IFLA Statement on Libraries and Artificial Intelligence

The adoption of artificial intelligence and machine learning systems in private and public spheres is rapidly growing. How can these developments impact both everyday practices and the broader social mission of libraries? IFLA's new statement sets out some key principles.

The rapid pace of AI development and adoption raises crucial questions about intellectual freedom, equity and privacy, automation, the evolution of necessary digital literacy skills, relevant Intellectual Property policy frameworks, and more.

The new IFLA FAIFE statement aims to outline key considerations for the use of AI and ML technologies in the library sector, and suggest the roles which libraries should strive to take on in a society with growing AI integration.

You can find the full text of the Statement and links to annexes with background information below, or download both as a .pdf.

IFLA Statement on Libraries and Artificial Intelligence

AI technologies can have deeply transformative capabilities, and their power can be put to the service of public welfare and innovation. With necessary preparations – and regard for ethical concerns and current limitations – libraries can responsibly use AI technologies to advance their social mission.

AI in the Library: integrating AI and machine learning technologies into everyday work.

As AI technology develops, several AI and machine learning (ML) applications may be able to introduce new services and functions to libraries. For example, AI and ML technologies could allow libraries to improve optical character recognition of texts, or make new uses of their machine-readable library collections (e.g. categorisation or discovery) – whether for libraries themselves, patrons, or researchers.

Similarly, AI and ML could have the potential to add new dimensions and approaches to knowledge management processes in libraries – particularly knowledge organisation, storage and integration. AI may be able to offer new dimensions to service provision when coupled with robotics. While there are cases where AI might be used to automate some of the existing library services (in such AI applications as chatbots or search and discovery tools), care should be taken to prevent negative impacts on quality of service and staffing (as described in Annexure i).

Libraries and library associations can, for example, interact with AI researchers and developers to create applications specifically for library use and/or in response to user needs, including by creating accessible services which have not been possible before.

The use of AI technologies in libraries should be subject to clear ethical standards, such as those spelled out in the IFLA Code of Ethics for Librarians and other Information Workers. For instance, AI applications which rely on extensive data collection – such as behaviour analytics software for performance evaluation - must not override patron privacy choices or equity considerations (further detailed in Annexure i)

Libraries can educate users about AI, and help them thrive in a society which uses AI more extensively.

A growing number of initiatives from governments and civil society organisations aim to educate the public at large about AI and the social implications of its use. Libraries are a trusted and credible source of knowledge, and they can help extend these efforts to their communities.

Today, many libraries work to help their patrons develop digital literacy competencies – the abilities to make meaningful use of technology, from basic IT skills to creative abilities - safely, ethically and legally. These efforts can be expanded to promote algorithmic literacy – the understanding of how algorithms and other digital processes impact the way users access and receive information.

Public libraries, for example, can be well-positioned to deliver such training to the wider public (Annexure ii), but this can require significant time and resource investment, as well as upskilling on the part of librarians. Librarians therefore need to be supported in these efforts – and they could also partner with other organisations or sectors to help deliver algorithmic and digital literacy education (further described in Annexure ii).

By helping their patrons understand AI, libraries could also enable them to participate in policy discussions on what it means for AI to be used “for good”, and which use cases are ethical and desirable for the public.

More broadly, libraries could play a role in helping their users adapt to some of the possible changes in the labour market caused by AI. Lifelong learning is likely to play a vital role in helping people navigate changing skill demands or career transitions. The tools and services libraries offer can provide equitable lifelong learning opportunities, particularly for vulnerable or marginalised populations. For example, the free digital literacy and IT skills classes many libraries offer may become increasingly important as these skills affect employability more than ever before.

Libraries can support high-quality, ethical AI research.

Many current ethical and inclusivity concerns linked to AI research and applications stem from incomplete, incorrect or biased training data (‘garbage in, garbage out’). Trained librarians can lend their expertise in data storage and licensing, data quality assessment, and safe and ethical information storage to help researchers address some of the concerns around data.

Libraries can also support ethical AI research and development by their procurement choices: purchasing AI technologies which abide by ethical standards of privacy and inclusivity. This would both reaffirm the trust of users in libraries, and send a message to the AI research field by increasing the demand for ethical AI technologies.

What do libraries need to adapt successfully?

The transformative nature of AI means that libraries need to be in a position to make use of the new resources as the technology landscape changes, adjusting their services as appropriate to meet changing societal needs.

To this end, librarians may benefit from a greater awareness of the technology landscape. Since AI can significantly reshape employment in the sector, they might also require training to adjust to the possible shifts in their workplace roles. Notably, libraries in different sectors could be affected differently – for example, government, legal and other special libraries could experience a greater emphasis on knowledge-based service delivery.

Libraries will need to have the laws, infrastructure and technologies necessary to be able to adopt and make use of AI. Text and Data Mining exceptions in copyright frameworks are crucial to allow libraries to continue carrying out their mission, and actively participate in AI communities.

Recommendations

Based on the above, IFLA makes the following recommendations to governments, libraries and library associations:

Governments (and intergovernmental organisations, where appropriate) should:

  • Include Text and Data Mining exceptions in copyright frameworks
  • Ensure that libraries or library networks have the required infrastructure and technologies to be able to adopt and make use of AI technologies.
  • Ensure that any regulation of AI protects privacy or equity principles, while also enabling efforts that support innovation and public interest goals.
  • Ensure that libraries are included in development and implementation of cross-sectoral AI programs and strategies.

Library associations and library training providers should:

  • Support library professionals to understand the impact of AI as well as its intersections with privacy and ethical principles. Library training providers should ensure that librarians are able to develop relevant digital skills and competencies.
  • Advocate for libraries to play a bigger role in changing education systems as they adapt to the labour market changes which AI might bring.
  • Engage with AI researchers and developers to create applications for library use, which meet ethical and privacy standards and respond specifically to the needs of libraries and library patrons.
  • Act as forums for exchanging best practices on ethical use of AI technologies in libraries.

Libraries should, where possible and appropriate:

  • Help their patrons develop digital literacies that include an understanding of how AI and algorithms work, and corresponding privacy and ethics questions.
  • Continue to focus their efforts to enable lifelong learning and, where possible, provide services for the unemployed.
  • Ensure that any use of AI technologies in libraries should be subject to clear ethical standards and safeguard the rights of their users.
  • Procure technologies that adhere to legal and ethical privacy and accessibility requirements.

Agreed by the IFLA Governing Board, 17 September 2020

Statements, FAIFE (Committee on Freedom of Access to Information and Freedom of Expression), Access to information, Big data, Information society, Artificial intelligence, machine learning

Last update: 21 October 2020