creating rubrics, topic ideation and brainstorming, analyzing and visualizing basic data, crafting social media posts, reviewing content, and writing code. Equally they also present the pitfalls of ChatGPT, highlighting its inability to adequately fact-check, cite sources, analyze and visualize complex data, evaluate or analyze current events, and incorporate or analyze private information. After providing this well thought review of functions and inabilities of ChatGPT, the authors then reassert their argument that it is imperative that information professionals and libraries must stay current and share information with their peers on the usages of ChatGPT – although this could be applied to all comparable AI softwares – and to educate their peers and library users on accurate information about ChatGPT, and especially its pitfalls. Overall, it is a well written and researched article that information professionals should read to ensure education on AI, and ChatGPT especially, is in their tool belt. In the current digital age washed with misinformation, an article highlighting the correct uses and detriments of AI is a valuable tool to have stored in one’s arsenal.

Attachment of Articles Reading and Watching List:

ALA Connect (American Library Association [ALA], 2023) for “ChatGPT” returns 60 “Discussion Threads”

“ChatGPT and Other AI Tools: A Guide for Faculty,” including pages for “ChatGPT/AI Use Cases & Strategies in Higher Education” 

“Readings, Podcasts, and Other Information about ChatGPT and AI” (Wayne State University Libraries, 2023).

In April 2023, the University of Washington community hosted a panel discussion, “Demystifying ChatGPT for Academics” (UW Video, 2023). 

Texas State University (Vaverek, 2023)

Stony Brook University (Kennedy, 2023)

Brown University (Morton-Aiken, n.d.)

DePauw University (Gilson, 2023)

Northern Illinois University (ChatGPT and Education | Center for Innovative Teaching and Learning, n.d.)

University of Minnesota (Matts-Benson, 2023)

New York University (Dai, 2023)

University of Georgia (Chaturvedi, 2023)

Virtual Reality (ALA Trends)

 APA Citation:

Virtual reality. American Library Association. (2017, June 16). https://www.ala.org/future/trends/virtualreality

Posted by: Federwisch, Erik

Summary:

This article provides a detailed overview from the ALA about the importance of AI in libraries in the upcoming years. While the article is from 2017, it provides several useful insights into the role AI might play in the future. Some of these reasons include providing equitable access to an emerging technology, assistance for patrons overcoming phobias, increased social interaction and implementation of new learning environments made possible through a 360 degree virtual space that users can traverse. The "Notes and Resources" section is where this article shines, though. It provides 30 resources to give librarians and library workers a better scope of the possibilities AI presents libraries.

Opinion:

While the article mostly provides an overview of AI as it applies to libraries, the resources it compiles in the "Notes and Resources" section are invaluable to understanding how you may implement VR in your own library to create better connections between patrons and resources

Link to full article:

https://www.ala.org/future/trends/virtualreality

AI in Indian Libraries: Prospects and Perceptions from Library Professionals

Posted by:

McQuiston, Regina

APA Citation

Subaveerapandiyan, A. & Gozali, A. (2024). AI in Indian Libraries: Prospects and Perceptions from Library Professionals. Open Information Science, 8(1), 20220164. https://doi.org/10.1515/opis-2022-0164 

Summary

    A 2024 study which uses a survey of 386 information professionals in a variety of institutions (“Colleges, Universities, Technical Institutes, Management Institutes, Medical Institutes, and Research Institutes”). The survey focuses on five areas: socio-demographic information about the participants, their fundamental comprehension of AI, a self-assessment of their proficiency with AI systems in libraries, Perspectives on AI Libraries, and usage of AI tools and services. Overwhelmingly, these surveys demonstrated that these information professionals recognize that AI is crucial, support its utilization as a tool, and believe that AI cannot replace librarians. The results indicate that these information professionals have a strong understanding of AI and its potential use as a tool that further enhances the capabilities of ever-necessary human librarians. Below is a figure of some of the survey's relevant findings.

Opinion

    While exploring AI’s (nearly inevitable) role in the future of libraries, I found myself wondering about Library Professionals’ experience with - and opinions of - the practical application of AI in libraries. Much of the discussion surrounding AI revolves around how AI works, what it can do, and the risks associated with its use. Many of the articles I have explored - even ones specific to the use of AI in libraries - have been centered on the technology itself, largely neglecting the perspectives of the professionals who will be expected to use it. This study, however, offered a more human-centered insight into relevant professionals’ understanding of AI as well as the way AI can be utilized as a tool to maximize library offerings. It was interesting to see such a positive outlook from information professionals regarding the use of AI as a tool to supplement human capabilities and further the advancement of libraries and information professionals.

A SWOT Analysis Approach to AI in Libraries

 

Petro, Corinne

International Federation of Library Associations and Institutions. (n.d.). Developing a Library Strategic Response to Artificial Intelligence. Retrieved from https://www.ifla.org/g/ai/developing-a-library-strategic-response-to-artificial-intelligence/

Summary:

This is a working document created by the International Federation of Library Associations and Institutions to create a strategic response to artificial intelligence (AI). First, the authors provide multiple definitions of AI, and discuss some of the ethical concerns that have arisen with the introduction of AI into the workforce. Then, the authors move to discussing how AI has impacted libraries in ways that are both “wide and deep.” There are various ways in which AI is currently used in libraries such as improving accessibility to library collections, aiding in metadata creation, supporting data scientist communities, translation, and text generation. Looking towards the future, AI’s predictive capabilities may be able to help libraries better understand user behavior and inform decision-making processes.To meet these changes, the authors stress that AI literacy is integral for both staff and patrons. In addition, libraries must think and act strategically within the context of AI’s impact on libraries by positioning themselves within institutional, sectoral, and national priorities and/or policies. The authors identify three categories of AI policies that are popular in different areas of the world: development, control, and promotion. Where a library is located will impact how it can develop and use AI. Then, the authors create a SWOT (Strengths, Weaknesses, Opportunities, Threats) for AI in libraries. One of the strengths they found is that libraries are trusted sources that have values like access to knowledge and privacy protection. These values could counteract some of the challenges that come with AI. One of the weaknesses is that commercial AI products can be costly and many libraries have limited budgets and limited technical capacities. One of the opportunities is that AI automates routine tasks and aids in professional tasks. Lastly, one of the threats is that there is a lot of fear surrounding AI, both in the public and in libraries. The authors move on to name ten institutional approaches to AI in libraries. These approaches involve hiring new staff that are already familiar with AI, upskilling current staff to be competent in AI, engaging with the library’s user base to see how they are using AI, studying best practices, etc. The authors state that some of these institutional approaches may be combined. They also list the pros and cons for each institutional response. Lastly, the authors outline what they believe to be the three most important strategies for libraries today. The first is using library AI capabilities to model responsible and explainable applications of descriptive AI. This includes using descriptive AI to improve the description and retrieval of library collections, ensuring usability and explainability of AI tools used in a library, documenting AI related projects thoroughly and openly. Some challenges with this include prioritizing collections, solving conceptual issues, and establishing sustainable services. The second strategy is using librarians’ data competencies to enhance organizational AI capability. This strategy includes using librarians’ data expertise to support data scientists and future AI applications. Activities include finding data sources, promoting data sharing and standards, and ensuring data quality and compliance. The third strategy is promoting AI literacy to enhance organizational and societal AI capabilities. This strategy includes libraries taking the lead in promoting AI literacy, especially in educational and public settings. Challenges with this strategy include librarians needing to understand the complexity of AI, including algorithmic literacy. Librarians will also have to grapple with the opacity of AI in infrastructure, especially Big Tech.

Opinion:

I found this article to be helpful in understanding what librarians can do now to facilitate the integration of AI into libraries. There are many articles on this topic that theorize a lot and do not offer much in the way of solutions beyond general advice. Although AI integration into libraries is still very new, the authors of this piece provide multiple strategies that librarians can use depending on their area’s relationship with AI. I liked that the authors used a SWOT analysis to break down this issue. Overall, I like how this article offered multiple approaches to a complex problem. Each approach was unique which allows for librarians to assess their current position in the AI landscape, and adopt whichever approach works best for them.

How AI might change academic library work

Hannah Prince



Cox, A. (2023). How artificial intelligence might change academic library work: Applying the competencies literature and the theory of the professions. Journal of the Association for Information Science and Technology, 74(3), 367–380. https://doi.org/10.1002/asi.24635


This article focuses on how AI can be applied to knowledge discovery. It analyzes eleven different potential approaches libraries might adopt - and these “applications are analyzed and their likelihood evaluated”. The paper also assesses how outside factors might influence how AI is adopted by those in academic libraries.

The paper covers a variety of topics, including analyzing the “theory of professions” and debating different iterations of what they call, “hybrid professional” roles - where a librarian can use AI to meld the skills of a librarian, a researcher, and an archivist all into one role. I did appreciate the multiple table examples used throughout the paper, it was a useful representation of data and helped to communicate the range of approaches that people are taking when incorporating AI into their profession.

Despite initial pessimism, this paper comes to the eventual conclusion that AI is inevitable in libraries, but doesn’t necessarily have to come at the expense of librarians' jobs. In fact, AI can integrate into librarians’ roles and the author goes as far as to say that, “it might be considered fair to be optimistic, therefore, that librarianship will adapt, even be strengthened through this process.” Increasingly we see librarians playing a hybrid role that is part managerial, part customer service, part academic - but often less of the “classically trained librarian”. AI can be incorporated into that hybrid role better than into the classical librarian role. As with most think pieces about AI, it summarizes ways that the library industry will be irrevocably changed by AI, but professes that the individuals in the industry will adapt rather than be wholly replaced. I did not think that this article was necessarily groundbreaking or saying anything new, but I liked how they approached the subject and how the author broke down the different sections. And as mentioned previously, I thought the use of tables was super helpful to get the author’s point across.

Introducing Chatbots in Libraries

Bachofer, Mark

McNeal, M. L., & Newyear, D. (2013). Introducing chatbots in libraries. Library Technology Reports, 49(8), 5-.

Summary:

For my Presentation Five in this class, I set out to explore the feasibility of utilizing artificial intelligence as a chatbot to answer the questions of digital patrons. While I’m definitely not the biggest fan of AI, especially in creative endeavors, at the branch where I work, two-thirds of my workdays are spent manning the ASK inbox. For the majority of the emails that we receive, I am able to cut and copy a preprepared response, and send it back out. In my mind, this would be an ideal use for such technology – scanning incoming emails, recognizing certain keywords, and deploying the premade advice. But before I fully condoned the use of AI chatbot technology, I wanted to make sure that I understood where it came from.

In Introducing chatbots in libraries, McNeal provided exactly the resource I was looking for. This article discusses the history of chatbots, and their more recent implementation in libraries. I was surprised to find that some branches have been using them since the mid-2000s, but given that the technology has existed since 1966, I suppose that’s not too unreasonable.

The first chatbot was designed as a therapist. Specifically, ELIZA acted as a Rogerian therapist, which reframe statements are questions to then induce further introspection, and allow the individual undergoing therapy to control the conversation. This type of conversation seemed ideally suited to a chatbot, as they specialize in recognizing keywords, and providing outputs based on those triggers. However, it also seemed ideal for a library scenario.

Indeed, many libraries have adopted chatbots – new systems include Emma the Catbot in Mentor, Ohio and Pixel at the University of Nebraska-Lincoln – but their overall purpose remains the same: to allow branch staff to get back to the desk. Interpersonal contact is one of the key duties of library staff, and if AI is able to increase the time that we are able to spend interacting with the public, I think that it is worth implementing.

Evaluation:

I very much enjoyed this article. It gave a solid history of chatbots, and was able to uncover the underpinnings of this technology. It’s important to understand where new technologies come from, how they’ve developed, and how they can be used in the future.

I did notice that there was not much discussion about the personal privacy issues of AI chatbots, which is still a concern I have. I think that while implementation of such a system in libraries is a good idea that could free up a lot of staff time, we will have to take precautions to avoid distributing people’s information.