Quick Poll Results: ARL Member Representatives on Generative AI in Libraries

Last Updated on May 9, 2023, 12:40 pm ET

As generative AI technologies continue to evolve, their potential to impact library services, operations, and user experiences is becoming increasingly apparent. We conducted a quick poll of Association of Research Libraries (ARL) member representatives in April 2023 to gather insights into their current perspectives on generative AI adoption, its potential implications, and the role of libraries in AI-driven environments. In this blog post, we summarize, synthesize, and provide recommendations based on the survey responses, aiming to offer valuable insights for senior library directors navigating the AI landscape.

We shared the survey link with ARL member representatives in April 2023; 19 ARL members completed the survey.

Key Findings

1. Perspectives on Generative AI: Most ARL member representatives who responded to the poll have a somewhat positive (53%) or very positive (11%) view of the potential of generative AI in enhancing library services in the next 12 months. However, they also acknowledge the limitations and challenges associated with these technologies. The remaining 37% are neutral on the potential of generative AI.

2. Adoption Stages: While some libraries are actively implementing generative AI solutions (11%), many are exploring potential applications (32%) or considering adoption in the near future (32%). A few libraries have limited interest (16%) or no plans to explore AI technologies (11%) in the next 12 months.

   

3. Impact on Library Operations and Services: Library directors anticipate that generative AI will have a transformative impact on library operations and services, including metadata generation, research discovery tools, and user services. They also foresee potential challenges related to misuse, biases, and copyright.

   

4. Institutional and Organizational Conversations: Libraries are engaging in organization-wide discussions on generative AI’s impact on academic integrity, publishing, authorship, and research integrity. They are participating in interdisciplinary forums, working groups, and collaborations with various institutional partners.
5. Information Literacy: Survey responses suggest that libraries can enhance information literacy programs to help users understand and evaluate AI-generated information through collaboration, interdisciplinary partnerships, and incorporating AI literacy into broader information literacy. Emphasizing librarians’ knowledge of AI, developing new AI-related literacies, and retraining staff and offering workshops are opportunities for libraries to exert leadership as research institutions navigate the AI era.

Specific ideas reflected in the quick poll results include:

• Produce collaborative guides and/or briefings on generative AI, targeted to researchers and the broader community.
• Leverage AI to aid in the identification of misinformation.
• Emphasize the importance of critical thinking to researchers, including community members, faculty, and staff, so they are able to assess the source of information and its reliability. Affirm how the information is generated (AI or otherwise) is less important than the ability to recognize what is reliable, what is not, and how those decisions are made.
• Collaborate with colleagues in teaching and learning as well as in research areas to share experience and develop joint initiatives focused on educating students and the community regarding effective and responsible use of generative AI tools in learning, research, and community engagement.
• Work with IT security/privacy teams to teach how to detect and evaluate AI-generated content and how to use this and other information responsibly, especially within public research libraries.

Opportunities for Collective Library Learning and Information Sharing

While we’re in the early stages of understanding the applications of this technology, and how law and regulations govern it, here are some opportunities for ARL members to consider within libraries, across their organization, and collectively across research libraries.

Exploring potential applications and uses for generative AI in library services is an opportunity that research libraries have undertaken since ChatGPT launched late last year. Last December, the University of Delaware held a half-day research sprint among library staff who support research and curriculum to collaboratively learn about large language models and explore the use and potential challenges of generative AI in library services. During this research sprint, the team explored Elicit, Chat GPT, ResearchRabbit, Scholarcy, and Paper Digest. After putting each tool through its paces and articulating its strengths and weaknesses, the team discussed where they feel their expertise fits into this landscape, and how they will need to support researchers as these tools gain greater adoption. At the University of New Mexico (UNM), in addition to a university-wide panel on “Teaching in the Age of AIs” held in March 2023, a task force of librarians and staff across departments will be convened this summer to experiment with GPT-4 and the value the paid version may bring to library services. The initiative at UNM consists of three phases: Introduction and Training (two weeks), Exploration and Experimentation (eight weeks), and Evaluation and Sharing (two weeks). The GPT-4 Exploration Group will produce a report detailing project summaries, participant feedback, and recommendations for future implementation while addressing copyright and data-privacy concerns.

As a result of these explorations, as well the shifting academic and research integrity implications of the use of generative AI, many research libraries are in the process of, or have plans to, modify their information literacy and digital/data literacy curriculum to be inclusive of these unique information-integrity issues. For example, in many research libraries information literacy librarians are revamping their training and online tutorials to equip researchers and community members with skills on the responsible use of generative AI, assessing the accuracy of responses, and how to cite the use of these tools, among others. The Stony Brook University Libraries has also hosted forums among faculty, students, and staff on generative AI. Through these forums, modifications to information literacy programs, and other services and infrastructure, the librarians at Stony Brook have become the go-to AI experts on campus – providing guidance on AI issues and governance.  Research libraries are also actively partnering with other departments to create resources, such as LibGuides, tutorials, and workshops.

To successfully integrate generative AI into library operations, research libraries are beginning to research, assess, and plan for any needed workforce development. This process is varied across libraries, but includes conducting a skills assessment, hiring new talent, partnering with external experts, and training existing staff to handle AI-driven tasks and services. The University of Florida has a natural language processing (NLP) specialist within the library who provides support, such as consultations, training, and tutorials to students, faculty, and staff in research that leverages NLP.

While there is much work happening within libraries on the use of generative AI and support for student and faculty research, libraries are also key partners and players in organization-wide discussions on generative AI. These organization-wide task forces may focus on implications for learning, academic integrity, and research integrity. By positioning themselves as a resource for their organizations, libraries play a pivotal role in shaping AI policies and practices on campus. At the University of Waterloo, the libraries have been engaged in organization-wide conversations around guidelines for use of AI tools in teaching and learning, including citing and using AI tools in reports, publications, assignments, and copyright consultations on the use of generative AI outputs.

As research libraries investigate, pilot, and launch new initiatives and services around generative AI, ARL will create opportunities for members to collaborate and share their experiences, knowledge, and best practices in implementing and integrating generative AI technologies. This will help other libraries make informed decisions about AI adoption and avoid potential pitfalls.

Embracing generative AI technologies in ARL libraries presents both opportunities and challenges. This quick poll was limited by its small sample size, but we aim to continue this conversation and seek to engage with more member representatives on this topic in the coming months.