Module 3: How AI Is Creating Library Workflow Efficiencies
AI Literacy Essentials for Academic Libraries: Beyond the Basics
Choice and Clarivate have teamed up once again to expand our eight-week, newsletter-based course on generative AI literacy for academic library workers by creating three new modules. Each module contains bite-sized, self-paced material such as readings, multimedia content, case studies, and interviews with thought leaders.
Make sure you’re registered to receive all three modules in your inbox! If you missed the original eight-part series, you can register here.
In this module, we will examine where AI can have the greatest impact on workflow efficiencies without eroding roles or institutional values.
Learning Objectives:
- Explore practical examples of how AI tools can streamline library workflows
- Identify repetitive tasks that could be appropriately automated with AI
- Consider guardrails or bounded uses that will enable viable workflow efficiencies
- Evaluate AI-generated results for accuracy, feasibility, and ethical considerations
Table of Contents
Introduction
Transforming Academic Library Workflows with AI: Enhancing Efficiency While Preserving Values
by Dr. Kari D. Weaver, Program Manager, Artificial Intelligence and Machine Learning, Ontario Council of University Libraries
As stewards of knowledge and operators of sophisticated information systems, libraries manage complex workflows, structured metadata, and user-centered services informed by our professional values. These are all areas where artificial intelligence (AI) is reshaping knowledge work within libraries. When implemented thoughtfully, AI can enhance values and improve service by automating repetitive work, enabling staff to focus on higher-value activities, and connecting siloed systems into more cohesive, responsive service environments.
Automating Repetitive and Rules-Based Work
Some library workflows consist of repetitive, rules-based tasks that are crucial but time-consuming. Examples include cataloging and metadata creation, acquisitions processing, and circulation management. AI technologies are well-suited to tasks with established standards and schemas, predictable and repeatable steps, or well-defined inputs and outputs.
In collections analysis, for instance, AI can compare spreadsheets of titles or suggest materials for weeding based on factors like circulation numbers, publication date, and classification area. In a weeding project, AI could use the spreadsheet data to make suggestions for shelf density and materials distribution. This approach shifts the librarian’s work toward professional expertise and oversight, providing quality assurance of the AI analysis.
Similarly, AI-driven tools can streamline acquisition workflows by automating invoice processing, vendor communications, and metadata extraction. In digitization projects, computer vision, AI tools that allow computers to interpret, analyze, and understand visual information such as images and videos, can automatically extract text from scanned materials, classify images, and detect errors. Chatbots and virtual reference assistants can handle routine patron inquiries, such as library hours, account status, or basic search guidance.
Automating these tasks can reduce backlogs, improve consistency, and accelerate service delivery. However, AI should be positioned as a tool that augments human expertise rather than replaces it. Human oversight is critical to ensure accuracy, mitigate bias, and uphold professional standards.
Enabling Staff to Shift to Mission-Aligned Work
By reducing the burden of repetitive tasks, AI can enable library staff to focus on more mission-aligned work. This shift is not only about efficiency but about re-centering human expertise where it has the greatest impact: teaching, research support, community engagement, and critical stewardship of knowledge.
In special collections and archives, for example, staff can move beyond basic description and processing to focus on interpretive work and community partnerships. AI-assisted transcription and metadata generation can make hidden collections more accessible, while human expertise ensures cultural sensitivity and contextual accuracy. The University of Virginia Archival AI Protocol provides guidance focused on using retrieval and controlled AI models for archival work to preserve context while AI enhances discoverability.
Crucially, this shift requires intentional workforce development. Libraries must invest in training staff to work effectively with AI tools, including topics like prompting, model comparison, ethical AI use, and governance. The individuals best positioned to consider how AI impacts workflows are, in fact, those already doing the work.
Connecting Siloed Systems and Data
Academic libraries operate a constellation of systems: integrated library systems (ILS), institutional repositories, learning management systems (LMS), and more. These systems frequently function in isolation, creating inefficiencies and fragmented user experiences.
AI can help bridge these silos by enabling interoperability and data integration. For example, an AI-powered discovery layer could integrate data from the catalog, institutional repository, and subscription databases to deliver more relevant, personalized search results. In research support contexts, AI could connect datasets, publications, and researcher profiles.
An electronic thesis submission to an institutional repository could use AI to check compliance with submission guidelines, automatically extract metadata, perform Optical Character Recognition (OCR) on the thesis for accessibility, and send notifications when complete. A human-in-the-loop review of the extracted metadata and OCR outputs would ensure quality and discoverability.
Preserving Roles and Institutional Values
While the efficiencies offered by AI are compelling, libraries must approach implementation with a clear commitment to their core values. Transparency, accountability, equity, and user privacy should guide decisions about where and how AI is deployed. Human oversight and explainable AI, systems that are designed to make their decisions understandable to humans, should be used to prevent the amplification of bias. Governance structures, including cross-functional committees and clear policies, can ensure that AI aligns with institutional missions and ethical standards.
Staff must also be active partners in AI considerations. Open communication, training opportunities, and participatory, evidence-based decision-making can help build confidence and foster a culture of innovation.
Finally, libraries should view AI as one tool among many, not a panacea. Its greatest impact lies in complementing human judgment, enhancing existing workflows, and enabling new forms of service that were previously impractical.
Conclusion
AI has the potential to enhance academic libraries by automating repetitive tasks, enabling higher-value work, and connecting siloed systems, and new data backs this up. According to Clarivate’s recent “Academic AI Impact” study, respondents reported consistent improvements in time use and efficiency, access and availability, throughput and capacity, and quality and oversight after incorporating AI tools into their workflows. By centering human expertise, investing in staff development, and governing responsible AI use, libraries can harness these technologies to enhance service, expand access, and better support the evolving needs of their communities.
In Practice
AI Implementation Case Studies
Librarians today face growing workloads and shrinking budgets, constantly asked to do more with less. Although skepticism and hesitancy around AI use are prevalent among the academic library community, the constraints of the moment create an opportunity to use AI in specific, controlled contexts to help reduce librarians’ workloads.
For this module, Choice spoke with librarians and library leaders from two universities to learn about how they are employing AI tools for this purpose. In the first case study, Cory Tressler, Assistant Dean for Technology and Digital Programs at The Ohio State University Libraries, discusses how Ohio State developed a proprietary AI tool to help the library meet the WCAG compliance requirements. In the second case study, Sarah Cogley, Digital Collections and Repositories Librarian at the University at Buffalo, SUNY, and Stacy Snyder, Digital Collections Projects and Compliance Librarian and Accessibility Coordinator at the University at Buffalo, SUNY, share how they evaluated different AI tools to generate alternative text and descriptions for images in the libraries’ digital holdings.
Together, these case studies offer practical examples that can help other librarians evaluate how AI tools might address similar questions, concerns, and operational challenges within their own institutions. By highlighting specific, bounded uses of AI in library settings, they provide a useful framework for thinking about where these tools can improve workflow efficiency, reduce administrative burden, and support staff in focusing on higher-value work.
Click here to learn about how different libraries have been implementing AI tools.
Case Study 1: Using AI to Meet Digital Accessibility Benchmarks
Cory Tressler, Assistant Dean for Technology and Digital Programs, The Ohio State University Libraries
How did you pinpoint an area of work where AI might help to make your library staff’s workflows more efficient?
We looked at areas within our established workflows wherein we either had technical or capacity constraints. We then looked at those areas that we identified to determine which we felt could be addressed through AI exploration and development. We landed on exploring how AI could assist with elements within our workflows related to meeting digital accessibility standards and guidelines. We then started to focus on identifying tools that could assist in that space, working with vendors and peers to co-develop new tools, testing those tools, and building consistency and trust with using those tools.
What AI tool(s) did you adopt to facilitate this process, and what evaluative criteria did you use to make this decision?
We did not find any off-the-shelf tools to use for full PDF remediation, so we partnered with Amazon Web Services and Arizona State University to develop a new tool that could meet our needs. We landed on a tool that utilizes AI-enabled APIs from Adobe and LLMs. After initial testing to determine what was possible, we then started giving specific employees access to the tool and gathering feedback on their experience. We then used that feedback to work with Adobe on anomalies we saw from their APIs and update our LLM prompts until we had consistent and mostly reliable outputs from the tool. We have moved this PDF Remediation tool into production and embedded it in our updated workflows. We continue to gather feedback and refine the tool as AI continues to evolve.
Who was involved in the decision-making process?
The Assistant Dean for Technology and Digital Programs came up with the original idea and acted as the executive sponsor for the development cycles to ensure the proper resources and stakeholders were available. Various Libraries faculty and staff who are associated with specific collections services and repositories were utilized as the testers, evaluators, and refiners of the tool. One system engineer was utilized to make technical updates and feature refinements to the tool.
Can you outline the process of implementing this tool? How did you measure its effectiveness?
We gradually rolled out the tool for testing and used employee feedback and digital accessibility checkers to gauge effectiveness. We added a pre- and post-accessibility checker to the full PDF remediation workflow, so that outlier PDFs that score low on the post checker can be manually reviewed by our one in-house digital accessibility expert.
How did staff respond to using this tool?
The initial response was a little excitement and a lot of skepticism. The excitement came from the scale of our digital accessibility work across our collections, and we didn’t have a human or financial resource pathway to meeting the updated federal guidelines. The skepticism came from the use of AI in general. We have a significant number of employees who are worried about the societal impacts of AI, the cost to the environment, and the growing control of big tech over all aspects of our world. We took baby steps forward at each iteration of our initial utilization and pushed for testing across our internal workflows where the tool could potentially help. We also provided multiple professional learning sessions on foundational AI information.
What was the outcome? Please quantify results if applicable (e.g., x% time or money saved, x% criteria met)
Remediating PDFs via a third-party vendor or via internal human resources costs $5+/page. The PDF Remediation Tool costs $0.07/page. A 100-page PDF can take a vendor or internal employees many hours to days to fully remediate. The PDF Remediation Tool can remediate a 100-page PDF in less than 3 minutes.
Do you plan to implement additional AI tools to address other areas of work?
Yes, we are currently expanding the development of tools and services related to digital accessibility, specifically with A/V and digitized items in our collections. Through that work we will be expanding to metadata and description development for materials being added to our collections.
Are there any key takeaways you would like to share with other academic librarians who are curious about implementing AI tools to streamline their own workflows?
Take the time to build trust through pilots, user testing, and multiple feedback loops with employees. There is some inherent distrust with AI within academia (as there should be), so you should spend the time on finding a use case that doesn’t cause deep ethical or employment-based anxiety for people, then lean into testing to build the trust in the AI’s output across multiple employee and service delivery perspectives. Be flexible, willing to adapt, and willing to stop if the AI output is not meeting your needs or if employee anxiety is too high.
| Demographic Data | The Ohio State University |
|---|---|
| Type of Institution: | Doctorate-granting univeristy |
| Institution Size: | 20,000+ students |
| Public/Private: | Public |
| Location: | Columbus, OH |
Case Study 2: Assessing AI to Generate Alt-Text for Digital Collections
Sarah Cogley, Digital Collections and Repositories Librarian, University at Buffalo, SUNY
Stacy Snyder, Digital Collections Projects and Compliance Librarian; Accessibility Coordinator, University at Buffalo, SUNY
How did you pinpoint an area of work where AI might help to make your library staff’s workflows more efficient?
In the fall of 2024, we started to look closely at what would be required for our digital collections to meet the new ADA Title II requirements. It quickly became apparent that we wouldn’t be able to make any significant progress without either dedicating additional staff resources or investigating technological options. We decided to explore generative AI tools to determine if the technology could create alternative text and descriptions for images in our digital collections.
What AI tool(s) did you adopt to facilitate this process, and what evaluative criteria did you use to make this decision?
We started testing ChatGPT, Gemini, and Claude in the spring of 2025. We identified nine broad categories of images: black-and-white posed, black-and-white candid, color posed, color candid, color built environment, black-and-white built environment, artifacts, illustrations, and postcards. Five images were selected for inclusion in each category, for a total of 45 images. We placed a copy of each image and the prompt into each generative AI tool.
We developed three prompts: a prompt to create alternative text, a prompt to create description, and a combined prompt to create alternative text and description.
Each response was copied into a spreadsheet. These steps were repeated for each image with each of the generative AI tools. Three librarians independently reviewed the AI-generated metadata using a standardized scoring rubric. Each librarian assigned scores from 1 to 5 across four measures: accuracy, relevance, clarity, and safety and ethical considerations.
We removed Gemini from the evaluation early because it would not describe people. We used the rubric to document this decision.
ChatGPT earned higher average scores than Claude across all image sets and evaluation criteria.
The results of the prompt testing indicated that prompt structure (requesting alternative text and descriptions in a single prompt versus using separate prompts) did not significantly affect the quality of output.
Who was involved in the decision-making process?
Testing was supervised by the Digital Projects and Compliance Librarian. The three librarians who scored the results are the Digital Projects and Compliance Librarian, the Digital Collections and Repositories Librarian, and the User Experience Strategist. The results of the test were shared with the Head of Digital Collections and Technology Strategy and the Associate University Librarian for Digital Strategies.
Can you outline the process of implementing this tool? How did you measure its effectiveness?
The University Libraries purchased a business license for ChatGPT so that everyone working on the project could use it without running into prompt limits. The Digital Projects and Compliance Librarian developed instructions for student assistants, who would be doing the work. The instructions included entering existing descriptive metadata for each image in a spreadsheet. The students then copied each image into ChatGPT, added the prompt, and copied the output (AI-generated alternative text and description) into the spreadsheet and the digital collections platform.
We measured the effectiveness by tracking the completion of accessibility metadata (alternative text and description) for images in our digital collections.
How did staff respond to using this tool?
Student assistants have been very eager to use ChatGPT in this way. The Digital Projects and Compliance Librarian has expanded the use of ChatGPT to print and handwriting transcription.
Some librarians involved in the creation of digital collections are less enthusiastic about using generative AI for descriptive metadata creation. There is no requirement for librarians to use generative AI to create descriptive metadata or for any other purpose. The primary use is remediating existing collections.
What was the outcome? Please quantify results if applicable (e.g., x% time or money saved, x% criteria met)
As of spring 2026, approximately 32 percent of images in the digital collection meet accessibility guidelines.
Do you plan to implement additional AI tools to address other areas of work?
Librarians are using ChatGPT to create transcripts for images of text. The UB Libraries are planning to test AI tools that transcribe handwriting and are currently testing AI tools for PDF remediation.
Are there any key takeaways you would like to share with other academic librarians who are curious about implementing AI tools to streamline their own workflows?
- The models are constantly changing, so what might not have been possible a month, six months, or a year ago might be possible now.
- Prompt engineering is necessary. The more specific the prompt, the better the output will be.
- Humans are still needed. AI-generated metadata must be reviewed by librarians.
- Testing must be iterative. Librarians re-ran tests in spring 2026 and found that Gemini and Copilot can identify people. Claude had improved dramatically. In fact, it was on par with ChatGPT.
| Demographic Data | University at Buffalo, SUNY |
|---|---|
| Type of Institution: | Doctorate-Granting University |
| Institution Size: | 20,000+ students |
| Public/Private: | Public |
| Location: | Buffalo, NY |
Reading Room
A selection of multimedia content to fit your time commitment.
5 minutes
- “AltGen: AI-Driven Alt Text Generation for Enhancing EPUB Accessibility.” Proceedings of the 2025 International Conference on Artificial Intelligence and Computational Intelligence, 2025.
- “Higher Ed Prepares for New Era of Accessibility.” Inside Higher Ed, 2026.
- “MetaEnhance: Metadata Quality Improvement for Electronic Theses and Dissertations of University Libraries.” ACM/IEEE Joint Conference on Digital Libraries (JCDL), 2023.
- “University Libraries Improves Access to Digital Materials.” The Ohio State University, 2025.
10 minutes
- “How Artificial Intelligence Might Change Academic Library Work: Applying the Competencies Literature and the Theory of the Professions.” JASIST, 2023.
- “It Takes a Village: A Distributed Model for AI-Based Chatbots.” Information Technology and Libraries, 2024.
- “PDF Accessibility: The Ohio State University.” Artificial Intelligence Cloud Innovation Center, Arizona State University, 2025.
- “The Viability of Using an Open Source Locally Hosted AI for Creating Metadata in Digital Image Collections.” The Code4lib Journal, 2023.
20+ minutes
- “Academic AI Impact: Measuring the Impact of AI Adoption on Academic Library Workflows.” Clarivate, 2026.
- “(Mis)Matching Metadata: Improving Accessibility in Digital Visual Archives through the EyCon Project.” Journal on Computing and Cultural Heritage, 2023.
- “Responsible AI Practice in Libraries and Archives: A Review of the Literature.” Information Technology and Libraries, 2024.
- “Seeing History Unseen: Evaluating Vision-Language Models for WCAG-Compliant Alt-Text in Digital Heritage Collections.” Anthology of Computers and the Humanities, 2025.
- “University of Virginia Archival AI Protocol.” University of Virginia Library, 2026.
Poll
Let us know about current or potential workflow efficiencies your library is implementing or exploring through these four brief poll questions.
Discussion Questions
AI gives us a lot to think about. Share your thoughts in the Leave a Reply section below!
- If your library has already implemented an AI tool to improve workflow efficiency, what tool have you chosen, and how have you implemented it?
- If your library is still considering AI tools, but hasn’t yet employed any, what are the factors influencing your considerations?
- Do you feel any hesitancy around embedding AI tools in your workflows?
Certificate of Completion
To receive your Certificate of Completion for the microcourse, answer the question in the embedded form below. Once finished, click the “Download Certificate” button on the results page. Please note: You must enter your name at the end of the quiz for it to appear on the certificate.
Take me to the Table of Contents.
Know someone who might be interested in this micro-course? Share the registration page with your colleagues! Make sure you’re registered to receive all eight weeks directly in your inbox.
Learn more about Clarivate and Choice’s LibTech Insights.
24 responses to “Module 3: How AI Is Creating Library Workflow Efficiencies”
-
If your library has already implemented an AI tool to improve workflow efficiency, what tool have you chosen, and how have you implemented it? My director and I both use ChatGPT to help support us with emails, images, creating a call # for items if we can’t find one elsewhere. We find it helpful in this capacity.
If your library is still considering AI tools, but hasn’t yet employed any, what are the factors influencing your considerations? The main thing is our budget.
Do you feel any hesitancy around embedding AI tools in your workflows? I am a little hesitant to embedding AI tools one the fact that it is not totally reliable depending on the source it is pulling from on the internet.
-
Our university library is currently exploring workflow efficiencies through an internal, voluntary AI working group rather than a full-scale operational rollout. While our campus provides institutional access to tools like Microsoft Copilot and Google Gemini, we are taking an intentional, step-by-step approach. We are currently developing an AI LibGuide and evaluating how bounded AI applications could assist with repetitive tasks, such as metadata drafting, LibGuides maintenance, and accessibility compliance.
The main factors shaping our cautious considerations are maintaining data privacy, preventing hallucinations in scholarly contexts, and avoiding uncritical adoption. Case studies like the ones featured here demonstrate that having human-in-the-loop validation and clear, bounded tasks are essential to preserving professional standards while gaining meaningful efficiencies.
-
At our library, we are not implementing AI in the library workflows, but we are using AI assistance in routine tasks.
-
Our library hasn’t specifically implemented any AI tools, but students have access to AI tools in our databases.
-
As far I know, our library has not implemented AI tools but I think technical services is using AI agents to assist in their work. I know we have not used them in public services. We still have a staff member monitor our chat. That staff member was thought to be a chatbot by a student asking a question. Questions asked at our research services desk tend to evolve as the questions and answers continue. Use of a chatbot may require a patron recover some aspect of their response if switched to a live person. One old tag line at our university was where learning gets personal. Interacting with chatbot removes the person. Chatbots may be more efficient but the personal connection is lost.
-
1. If your library has already implemented an AI tool to improve workflow efficiency, what tool have you chosen, and how have you implemented it?
Our library hasn’t implemented AI tool to improve workflow. There’s a task team that has been formed called Digital Transformation and AI. Its role is to benchmark the best practices on AI and how best it can assist to improve workflow efficiency. With the knowledge I have gained in this course, I think I can also assist the task team when it comes to selection of the best AI tool to improve workflow efficiency and allow staff to focus more on mission-aligned work.2. If your library is still considering AI tools, but hasn’t yet employed any, what are the factors influencing your considerations?
As indicated, our library is considering AI tools, and the task team is working on that. As we are still working on this project. The first critical step which I fully support is by Dr. Kari. D Weaver. When considering implementing AI tools, libraries must approach implementation with clear a commitment to the values, transparency, accountability and equity. User privacy should guide the decisions about where and how AI is deployed. Governance structures, cross-functional committees and clear policies should be developed. The latter will ensure that AI aligns with institutional mission and ethical standards. Mastering this will enable libraries to have a clear picture on what the chosen AI tools can benefit them.3. Do you feel any hesitancy around embedding AI tools in your workflows?
Since there’s already a task team formed to benchmark the best practices and implement the best AI tool I don’t think there will be hesitancy around this. I would love to be part of that task team to guide and share the knowledge I have gained in this course. AI is an aid not a complete replacement, it can improve accessibility workflow human evaluation will forever be important. -
Our small community college is in the process of developing an AI policy. Until there is an official policy, we are not implementing any AI software tools or implementing training. AI is in its infancy and is like the Wild West with no laws, no guardrails, and no sheriffs. I imagine things will change for the better over time as AI is here to stay. I am concerned about the accuracy of results, the ethics, plagiarism, and the loss of critical thinking skills in our students. This has been an interesting module & the material provided will be useful in future. Thank you.
-
My library has not implemented AI tools apart from minor patron services, i.e. AI Chatbot embedded on the Library website to handle simple enquiries, and for personal or communications use, i.e. institutional access to Copilot that is generally used by individuals for email drafting or other language-based tasks. As far as I know, my library does not have any plans (or budget) to implement AI tools to the collections, nor do we have any strategic plans or protocols in work.
While not directly AI tools, we currently do use several automated services and workflows (such as records and metadata harvesting) that could possibly benefit from AI integration. However, this would likely be implemented by the vendor/upstream instead of by the Library. From our experience, these records still require significant human intervention to validate.
-
An internal policy on staff usage of AI has been approved and distributed. AI can only play an integral part in our work if we obtain the necessary permissions to deploy AI.
Based on the second case study, where the testing approach was described more transparently, this seems something that staff in the library I work in, would be able to replicate. I am sure where AI tools are involved, that such testing has been applied to use the best AI tool available. It just emphasises that there will always need to have a series of human checks to make sure that the created text meets the criteria or standards followed by their institution. I would say, in relation to budgetary considerations, there would need to be some consideration for smaller institutions to balance the flexibility that could be provided by staff versus the limitations that could be provided by AI tools (and vice versa). This is my opinion. After all, AI-produced work could prove burdensome for these organisations who might not be in a position to “expand” in response.
-
At the ANU library staff are using AI in their workflows to assist in more routine tasks, such as assistance in composing emails, writing reports, writing procedure documents, etc. In my position, I have used AI to create documents for statistics. For example, I used Co-pilot to create a graph of the number at the exit gate of the library at the start of each morning and when we close the service desk at night. This provides as with a number of how many library users have used the library during the day and overnight, as we have 24/7 access for students and academic staff. I record the gate number in the morning and afternoon each day in an excel sheet for a month. At the end of the month, I use AI to collate the gate numbers and to create a graph to demonstrate the different times of the day and week when the library is busiest with library is not as busy. This has helped the library team to know when to staff the service desk and when not staff the service desk but have a buzzer for library users to use to inform staff that a library user needs assistance, and we can go out to assist them with their needs.
-
I feel AI has some positive outcomes when monitored continually…relying on human decisions for important details. As with all new technology I believe caution and monitoring need to be in place before releasing AI into our systems without knowing the complete outcome.
-
“Do you feel any hesitancy around embedding AI tools in your workflows?”
Yes, because I have not yet seen the value-add for the regular tasks that I do. However, I’m sure there is probably a strong use case for atypical work tasks that involve higher levels of analytical work that I normally do. Having said that, I would need higher confidence in the ability for the existing AI tools I have institutional access to before I start pursuing that more in earnest.
-
Exactly. Like one of the examples in the article: “In collections analysis, for instance, AI can compare spreadsheets of titles or suggest materials for weeding based on factors like circulation numbers, publication date, and classification area”. I’m sure it COULD, but spreadsheet filters are faster, and don’t need paragraphs of custom instructions in order to just give me a list without extra fluff.
-
-
We have only implemented AI in a few specific cases. The one case where it’s been most successful is our archival videos and oral histories. We decided early on that in order to make these large quantities of audiovisual materials accessible in any reasonable amount of time, we would AI for transcripts. Several clips were selected as our test set and then we evaluated the results from several different AI models. In the end, we chose Rev as it performed best and the materials would remain within its closed environment. All transcripts are reviewed by students and/or staff prior to being made available on our IR. It’s actually created an opportunity for collaborating with a Digital Humanities history class where they’ll be working on several of the oral histories, reviewing and revising the AI transcripts according to our style guide.
-
We use Copilot and I use it in my e-mail. every morning it prioritizes what emails are the most important and the ones that need a response. We also use copilot for online meetings. I still feel a little uneasy about using AI but I am trying to use it more and take classes and webinars on AI to feel less uneasy.
-
In my small library, we utilize the applications and new tools offered by publishers and developers of academic and research information systems. This requires preparation, a process we undertake with genuine interest and enthusiasm.
We look forward to—and actively seek—opportunities to contribute to institutional goals with greater determination, moving beyond the role of mere support staff providing traditional services. Members of the academic community approach us with questions about AI in various ways; consequently, I believe that not only does an opportunity exist, but this professional engagement with AI is already taking place—albeit informally. This encourages me to keep going! -
My academic library approaches AI adoption through a balanced lens of cautious innovation and value preservation, utilizing vendor-supported tools for specialized tasks while maintaining human oversight for quality assurance, especially for our students. Institutional adoption at my University hinges on factors like budget constraints, data privacy, output accuracy, while also driven by a hesitancy surrounding algorithmic bias, loss of cultural nuance, and shifting staff roles, and user sentiment.
-
My library has not implemented AI and is, to my knowledge, not considering AI tools. I could see how it would be useful for a public or larger library, but we are a small and specialized library so it would not be worth the cost.
-
Ai adoption in libraries isn’t straightforward because a lot of aspects around AI use in general (for example, ethics) and about particular issues found in each library (for example, readiness of staff and users, specific processes) are complex, they are not straightforward themselves. An audit of needs, priorities, and readiness is necessary.
-
Our library hasn’t implemented an AI tool, but we use AI assisted elements in our day-to-day work
-
Our library hasn’t implemented an AI tool, but we use AI assisted elements in our day-to-day work
-
AI training is important to build staff skills and confidence.
The choice should depend on institutional needs, ethics, privacy, and resources.
A controlled evaluation can help libraries adopt AI safely and effectively.
The module increased my interest in AI, but more practical examples are needed.
-
My library hasn’t implemented an AI tool, but we use AI assisted elements in our day-to-day work. We are considering a tool, but right now it is in the initiation phase. One of the main concerns is, should we come up with something new or should we use already existing AI Tools. I am exciting about AI tools in the workflow. Many things have changed over the years, and this is another thing that I believe libraries should get ahead of and play an integral and how it is implemented in the library.
-
Our library (the V. I. Vernadsky National Library of Ukraine) is already actively using AI-based tools to streamline processes. In particular, we have implemented a specialized anti-plagiarism program (StrikePlagiarism) that employs AI and machine learning algorithms to check research articles, academic papers, and dissertations. Librarians have undergone training to correctly interpret similarity reports, as AI helps identify matches, but the final determination regarding the presence of plagiarism rests with the expert. This has significantly reduced the time spent on routine text screening and enhanced the quality of our academic workflow.
Take me to the Table of Contents.
Leave a Reply