AI Resources Clearinghouse Mockup

This collection includes a variety of resources relating to artificial intelligence (AI) and libraries, including the role AI plays in optimizing library operations and serving communities. The resources are intended to help the library community and vested partners learn and stay informed about artificial intelligence and libraries, including Generative, Descriptive, and Discriminative AI.

Filter By

Type

Access

Tags

Date Range

46 Items

Artificial Intelligence: A Policy-Oriented Introduction

Policy Brief

By Anne Bowser, Michael Sloan, Pietro Michelucci. Eleonare Pauwels

Published November 1, 2017 by Wilson Center

General Background, Policy, Human Computation, Risk/Benefit Analysis

Technological advances in artificial intelligence have the potential to transform society through better decision-making and improvements to the human condition. However, without adequate risk assessment and mitigation, AI may pose threats to existing vulnerabilities in defenses, economic systems, and social structures. The benefits of AI include augmenting human capacity, improving the human condition, and solving complex problems. Concerns about AI include job loss, risks to human safety and security, biased algorithms causing discrimination, and the potential for a global arms race. Policy recommendations suggest incorporating human participation into complex socio-technical systems to ensure safe and equitable development of automated intelligence.

The scholarly footprint of ChatGPT: a bibliometric analysis of the early outbreak phase

Academic Article

By Faiza Farhat, Emmanuel Sirimal Silva, Hossein Hassani, Dag Øivind Madsen, Shahab Saquib Sohail, Yassine Himeur, M. Afshar Alam, Aasim Zafar

Published January 5, 2024 by Frontiers in Artificial Intelligence

General Background, Bibliometric Analysis, Academic Review

The rapid advancements in artificial intelligence have led to the development of sophisticated language models like ChatGPT, which has garnered significant attention in various domains. From November 2022 to early June 2023 there was a surge in scholarly publications and collaboration among researchers worldwide. This text serves as a reference guide for new researchers in GPT research, highlighting influential authors, studies, and institutions.

Artificial intelligence in the work context

Academic Article

By Mohammad Hossein Jarrahi, Christoph Lutz, Karen Boyd, Carsten Oesterlund, Matthew Willis

Published December 21, 2022 by Journal of the Association for Information Science and Technology

General Background, Academic Review, Editorial, Ethics

Artificial intelligence systems have unique characteristics that can impact work practices, with their logic often not transparent to humans. The transformative power of AI lies in descriptive and predictive models, big data utilization, and changes in work practices. The interdependence of AI models, data, and work practices has consequences for regulations and policies. Ethical questions and power dynamics arise in AI models related to medical work practices. The integration of AI may lead to new divisions of work between humans and intelligent systems. Future research on AI and work will need to involve transdisciplinary collaboration to study AI’s impact on various industries and occupations. This collaboration should include computer scientists, social scientists, and ethical/legal scholars to provide a holistic perspective. Industry partnerships and multistakeholder collaborations should also be encouraged to ensure critical scrutiny of AI systems and avoid ethicswashing.Artificial intelligence systems have unique characteristics that can impact work practices, with their logic often not transparent to humans. The transformative power of AI lies in descriptive and predictive models, big data utilization, and changes in work practices. The interdependence of AI models, data, and work practices has consequences for regulations and policies. Ethical questions and power dynamics arise in AI models related to medical work practices. The integration of AI may lead to new divisions of work between humans and intelligent systems. Future research on AI and work will need to involve transdisciplinary collaboration to study AI’s impact on various industries and occupations. This collaboration should include computer scientists, social scientists, and ethical/legal scholars to provide a holistic perspective. Industry partnerships and multistakeholder collaborations should also be encouraged to ensure critical scrutiny of AI systems and avoid “ethicswashing”.

ChatGPT is not all you need. A State of the Art Review of large Generative AI models

Academic Article

By Roberto Gozalo-Brizuela, Eduardo C. Garrido-Merchan

Published January 11, 2023 by arXiv

General Background, Academic Review, Taxonomy

Generative AI refers to artificial intelligence that can generate novel content, using discriminator or transformer models trained on datasets to create new content. This text discusses developing a taxonomy for Generative AI models by focusing on the applications of each model rather than technical aspects like transformers, etc. Various Gen AI approaches have been developed, such as Galactica, ChatGPT3, and DALL·E 2, each with unique capabilities and limitations. Some models focus on converting text to images, while others, like Codex and Alphacode, translate text to code. Additionally, models like Flamingo and Soundify work on tasks involving vision and audio synthesis. The field of generative AI is rapidly evolving, with ongoing challenges related to data availability, model training, and ethical considerations.

Natural language processing in the era of large language models

Academic Article

By Arkaitz Zubiaga

Published January 12, 2024 by Frontiers in Artificial Intelligence

General Background, Academic Review, Ethics, Language

Large language models (LLMs) are based on traditional language models from the 1980s and leverage statistical language properties from large text collections. The increase in computational capacity has led to the rise of deep learning models and LLMs, impacting NLP research. They can be pre-trained on large datasets and fine-tuned for specific domains. However, research has shown that LLMs suffer from important limitations such as their “black box nature” hindering reproducility/lack of explainability, being sensitive to data contamination, integrating social biases, generating offensive content, violating individual’s right to privacy, and suffering from imperrfect accuracy and “hallucination.” Future research will need to focus on addressing the limitations of LLMs, particularly in data curation to mitigate risks related to fairness, privacy, and harm.

Prompt Design and Engineering: Introduction and Advanced Methods

Academic Article

By Xavier Amatriain

Published February 8, 2024 by arXiv

General Background, Prompt Engineering

A prompt in generative AI models is the textual input provided by users to guide the model’s output. Prompt engineering involves tailoring prompts for different contexts using domain knowledge and understanding of the AI model. Techniques like Retrieval Augmented Generation and Chains enhance the capabilities of Large Language Models (LLMs). Self-Consistency and Automatic Multi-step Reasoning and Tool-use (ART) are important prompt engineering techniques. The field of prompt engineering is evolving rapidly, offering tools and frameworks to optimize LLM performance and reliability.

Should We Trust a Black Box to Safeguard Human Rights? A Comparative Analysis of AI Governance

Academic Article

By Scott J Shackleford, Isak Nti Asare, Rachel Dockery, Anjanette H. Raymond, Alexandra Sergueeva

Published January 1, 2022 by UCLA Journal of International Law and Foreign Affairs

Policy, Ethics

Various entities are developing AI strategies to capitalize on economic, security, and social opportunities, but there is a need to analyze if common AI norms like transparency and accountability are being incorporated into these strategies for future norm development. Content analysis of national AI strategies reveals areas of convergence and divergence, highlighting opportunities for norm development. Recommendations include sustainable development as a driving concept for international AI efforts. Challenges include the regulation of AI and the balance between security and privacy.

Universal Guidelines for Artificial Intelligence

Statement

Published October 23, 2018 by The Public Voice

Policy, Ethics

Advancements in Artificial Intelligence are changing various aspects of society, including decision-making processes that impact people’s lives, raising concerns about fairness, transparency, and accountability. Twelve “Universal Guidelines” are proposed to guide the design and use of AI, emphasizing the importance of ethical standards and transparency. The responsibility for AI systems should reside with institutions, individuals have the right to know the basis of AI decisions, and the true operator of AI systems must be disclosed. The rise of AI decisionmaking impacts fundamental rights, and the guidelines should be incorporated into ethical standards, national laws, and international agreements.

The uselessness of AI ethics

Academic Article

By Luke Munn

Published August 23, 2022 by AI and Ethics

Ethics

AI ethical principles are ineffective in addressing the negative impacts of AI technologies due to their contested, isolated, and toothless nature, leading to a dangerous distraction from more effective approaches to AI justice. The focus on ethics is used as a substitute for regulation, leading to a gap between principles and practice. The development of AI technologies must consider broader societal impacts and engage with social and political questions. The proliferation of AI ethical frameworks has led to a need for more specific recommendations and a focus on core principles such as beneficence, non-maleficence, autonomy, justice, and explicability. Organizations should engage with marginalized groups impacted by AI and work towards addressing historical inequities. Grassroot efforts and conventional governance structures are suggested as ways to ensure AI contributes to well-being and does not exacerbate inequalities.

AI under great uncertainty: implications and decision strategies for public policy

Academic Article

By Maria Nordström

Published September 7, 2021 by AI & Society

Policy

Decisions on public policy regarding the implementation of machine learning and AI for public use are being made under great uncertainty due to the vague definition of AI, uncertain outcomes of AI implementations, and pacing problems. This paper suggests that decision-makers adopt strategies from decision theory to mitigate this uncertainty. The argumentative approach is proposed to mitigate uncertainty challenges. The importance of acknowledging uncertainty in AI policy decisions is highlighted, with a focus on the need for cautious regulation. The paper also addresses the role of public policies in ensuring Trustworthy AI and the challenges faced by decision-makers in the context of AI development. Various types of uncertainty in AI policy are identified, emphasizing the need for proper governance mechanisms and iterative assessment.

Executive Order N-12-23

Executive Order

By Governor Gavin Newsom

Published September 6, 2023 by Office of the Governor of the State of California

Policy

How artificial intelligence might change academic library work: Applying the competencies literature and the theory of the professions

Academic Article

By Andrew Cox

Published March 7, 2022 by Journal of the Association for Information Science and Technology

AI in Libraries, AI in Libraries – Academic Libraries, LIS Profession

Artificial intelligence is expected to have a number of different impacts on libraries and librarians. AI applications can be anticipated to play a role in a number of library services, including knowledge discovery (such as searching across the web, library collections, and licensed data products), customer relations (such as chatbots and virtual assistants), as well as for behind the scenes library work (such as library analytics, robotics/process automation, and “smart library” tools). Overall, the ability of the academic library profession to adapt to the changing technology landsape will depend on the ability of librarians to develop new skills and competencies that complement and work with AI tools. Given the new reality of modern LIS, it might be best to conceptualize LIS as a hybrid profession encompassing the role of both traditional LIS and IT professions. The text concludes with a discussion of different institutional approaches that academic libraries might take to respond to the emergence of AI, along with skills and resources required.

Adoption of Artificial Intelligence (AI) in Library Parlance: Issues and Benefits

Academic Article

By Asom Faga, Aliyu Olugbenga Yusuf

Published Summer 2023 by Library Philosophy and Practice (e-journal)

AI in Libraries, Risk/Benefit Analysis

Artificial Intelligence has become essential in improving efficiency and productivity in various sectors, including libraries. AI technologies such as Augmented Reality, Virtual Reality, and gesture recognition have transformed user experiences in libraries. Despite concerns about bias and privacy, AI offers significant benefits in library operations by enhancing efficiency, personalized services, and research recommendations.

5 Ways Artificial Intelligence Impacts Libraries

Trade Article

By Charla Viera

Published April 4, 2023 by American Journal Experts

AI in Libraries

Libraries and librarians have always been responsible for organizing, providing access to, and protecting information, making them leaders in adapting to new technologies throughout history. Ongoing trends in generative AI highlights the potential for new tools that can enhance information organization, accessibility, user services, and library analytics. Generative AI has the potential to improve management systems and offers a new role foor libraries in promoting AI literacy among patrons. The future of AI in libraries involves creating ‘Smart Libraries’ and incorporating AI into library analytics for real-time data analysis. Library professionals are adapting to AI tools and software to enhance their roles as information guardians and community partners. The article also highlights the slow adoption of new technologies by libraries and the importance of AI literacy for patrons to engage with society’s increasing use of AI tools.

Making and Using AI in the Library: Creating a BERT Model at the National Library of Sweden

Academic Article

By Chris Haffenden, Elena Fano, Martin Malmsten, Love Börjeson

Published January 1, 2023 by College & Research Libraries

AI in Libraries, AI in Libraries – National Libraries, Natural Language Processing

The advent of artificial neural networks offers tantalizing possibilities for libraries to classify, organize, and make huge digital collections searchable with the help of artificial intelligence (AI). To this end, various academic and national libraries have established data labs as testing sites to explore and harness such potential. This text discusses the implementation of a BERT language model at the National Library of Sweden, highlighting its outperformance of existing models for Swedish. The model has various potential use cases in libraries, such as improving access to collections for researchers and enhancing searchability. There are also broader opportunities for integrating AI with libraries; however fully capitallizing on those opportunities will require funding and collaboration to optimize AI development in this context.

ChatGPT: Implications for academic libraries

Trade Article

By Christopher Cox, Elias Tzoc

Published March 1, 2023 by College & Research Libraries News

AI in Libraries, AI in Libraries – Academic Libraries

ChatGPT offers an alternative to traditional search engines like Google by providing relevant resources and assisting in knowledge production. It can generate rough drafts for inspiration and assist in creating art. Librarians can integrate AI tools to improve research and teaching, while entrepreneurs are using AI-generated art for profit. ChatGPT can also assist in writing emails and generating book lists. Libraries can embrace AI tools to enhance services and support their use.

A Survey on Artificial Intelligence Aided Internet-of-Things Technologies in Emerging Smart Libraries

Academic Article

By Siguo Bi, Cong Wang, Jilong Zhang, Wutao Huang, Bochun Wu, Yi Gong, Wei Ni

Published April 13, 2022 by Sensors

AI in Libraries, Internet of Things

The rise of artificial intelligence and Internet-of-Things has led to the integration of smart devices in society, improving public services and management, including the emergence of “Smart Libraries” utilizing AI and IoT technologies for smart service, sustainability, and security, with a focus on future trends. This text discusses the rapid development of artificial intelligence and Internet-of-Things in smart libraries, highlighting improvements in service, sustainability, and security. It covers key technologies such as NLP, RFID, and BLE, as well as challenges and future directions. The integration of AI and IoT enhances operations, security, and sustainability in libraries, with applications in book finding, recommender systems, and security measures. Overall, AI-aided IoT plays a crucial role in making libraries smarter and more efficient.

Application of Artificial Intelligence (AI) In Libraries and Its Impact on Library Operations Review

Academic Article

By Subaveerapandiyan A

Published June 24, 2023 by Library Philosophy and Practice (e-journal)

AI in Libraries, AI in Libraries – Academic Libraries

The rapid advancements in Artificial Intelligence have transformed libraries by introducing AI chatbots, intelligent libraries, robots, and smart libraries. AI technologies in libraries can improve information retrieval, automate tasks, personalize user interactions, and provide innovative services. Intelligent libraries equipped with AI can streamline processes and enhance user experiences. Further research is needed to explore the long-term impact of AI on library operations. AI has the potential to make libraries more user-centered and accessible. This literature review discusses the implications of AI on libraries, challenges, and opportunities. Despite challenges, AI tools like ChatGPT can enhance library services. Various studies and conferences explore the integration of AI in library management and development.

Leadership Brief: Explorations of Generative AI for Library Systems

Policy Brief

Published October 1, 2023 by Urban Libraries Council

AI in Libraries, Ethics

While generative AI remains a powerful tool with amazing capabilities and enormous economic potential, it has short falls, challenges and potential risks. One of the major issues with generative AI is its inability to verify the sources of the information compiled, as well as its potential to sometimes hallucinate and provide made-up information. Building more powerful generative AI models could lead to the development of systems with behaviors the creators might not be able to anticipate or control. Other emerging concerns around AI’s acceleration are job security and creative licensing such as the demands of striking Hollywood writers for regulation that stops AI from writing or rewriting any film materials, as well as copyright campaigns by the Authors Guild calling on Congress to ensure that authors are adequately compensated for the intellectual property used to train AI models. As generative AI tools become more accessible, effective and less expensive, new opportunities for libraries to lead around information literacy and responsible AI are emerging.

IFLA Statement on Libraries and Artificial Intelligence

Statement

Published October 1, 2020 by International Federation of Library Association and Institutions

AI in Libraries

The adoption of Artificial Intelligence and machine learning in private and public spheres is rapidly growing, with libraries playing a crucial role in promoting algorithmic and digital literacy. Librarians can partner with other organizations to deliver education on ethical AI use and support the development of AI technologies that adhere to privacy and inclusivity standards. AI applications, such as automated content moderation and deepfakes, raise concerns about intellectual freedom and privacy. Libraries can help ensure AI benefits society by promoting AI literacy and responsible AI use. Recommendations include incorporating Text and Data Mining exceptions in copyright frameworks and procuring technologies that meet legal and ethical standards.

7 ways artificial intelligence is changing libraries

Trade Article

AI in Libraries

Artificial intelligence in libraries is revolutionizing the way information is accessed and managed. AI tools like indexing automation and content summarization are improving efficiency and accuracy in research. Librarians are embracing AI to enhance services and adapt to the changing information landscape. The focus is shifting towards advanced technologies to better serve the upcoming generations. The impact factor is evolving with AI algorithms validating research arguments for broader readership.

Recommend removal

Published January 11, 2021 by iris.ai

Developing a library strategic response to Artificial Intelligence

White Paper

By Andrew Cox

Published November 20, 2023 by International Federation of Library Association and Institutions

AI in Libraries, Ethics, LIS Profession

This working document discusses considerations for libraries responding to Artificial Intelligence (AI), including definitions of AI, ethical concerns, and the importance of AI literacy training. It emphasizes the need for libraries to strategically position themselves in relation to wider priorities, engage with users to understand the impact of AI, and collaborate internally and externally to increase capacity and influence. The document also highlights the potential impacts of AI on library services and the importance of aligning new tools with organizational goals. Additionally, it addresses the need for transparency, data sources, bias, privacy, and ownership of user data in AI implementation. The draft received input from various individuals, including librarians from higher education, further education, and health sectors. The data was gathered from a survey conducted at an April 2023 event in Sheffield and recirculated in July, primarily reflecting opinions from the UK audience.

Generative AI for library and information professionals

White Paper

Published April 1, 2024 by International Federation of Library Association and Institutions

AI in Libraries, LIS Profession

Generative AI systems can create new text, images, and media, but they may exhibit biases due to training data. Concerns about bias have led to calls for AI regulation. Users should critically evaluate and understand how generative AI tools work. Questions remain about the legitimate uses of generative AI, especially in research. Despite potential benefits, ethical considerations should be taken into account when using generative AI like ChatGPT. Generative AI has various applications, including text summarization in libraries. Proper use and citation of generative AI tools are important for ethical and accurate information dissemination.

Really a list of resources, with liitle/no analysis. Evaluate resources for inclusion inn reading list and remove?

By Andrew Cox