Use of Generative AI in Aiding Daily Professional Tasks: A Survey of Librarians’ Experiences

Abstract

This study examines how librarians are using third-party generative AI (GAI) tools such as ChatGPT to aid their daily professional tasks. An online survey of 272 librarians found that text-generating AI tools were the most popular. The majority of respondents felt that GAI tools were effective in improving productivity. Key challenges included ensuring content accuracy and designing effective prompts. Top suggestions for better preparing librarians to use GAI include practical training on using GAI, establishing AI policies and guidelines, fostering collaboration and communities of practice, and providing access to useful GAI resources. The study highlights popular use cases that can inform professional development, while underscoring the need for hands-on training, institutional policies, opportunities to experiment with GAI, and access to enhanced tools. As GAI evolves, supporting librarians’ adoption will be crucial for harnessing its potential benefits.

Introduction

Artificial intelligence (AI) enables intelligent machines to work and react more like humans via deep learning, machine learning, and natural language processing (American Library Association 2019). Generative AI (GAI) refers to a type of AI technology that has the capacity to autonomously generate new material, including text, image, audio, and video (Feuerriegel et al. 2023). In recent years, GAI has been recognized as a transformative force not only having an impact on the daily lives of individuals but also exerting consequences for organizations (Cardon et al. 2023). It is increasingly being applied in a variety of disciplines as it aids and accelerates the content production process (Morris 2023). Organizations of all sizes are actively exploring the use of GAI to increase efficiency and address productivity challenges (Al Naqbi et al. 2024). The library world is no exception. The literature abounds with discussions about the diverse ways GAI can be integrated into library operations and services to improve information organization and retrieval, automate routine tasks, enhance user experience, and stimulate innovations.

Thus, it is crucial for librarians and information professionals to stay informed and involved as they navigate the AI landscape. In his interpretation of the US Department of Education’s AI report, Lo (2023) opines that librarians must become closely involved in AI implementation, promoting AI literacy, developing guidelines for AI use, preparing for AI issues, and collaborating with other stakeholders in order to fully harness the technology’s transformative potential. To better understand librarians’ preparedness in embracing this AI wave, this study seeks to examine librarians’ adoption of GAI in their daily professional practice by answering the research question “How do librarians use third-party generative AI tools in aiding their daily professional tasks?”

This study explores the role of GAI tools that are either stand-alone applications such as ChatGPT or embedded in existing technologies such as Grammarly in assisting librarians in their professional work. The focus is on investigating individual librarians’ own efforts in experimenting with and employing these tools to improve their productivity at work. Findings of the study will further the understanding of how to optimize GAI use among librarians, inform professional development initiatives by identifying areas where librarians may require additional training or support to maximize the benefits of GAI tools in their work, and guide future decisions about the integration of GAI tools in library settings.

Literature Review

AI and Productivity

GAI’s positive impact on productivity is well documented in the literature. In a survey among early adopters of ChatGPT, Skjuve et al. (2024) summarize six motivations for using GAI, and productivity is the top one on the list, followed by novelty, creative work, learning and development, entertainment, and social interaction and support. In Cardon et al.’s (2023) study of employee perspectives on ChatGPT benefits, they found that many respondents used ChatGPT for professional purposes—about 42 percent used it for researching a topic or generating ideas; 32 percent, for drafting messages; 26 percent, for drafting longer documents; and 22 percent, for editing text. The majority believed that GAI can make them more productive, such as helping them generate ideas for work, communicate more effectively, and improve the quality of their work. Brynjolfsson et al. (2023), in their study of 2,179 customer support agents’ job performance after the staggered introduction of a GAI-based conversational assistant, found that access to this tool increased productivity, as measured by issues resolved per hour, by 14 percent on average, including a 34 percent improvement for novice and low-skilled workers. They conclude that AI technologies enable the dissemination of the best practices of more able workers and help newer workers improve and that they ultimately lead to increased customer satisfaction and employee retention. In their study of one hundred of the largest publicly traded US companies, Yu and Qi (2023) compared changes in firms with high GAI exposure with those in companies with low GAI exposure before and after the launch of ChatGPT and found that GAI created positive and statistically significant effects on labor productivity. Noy and Zhang (2023) conducted an experiment to assess ChatGPT’s impact on productivity. They assigned writing tasks to 453 college-educated professionals and randomly exposed half of them to ChatGPT. Their findings show that ChatGPT substantially raised productivity, in that the average time taken decreased by 40 percent and output quality rose by 18 percent. Similarly, in the creative realm, Doshi and Hauser (2023) studied the causal impact of GAI on the production of a creative output in an online experiment, and they conclude that access to GAI led to an increase in the writers’ creativity, with stories being evaluated as better written and more enjoyable, especially among less creative writers. Echoing the empirical evidence presented in the aforementioned studies, Al Naqbi et al. (2024), in their literature review of GAI applications in eight separate fields, found that research publications in fields ranging from engineering to agriculture all emphasize significant benefits of GAI, highlighting increased efficiency and productivity.

AI Application in Libraries

Libraries have been experimenting with integrating AI into library operations and services in a variety of ways. Information organization and discovery is the area where most library AI experiments have occurred. Coleman et al. (2022) applied machine learning in classifying images in order to improve the accessibility of collections. Harper et al. (2022) used machine learning to make the discovery of theses and dissertations more efficient. Milholland and Maddalena (2022) attempted to enhance the discoverability of metadata in an institutional repository with the support of robotic process automation. The National Library of Sweden developed an AI-based language model for its collections that enabled automated classification, enhanced searchability, and improved optical character recognition cohesion. Researchers can use the AI model to build data- sets of various Swedish materials such as books, newspapers, radio and TV broadcasts, internet content, Ph.D. dissertations, postcards, menus, and video games to meet their specific needs (Haffenden et al. 2023). Ahmed et al. (2023) report their design of an open-source AI environment named Annif and how it was used in indexing large volumes of documents. Their experimentation indicates that the convergence of data carpentry and AI/machine learning–based knowledge processing may lead to breakthroughs in designing systems for the autogeneration of class numbers and proper subject descriptors for large datasets. Brzustowicz (2023) explores Chat-GPT’s application in assisting cataloging and demonstrates that it could generate accurate machine-readable cataloging (MARC) records using Resource Description and Access (RDA) and other standards such as the Dublin Core Metadata Element Set, highlighting its potential as a tool for streamlining the record creation process and improving efficiency in library settings.

With the rapid growth of GAI in recent years, libraries’ AI experimentation has become more diverse. Zayed University Library in the United Arab Emirates implemented a custom chatbot using the GAI tool Chat-GPT to provide reference services to students and faculty outside the library’s regular operating hours. This endeavor revealed that generative chatbots could be more accurate and efficient than traditional rule-based or retrieval-based chatbots, exhibiting great potential for providing personalized, accessible, and cost-effective support to library users (Lappalainen and Narayanan 2023). Librarians at Palo Alto Library tested the search engine optimization potential of ChatGPT plug-ins and used Chat-GPT to analyze catalog usage data in order to produce taxonomy terms better aligned with library users’ interests and ultimately provide meaningful and relevant personalized promotions for library users (Hess and Markman 2023). These experiments confirm AI’s efficiency, adaptability, and capacity in solving a wide range of library problems and accelerating project deliverables. Gupta and Gupta (2023) present a case study of using AI-generated videos to promote library events on social media. In this case, a librarian created short videos about entrepreneurship events organized by the library in collaboration with external experts and posted them on the library’s social media platform. As a result, user engagement improved, and user interest in entrepreneurship events grew, which encouraged future explorations of using AI-generated multimedia content in marketing and outreach.

While the empirical evidence reported in the literature supports AI’s potential in enhancing library services and operations, it is widely recognized that a host of challenges need to be carefully considered and addressed in AI adoption, such as cost, data privacy and bias, and equitable access to information (Cox and Tzoc 2023; Michalak 2023). Critical factors, including the innovation environment, infrastructure, budget, leadership focus, and librarians’ prior experience with AI applications and attitudes toward them, will also influence libraries’ adoption of AI (Huang 2022; Hussain 2023; Pence 2022).

Librarians’ Awareness, Perception, and Use of AI

Before the widespread attention to GAI, Wood and Evans (2018) surveyed academic librarians about the impact of AI on the library profession and found that most librarians had not read about AI in the professional literature and that only 47 percent of respondents were interested in attending a workshop on AI in the library. A few years later, in Yoon et al.’s (2021) study about how AI technologies are perceived by librarians, the percentage of librarians interested in AI-related training increased to 68 percent, suggesting a growing need for training and education for librarians on AI. Drawing on their experience with GAI adoption at Palo Alto Library, Hess and Markman (2023) identify four skills that library staff should develop to prepare for the inevitable AI reality: (1) maintaining freshness and providing “just in time” information when working with GAI; (2) developing prompt engineering skills, which requires writing lots of bad prompts and analyzing those results; (3) understanding how to conduct the “Reverse Reference” Interview to better converse with GAI; and (4) having data remediation skills to prepare data for use with GAI.

With the rapid development of GAI, librarians and information professionals need to stay informed, engage in practical training, and commit to continuous learning about AI (Williams 2023). One approach to do so is to constantly experiment with using the technologies and evaluate and investigate them to explore their potential in various aspects of library work (Panda and Kaur 2023; Pival 2023). Hosseini and Holmes (2023) spoke with eight librarians about GAI and found that they had been experimenting with GAI tools such as ChatGPT in the following ways: brainstorming for lessons and helping draft scripts for videos, generating examples that would be understandable for students or researchers at a certain level, summarizing text from a critical appraisal perspective, enhancing coding capabilities, and improving survey design. Hosseini and Holmes point out that “those who fail to learn about and incorporate these tools in their daily workflows may find themselves outperformed by those who have embraced it” (2023, 840).

While librarians recognize the potential benefits of AI applications in libraries, they are also aware of the challenges associated with the technologies and caution that a critical perspective is needed when using AI (Hosseini and Holmes 2023). Meanwhile, some librarians have expressed concerns about AI replacing human intelligence within libraries and hence potential job losses resulting from AI adoption (Ajani et al. 2022; Subaveerapandiyan and Gozali 2024). This sentiment tends to be more prevalent among librarians in developing countries (Yoon et al. 2021).

The literature review has established that AI can have a positive impact on productivity and enhance library services and operations, and it is critical for librarians to receive AI-related training and education and stay informed about the development of the technologies. Thus, more empirical research is needed to examine librarians’ experience in experimenting with AI in their daily workflow and their preparedness for AI adoption. This study was prompted by this need and seeks to investigate how librarians use third-party generative AI tools in aiding their daily professional tasks and improving productivity.

Methodology

Librarians who work full-time in a library and have used GAI tools in their professional practice were considered to be the study population. Individual members of this population are not identifiable via any sampling frame, which rules out the possibility of a probability sampling design. Thus, a nonprobability sampling technique, judgmental sampling, was used to select a sample from the population. Judgmental sampling is a type of nonprobability sampling in which the study participants are selected on the basis of the researcher’s judgment about which ones will be the most useful or representative (Babbie 2012). An educated judgment was made that librarians are likely to be subscribers to professional email listservs in their respective domains. Thus, study subjects were recruited from the following listservs covering different areas of librarianship:

• • Government documents and government libraries: GOVDOC-L

• • Reference and information services: LIBREF-L

• • Public libraries: PUBLIB-L

• • Collection development: COLLDV-L

• • Electronic content licensing for academic and research libraries: LIBLICENSE-L

• • Serials in libraries: SERIALST

• • Electronic resources: ERIL-L

• • Library acquisitions: ACQNET

• • Medical libraries: MEDLIB-L

• • Recorded sound archives: ARSCLIST

• • Association of College and Research Libraries (ACRL) Discussion Lists on ALA Connect

An online survey was employed as the data collection instrument. A message containing the study introduction, consent information, and a link to the survey was sent to the above listservs. The survey was launched in March 2024 and remained open for three weeks.

The main variable in the study, “librarians’ use of third-party GAI tools to aid their professional tasks,” was measured by three principal indicators: the ways in which librarians used text-generating, image-generating, audio/voice-generating, and video-generating AI tools to assist them at work; librarians’ perception of the effectiveness of GAI tools in improving their productivity; and challenges/concerns experienced by librarians in their use of GAI tools. In addition, demographic information was collected about respondents’ work setting and length of work experience. Finally, suggestions were elicited from respondents on how to better prepare librarians for adopting GAI technologies.

Descriptive statistical analyses were employed to examine responses to close-ended questions. Responses to open-ended questions were inductively and thematically coded, and patterns discerned from the idiosyncratic responses are presented in the analysis.

Results

A total of 272 survey responses were received. The majority of the respondents were academic librarians (75.7 percent), followed by medical librarians (7.1 percent), public librarians (6.2 percent), law librarians (2.9 percent), special librarians (2.4 percent), museum and archive professionals (2.4 percent), government librarians (1.9 percent), and library consortium staff (1.4 percent). Most of the respondents had relatively longer work experience as a full-time librarian—33.3 percent had more than twenty years of experience, and 35.7 percent had eleven to twenty years of experience. Less than one-third had only worked for five to ten years (15.2 percent) or less than five years (15.7 percent) as a librarian.

Nonuse of GAI Tools

Out of the 272 respondents, 181 used GAI in their professional practice, and eighty-one did not. The nonusers were asked about their main reason for not using GAI, and as shown in table 1, close to 40 percent of the respondents indicated a lack of interest in using GAI tools.

Use of GAI Tools

Among the 181 GAI users, 79.1 percent were academic librarians; 5.0 percent, medical librarians; 4.3 percent, law librarians; 3.6 percent, public librarians; 2.9 percent, library consortium staff; 2.2 percent, government librarians; 2.2 percent, special librarians; and 0.7 percent, museum and archive professionals. In terms of work experience, 35.3 percent had eleven to twenty years of work experience as a full-time librarian, 33.1 percent had more than twenty years,15.8 percent had five to ten years, and another 15.8 percent had less than five years.

A total of 165 GAI users used text-generating AI such as ChatGPT or Gemini to assist them in their professional practice. Table 2 presents the variety of ways in which they used these tools. Particularly, respondents used such tools to generate and edit a multitude of content types at work, the most popular use being generating reports, manuscripts, or presentation slides. In addition, respondents generated content to support library services such as reference and information literacy instruction. As one respondent commented, “[I use the tools to] create examples to use in class activities that students then critique, to generate keywords from research questions and to then turn those keywords into search strings.”

Table 1.

Respondents’ main reason for not using generative AI (GAI) in their professional practice.

Table 2.

Specific ways of using text-generating AI tools.

Regarding image-generating AI such as DALL-E and Midjourney, only fifty-nine respondents reported using it in their professional practice. Table 3 showcases the different ways these respondents used image-generating AI tools. Close to 60 percent of them generated images to use in instructional/teaching materials, representing the most popular type of use of image-generating AI tools.

Table 3.

Types of use of image-generating AI tools.

Table 4.

Types of use of AI features embedded in existing tools.

The number of respondents who used audio/voice-generating AI such as Stable Audio and PlayHT sharply declined; merely six people used such tools to assist them at work. Among them, five generated audio/voice for instructional/teaching materials, three generated reports or presentations, and one generated marketing/outreach materials. Video-generating AI such as HeyGen and Synthesia was the least used among the respondents. Only three respondents used these tools, generating video to use in instructional/teaching materials and in reports or presentations.

In addition to the stand-alone GAI tools, eighty respondents also used AI features embedded in existing tools. Table 4 provides a breakdown of the different types of use among these respondents. Close to three-fourths used AI-powered writing assistance applications such as Grammarly, a number far higher than the number of users of other types of AI-powered tools.

Perceived Effectiveness of GAI Tools

Respondents were asked to rate the overall effectiveness of GAI tools in aiding their daily work as a library professional on a ten-point scale, where 0 represents “not effective at all” and 10 represents “extremely effective.” The average rating was 6.76, and the median was 7.00, suggesting a generally positive view of GAI tools’ efficacy among the respondents.

Table 5.

Methods of handling the cost of generative AI tools.

Handling the Cost of GAI Tools

In terms of how respondents handled the cost of GAI tools, almost 70 percent used free tools, as shown in table 5. It is also worth noting that 11 percent of the respondents personally paid for GAI subscriptions.

Concerns and Challenges Encountered in GAI Use

When asked about the concerns or challenges they had encountered when using GAI tools to assist them at work, 105 respondents shared their thoughts. Table 6 presents the top six categories of these concerns and challenges. Most fell under the top three categories—more than 40 percent of the respondents felt that AI could not consistently generate accurate and reliable information and often requires verification, followed by one-fourth feeling concerned about the ethical use of AI and issues related to data privacy and security and about one-fifth voicing their frustration in creating prompts to effectively communicate with GAI. In addition to the top categories shown in table 6, respondents also mentioned the following concerns and challenges: libraries not being able to afford paid versions of GAI tools with more functionality (3.8 percent), paywall barriers (3.8 percent), ensuring equitable access to GAI tools (3.8 percent), the stigma associated with using GAI (2.9 percent), loss of critical-thinking skills (1.9 percent), and threats to librarian jobs (0.9 percent).

Preparing for GAI Adoption in Professional Practice

The survey also asked how librarians can be better prepared to adopt GAI in their professional practice. A total of ninety-three responses were received. The following themes arose from these responses:

• • Training and education (55.9 percent). It is important to provide free or affordable training and education opportunities (e.g., tutorials, webinars, workshops, certificate programs) to help librarians understand GAI models and learn the prevalent GAI tools and specific skills to effectively use them to improve productivity at work. As one respondent commented, “Most education for librarians up until recently has been more of a bracing for impact exercise, and it needs to become more technical and task-oriented so we can learn actual skills.” Particularly,

Table 6.

Concerns and challenges encountered by respondents when using generative AI (GAI).

• • respondents wanted to receive training on prompt design and see successful examples of specific GAI uses in librarians’ daily work.

• • Hands-on practice and experimentation (33.3 percent). Exploring GAI tools through hands-on practice is a quick way to learn about these tools. As shown in the comment, “The information is coming fast and there are so many small things that librarians can do to prepare themselves and their patrons. They don’t need to know everything, but they need to try these tools. Both big and small asks. Let ChatGPT write your paper, and see what our patrons are seeing,” experimenting with GAI technology can start with small attempts and gradually building one’s knowledge about these tools.

• • Institutional policy and support and professional guidelines (24.7 percent). It will be helpful for professional organizations for librarians to provide guidelines, such as those of the ACRL AI Competencies for Library Workers Task Force, that could guide training and education. Meanwhile, librarians will benefit greatly if their institutions provide support for their AI-related professional development and have clear policies in place to inform librarians’ AI use.

• • Understanding how to provide services to library users in the age of AI (19.4 percent). It is necessary for librarians to be aware of how patrons are using AI. They should be able to explain AI and teach it to patrons. Librarians should also support users in their AI-related activities. Additionally, they need to ensure that AI tools are accessible to everyone through library services.

• • Critical thinking and ethical use of AI (14.0 percent). Issues related to critical and ethical uses of GAI technology need to be discussed so that librarians are fully aware of them as they adopt these tools.

• • Collaboration and community of practice (12.9 percent). Collaboration among librarians and the creation of a community of practice can be useful in sharing knowledge and experiences with AI. Specific methods include doing group reviews of AI tools, joining listservs, participating in AI-focused professional committees and task forces, attending AI-themed conference sessions, and exploring relevant LibGuides and literature. Librarians are also encouraged to collaborate with other members of their institution (e.g., other units on campus) to explore the use of GAI.

• • Attitude adjustment (6.5 percent). As one respondent commented, “The hand-wringing about ethics is warranted but not useful.” GAI has become the reality, and a shift in attitude toward AI is unavoidable.

• • Access to useful GAI resources (5.4 percent). Resources such as prompt libraries, a list of free GAI tools, and systems that provide access to aggregated GAI tools would be beneficial in librarians’ GAI exploration and experimentation.

• • Keeping abreast with rapidly developing AI technology (5.4 percent). Although it is challenging, librarians have to keep up with the rapid pace of AI development.

Discussion

The results of this study provide valuable insights into how librarians are adopting and using generative AI tools in their daily professional tasks. With a high adoption rate of 66.5 percent among respondents, it is clear that many librarians, particularly in academic settings, are embracing these new technologies. The findings reveal a diverse range of GAI applications in library settings, from text generation to research support, demonstrating the versatility of these tools. The positive perceptions of GAI’s impact on productivity align with studies in other sectors, suggesting potential long-term implications for library operations. The study also highlights important challenges and concerns that need to be addressed as libraries continue to integrate GAI technologies. By analyzing these results in the context of the existing literature and considering their implications for library practice and policy, this discussion aims to contribute to the ongoing conversation about the role of AI in libraries and guide future research and professional development initiatives in this rapidly evolving field.

GAI-Related Professional Development for Librarians

Overall, librarians in this study had used GAI technology and found it effective in aiding their daily work as a library professional. Librarians had used tools such as ChatGPT, Perplexity, and Gemini to generate content and assist them at work in myriad ways. At the same time, study findings indicate that many librarians would like training and education that focuses on the practical aspect of GAI applications and the establishment of communities of practice to support GAI learning.

Use cases from librarians who are using GAI effectively can be a valuable source to support GAI-related professional development for other librarians. Professional conferences may consider offering sessions dedicated to GAI and productivity where presenters demonstrate their GAI use at work; journals may create columns to invite submissions from librarians where they discuss and reflect on their specific uses of GAI; and libraries can organize brown bags or “lunch and learn” sessions for people to share their experiences using GAI to assist them in completing professional tasks. Specific case studies can provide librarians with a better understanding of GAI in real-world contexts and the potential benefits and limitations of GAI. As more and more use cases are shared, librarians will gain insights into best practices for adopting GAI. For example, in this study, content accuracy and prompt design were identified as two major challenges in librarians’ GAI use. Seeing how these are addressed by other users would enable librarians to learn from the successes and failures of others, avoiding common pitfalls and focusing on strategies that have proved effective. Furthermore, studying different examples of GAI use can help librarians think creatively about how they can integrate AI into their daily tasks and broader library services. It can spark ideas for new services, programs, or workflows that leverage GAI to improve efficiency and user engagement.

Building a repository of detailed case studies that librarians can access and learn from is also worth considering, as it can further help demystify GAI and showcase its practical applications. For example, a repository of “GAI Success Stories and Challenges” could provide specific examples, such as using GAI for grant writing, improving user engagement through personalized recommendations, or managing large-scale data analysis projects. An “Interactive Case Study” database could allow librarians to browse through categorized examples, filter by library type or GAI tool, and even contribute their own experiences.

Ultimately, as the sharing continues, it might lead to networking and collaboration opportunities, allowing librarians to connect with peers and explore partnerships for GAI-related projects. Libraries can form consortia to codevelop AI solutions or share access to GAI tools, spreading costs and benefits across multiple institutions. Opportunities may also arise for libraries to collaborate with AI developers, universities, or other sectors to cocreate tools specifically tailored to library needs, ensuring that solutions remain practical and relevant.

Ethical Use of GAI

More than one-fourth of the librarians in the study expressed concerns about the ethical use of GAI, data privacy, and security. They pointed out the need for their institutions to establish meaningful AI policies to guide their AI use, as shown in this comment: “All employers should have policies on how tools are to be used for work.” This echoes findings from Cardon et al.’s (2023) study where employees in organizations with GAI policies had positive views of those policies, suggesting that such policies support more comfort in using GAI for work, improve trust and efficiency, and provide legal protections. Recognizing the necessity and importance of institutional AI policies, Michalak (2023) proposes the Academic Librarian Framework for Ethical AI Policy Development, calling for librarians to be actively involved in their institutions’ development of policies that are comprehensive and ethical and prioritize social responsibility and respect for human rights. Librarians may use their expertise to help their institutions define the core values and ethical principles to guide AI policies and collaborate with stakeholders from diverse backgrounds, including ethicists, legal experts, technologists, business leaders, and representatives from marginalized groups to oversee the development, implementation, and revision of AI policies. If possible, librarians may consider actively participating in their institution’s AI governance, for instance, advocating for librarian representation on AI ethics or governance committees at the institution to ensure that library concerns are considered in broader AI policy decisions and engaging in ongoing discussions to review and update AI policies based on new developments, emerging risks, and lessons learned from real-world use.

Institutional Support for GAI Experimentation

Institutional support is another important element in preparing librarians for the adoption of GAI at work. In addition to providing financial support for librarians to participate in AI-related training and education programs, institutions may also offer opportunities for librarians to gain hands-on experience. One-third of the librarians in the study identified hands-on practice and experimentation as a pathway toward AI preparedness. Thus, institutions may encourage a culture of experimentation and launch small-scale AI projects where librarians can test new tools and approaches in a controlled environment. For instance, several librarians expressed their interest in experimenting with customizing GAI with library data, and this could be a viable pilot project that libraries could invest in. A supportive infrastructure and access to GAI tools would allow librarians to explore GAI technology with more flexibility and confidence. The majority of librarians in the study only used free GAI tools, suggesting that they could benefit from institutional subscriptions to tools with enhanced features and functions. One respondent mentioned that “Harvard’s sand-box, a university wide wall-off access to seven selected AI tools, is a good example of best practice,” suggesting Harvard’s AI Sandbox as an exemplary institutional support for GAI exploration in a secure environment that mitigates many security and privacy risks.

Meanwhile, as librarians gain experience with GAI, institutions should ensure that successes and insights are shared across the organization. For instance, knowledge-sharing platforms can be established for librarians to document and share the outcomes of their AI experiments, including case studies, lessons learned, and recommended best practices. This allows others to replicate or build upon successful initiatives. Internal showcases and demonstrations can be regularly organized for librarians to display the AI projects they have worked on, allowing their colleagues to learn from their experiences and explore new possibilities. Forming learning communities within the institution should be encouraged so that librarians can exchange ideas, seek advice, and collaborate on AI-driven projects.

Getting More Librarians to Explore GAI

Among the librarians that did not use GAI technology, almost 40 percent lacked interest in using GAI tools, making it the top reason for nonuse. This finding highlights the importance of not only more GAI exposure among librarians but also a positive culture around AI exploration adoption both in the profession and in librarians’ institutions. The literature has revealed a consensus that

generative AI is making a major impact on our work and lives to the point that working and collaborating with generative AI will soon become a norm, if not already a norm. Continuous learning and adaptation are necessary to upskill, reskill, and retool the workforce as AI continues to advance and redefine our workplace and our lives.

(Nah et al. 2023, 296)

In order for librarians to engage in effective learning and adaptation, a positive culture is indispensable in shifting their attitude and embracing GAI technology. Professional organizations and institutions can foster such a culture by acknowledging and celebrating successful AI projects and initiatives led by librarians, providing clear communication to address fears and misconceptions about AI, and promoting a mindset that values creativity and innovation when exploring new AI applications.

Conclusion

This study provides a detailed depiction of how librarians currently use GAI tools in aiding their professional practice and to improve productivity at work. This information can be used by library administrators, policy-makers, and technology developers to prioritize investments and developments that align with the needs of librarians and their communities. Library educators may also use this information to design meaningful and up-to-date curriculum for library school students. Building on this study, future research may continue assessing librarians’ professional development needs related to GAI, identify best practices, conduct comparative studies with other professions that have adopted AI to examine how they approach professional development, evaluate the efficacy of AI training/education programs, and analyze job descriptions for librarians to identify the AI-related skills and knowledge increasingly sought by employers.

Disclaimer

Before submission to Library Trends, the author used Perplexity, a generative AI tool, to proofread this manuscript.