The Association of Research Libraries (ARL) has issued a set of “Research Libraries Guiding Principles for Artificial Intelligence.” AI technologies, and in particular, generative AI, have significant potential to improve access to information and advance openness in research outputs. AI also has the potential to disrupt information landscapes and the communities that research libraries support and serve. The increasing availability of AI models sparks many possibilities and raises several ethical, professional, and legal considerations.
Articulating a set of research libraries guiding principles for AI is useful to influence policy and advocate for the responsible development and deployment of AI technologies, promote ethical and transparent practices, and build trust among stakeholders, within research libraries as well as across the research environment. These principles will serve as a foundational framework for the ethical and transparent use of AI and reflect the values we hold in research libraries. ARL will rely on these principles in our policy advocacy and engagement.
Developed through the ARL Scholars and Scholarship Committee and the ARL Advocacy and Public Policy Committee, these principles have been created and refined after multiple committee meetings, a member-wide open office-hours session, and an open comment period. During the open comment period alone, we received over 60 comments that have been taken into consideration in the development of these principles.
Principles
- Libraries democratize access to artificial intelligence tools and technology to foster digital literacy among all people. Libraries embrace the opportunity to educate library users and information professionals about AI, equipping them with the knowledge and skills needed to thrive in a society increasingly reliant on artificial intelligence. This principle underscores the library’s commitment to inclusivity, knowledge dissemination, and the empowerment of individuals.
- Libraries commit to understanding where distortions and biases are present in AI models and applications. Recognizing the inherent susceptibility of all AI to distortion, we leverage our expertise to raise user awareness regarding distortions present in AI systems and their outputs. We strive to enhance transparency and understanding, fostering a library and research environment that empowers users with the knowledge to navigate and critically assess AI-driven information and services.
- Libraries champion transparency and information integrity. Libraries will advocate for openness and transparency in algorithms, training data, and methodologies used to create AI technologies that we license.
- Libraries believe “no human, no AI.” This principle underscores the importance of human involvement in critical decision-making junctures, enhancing accountability, ethical considerations, and the overall reliability of AI-driven processes within the research environment.
- Libraries prioritize the security and privacy of users in the use of AI tools, technology, and training data. Libraries will continue to advocate for laws and regulations that protect the personal information of library users and employees.
- Libraries assert that copyright law in the US and Canada is flexible and robust enough to respond to many copyright issues that arise from the intersection of technology and artificial intelligence. Libraries have a stake in preserving library rights for nonprofit research and education uses. More on copyright law in the US and Canada may be found in the “Library Copyright Alliance Principles for Copyright and Artificial Intelligence” and “CARL’s Response to the Consultation on Copyright in the Age of Generative AI.”
- Libraries negotiate to preserve the scholarly use of digital information. As champions of fair use and fair dealing, intellectual freedom, and freedom of information, libraries assert that licenses should not restrict the scholarly use of digital information.
Resources
- “ICOLC Statement on AI in Licensing,” International Coalition of Library Consortia (ICOLC), March 22, 2024.
- Transparency is actively pursued through the application of insights from Explainable AI (XAI), embodying the core premise of the “right to an explanation.” See “Explainable Artificial Intelligence,” Michael Ridley, Research Library Issues, no. 299 (2019): 28–46.
- “Library Copyright Alliance Principles for Copyright and Artificial Intelligence,” Library Copyright Alliance, July 10, 2023.
- “CARL Response to the Consultation on Copyright in the Age of Generative AI,” Canadian Association of Research Libraries (CARL), December 15, 2023.
Download a PDF of the “Research Libraries Guiding Principles for Artificial Intelligence.”
About the Association of Research Libraries
The Association of Research Libraries (ARL) is a nonprofit organization of research libraries in Canada and the US whose vision is to create a trusted, equitable, and inclusive research and learning ecosystem and prepare library leaders to advance this work in strategic partnership with member libraries and other organizations worldwide. ARL’s mission is to empower and advocate for research libraries and archives to shape, influence, and implement institutional, national, and international policy. ARL develops the next generation of leaders and enables strategic cooperation among partner institutions to benefit scholarship and society. ARL is on the web at ARL.org.
