AI + Libraries Call for Proposals

March 25, 2024 -Slides used during the presentation

Your Name and Title:
Paul R. Pival, Research Librarian – Data Analytics

Library, School, or Organization Name:
University of Calgary

Co-Presenter Name(s):
N/A

Area of the World from Which You Will Present:
Alberta, Canada

Language in Which You Will Present:
English

Target Audience(s):
Librarians and Library Staff

Short Session Description: 
While acknowledging the crucial role of other literacies (e.g., data, information, digital) within academic libraries, there is currently no agreed-upon literacy for library/librarian use of Artificial Intelligence.

Full Session Description: 
While acknowledging the crucial role of other literacies (e.g., data, information, digital) within academic libraries, there is currently no agreed-upon literacy for library or librarian use of Artificial Intelligence. This presentation will make the case for the need for a framework for AI literacy within libraries, specifically around the use of Large Language Models (LLMs). AI literacy for librarians should include, at a minimum, the following:

• How LLMs work: Librarians should know the basic principles of how LLMs are trained, how they generate text, and what are the main components and architectures of LLMs. This would help them to evaluate the quality and reliability of LLM outputs, as well as to use LLMs effectively for their own tasks.
• What biases might be inherent within a LLM: Librarians should be aware that LLMs can inherit and amplify biases from their training data, such as gender, racial, cultural, or ideological biases. These biases can affect the content and tone of LLM outputs, as well as the representation and inclusion of different groups and perspectives. Librarians should be able to identify and mitigate these biases, as well as to educate their users about them.
• What are hallucinations: Librarians should understand that LLMs can produce hallucinations, which are false or misleading information that is not supported by the input data or the real world. One of the most common type of hallucination experienced in library-use of LLMs is that of the non-existent citation.
  
• What are some of the ethical considerations: Librarians should consider the ethical implications of using LLMs, such as the impact on privacy, data security, intellectual property, accountability, transparency, and human agency. Librarians should follow ethical principles and guidelines, such as the ALA Code of Ethics, when using LLMs.

By building their AI literacy to an agreed-upon standard framework, library professionals can ensure they remain at the forefront of information provision, effectively adapting to the evolving AI landscape. This presentation will act as a starting point for discussion, with time left for feedback and participation.

Websites / URLs Associated with Your Session:

https://libguides.ucalgary.ca/artificialintelligence
https://www.ala.org/tools/ethics
https://www.ifla.org/units/ai/
https://www.ifla.org/units/information-literacy/
https://www.ala.org/acrl/standards/ilframework