What Can ChatGPT Do in the Library?

Imagine a library where patrons can get assistance with research at 3 a.m., effortlessly generate an essay draft, or explore complex topics through instant conversations. In the past few years, AI has gone from a futuristic idea to a practical tool that is used daily by millions around the world, including in libraries, making this vision a reality. But not without raising questions.

Product Overview

ChatGPT is an artificial intelligence language model created by OpenAI that generates human-like text based on a prompt entered by the user. It is programmed to scan a vast range of internet data to answer questions, give feedback, assist in writing, generate ideas, and simulate conversations. It is often used for drafting emails, coding, brainstorming, and creating content. But the program is not foolproof, meaning it can sometimes generate inaccurate information or biased responses. A message at the bottom of the screen reminds users of this, as seen in Figure 1.

ChatGPT has the potential to significantly impact libraries and how they operate, so it is important that librarians have at least a basic understanding of how it operates and is used. Even if they don’t use ChatGPT in their own workflow, librarians should understand it so they can more thoughtfully integrate AI into library services and, as experts on information ethics, take the lead in addressing the concerns and questions ChatGPT raises about data privacy, bias, and misinformation.

Applications in Academic Libraries

Academic libraries continuously adapt to new technologies to better meet the needs of students and patrons, and ChatGPT is no exception. While it cannot replace the expertise and critical thinking of librarians, it can serve to greatly complement their work.

Libraries worldwide are already integrating ChatGPT into their patron chat tools to answer basic reference questions and to redirect more complex ones to librarians. Below, in Figure 2 and 3, is a typical question a library might receive in its chat and ChatGPT’s response. ChatGPT can also help libraries provide 24-7 support for patrons through reference portals like LibAnswers, whose chatbot feature allows libraries to connect users with ChatGPT or similar tools. Librarians can guide users on how to frame effective questions.

Libraries can integrate ChatGPT with databases, catalogs, and research guides to help users find academic sources faster. They can train ChatGPT on specific collections, enabling it to offer guidance tailored to what a library has to offer.

This integration also enables ChatGPT to enhance engagement by providing personalized recommendations for resources based on user queries and interests and to assist in advanced research and data analysis by summarizing complex datasets and generating insights from large volumes of text. For example, if a university library is working on analyzing thousands of academic papers on climate change over the past 20 years, ChatGPT could help with things like summarizing research trends, extracting emerging research areas, and identifying a category for each article. Users could either directly upload files, copy and paste the text, or for this type of large-scale analysis, integrate ChatGPT with their research databases via application programming interfaces. Additionally, ChatGPT holds potential in cataloging and managing metadata, where it could help librarians describe and access resources, improving discoverability within library systems.

Some library users, especially undergraduates, may also find ChatGPT can help with research. It can be used for brainstorming ideas and topics, identifying key terms, and drafting outlines. For example, if someone wants to research an essay on gender roles and their portrayal within media, ChatGPT could help them narrow that topic down as well as provide guidance on how to structure the essay. It can also generate examples of annotated bibliographies and summarize lengthy academic articles or datasets to provide a patron a quick overview (Cox & Tzoc, 2023).

Librarians can support students using ChatGPT in all of these ways by conducting workshops on how the tool generates content, including its shortcomings, and offering demonstrations on how to use the tool effectively to refine research questions. Librarians can also teach students how to assess the credibility of the information ChatGPT provides and how to stay within the bounds of academic integrity while using ChatGPT.

Finally, librarians can use ChatGPT to automate routine tasks such as drafting emails, creating research guides, or suggesting key words for searches, freeing them to focus on higher-level responsibilities.

Challenges and Ethical Considerations

ChatGPT is one of many AI platforms that has emerged in the past few years. Its highly advanced capabilities—such as producing high-quality text, engaging in dialogical interactions with users, consistently improving the content quality, and tailoring its outputs to specific language styles—all raise ethical concerns.

One of the most prominent is that since ChatGPT produces human-like text, it can be difficult to distinguish proper attribution of authorship (Stahl & Eke, 2024). This issue comes up predominantly in schools where students have been assigned to write essays or complete text-based assessments.

Not only that, but ChatGPT does not always output accurate or reliable information. A 2024 study on using AI for library reference services found that ChatGPT recommended many imposter articles for which it made up fake citations. When asked why it did this, ChatGPT responded that as an AI model, it just tries to provide helpful information but can sometimes make mistakes due to limitations in its current training and the nature of the program (Yang, 2024).

Another concern is ChatGPT’s impact on the job market, particularly in fields like computer programming (Stahl & Eke, 2024). Privacy and data protection are also critical issues for the library community. ChatGPT processes user input to generate responses, raising questions about how this data is used, stored, and protected. Libraries must consider the implications of exposing patron queries to external AI systems and ensure that data privacy standards are upheld.

But the biggest challenge I see with ChatGPT right now is its environmental impact. Recent studies highlight that AI systems require a large amount of water to operate, both directly and indirectly. This includes water used for cooling data centers and during the manufacture of the necessary equipment. (George et al., 2023). On average, ChatGPT consumes a 500 ml bottle of water for every ten to fifty responses, and this amount is likely to increase as newer versions continue to emerge (Li et al., 2023). Extracting such large volumes of water from rivers and streams can disrupt natural water flows and potentially reduce the availability of water for human consumption and agriculture (George et al., 2023).

ChatGPT also generates large amounts of wastewater that contain a range of pollutants. When not properly treated, it can contaminate local water supplies and degrade aquatic habitats. Generating the energy that powers AI, as well as the production and transportation of hardware for AI systems, also creates concerning amounts of greenhouse gases (George et al., 2023).

ADA compliance is another vital consideration. ChatGPT supports text-to-speech functionality and offers multilingual support, which aids in inclusive education. Its integration with assistive technologies helps users with disabilities access information more effectively. However, continuous evaluation is necessary to ensure these features meet ADA standards and support all types of user needs.

Strategies for Responsible Integration

As librarians continue to navigate the challenges and opportunities that ChatGPT presents, they must adopt thoughtful strategies for its integration into library services that ensure ChatGPT—and all AI tools—complement rather than replace human expertise and maintain a library’s commitment to equitable access, privacy, and information literacy.

This means first making sure librarians understand ChatGPT’s capabilities and limitations. Providing professional development opportunities such as workshops or webinars can help staff build their skills and confidence when it comes to using this tool. Trainings should cover topics such as AI-generated misinformation, how to guide patrons in framing effective prompts, and troubleshooting.

Libraries should also empower their patrons to effectively use AI tools. Libraries can host workshops, provide online guides or tutorials, or offer one-on-one consultations to share information on topics like generating search terms, assessing credibility, and avoiding overreliance on AI. My university—Utah State University—has a number of Libguides available to patrons to help them better understand AI. One example can be seen in Figure 4.

It’s also important that libraries balance automation and librarian expertise. While ChatGPT can handle routine tasks and questions, libraries should prioritize maintaining human interactions for complex and sensitive inquiries. Librarians should position themselves as essential guides, helping patrons critically evaluate AI-generated content and use it responsibly.

Successful integration of ChatGPT into libraries also requires collaboration. Libraries should partner with IT departments to ensure that technical and security measures are in place and work with faculty and academic departments to create opportunities to use ChatGPT for classroom instruction and research support. Given the environmental and ethical concerns associated with AI, libraries should seek sustainable practices when possible. This might include choosing an energy-efficient AI platform or advocating for greener data center operations. Choosing smaller, more specialized AI models or open-source AI could reduce environmental impact. Some AI providers are reportedly investing in carbon-neutral and renewable energy-powered data centers, including Google’s Gemini and Microsoft’s OpenAI, although both companies still have a large carbon footprint.

Libraries can also ensure inclusivity by providing equitable access to AI tools for underserved populations.

Contracting and Pricing

ChatGPT has three plans for individual users as well as group plans for teams and organizations. The most basic plan is free for all to use, with or without an account. The free plan gives users access to a more advanced model, GPT-4o, for a limited number of queries in each five-hour period, after which they can continue with a more basic model, GPT-3.5, which has a knowledge cutoff of January 2022, meaning it doesn’t include information from beyond that date. The plan also includes about 15 minutes of talking to ChatGPT in “voice mode” each month.

The next plan up, “Plus,” costs $20 per month, and gives users access to GPT-4o, which includes real-time web browsing capabilities, allowing the program to retrieve the most current information available online.

The most advanced individual plan, “Pro,” is $200 a month. It provides everything offered in the Plus plan, unlimited access to o1 models (which offer improved reasoning and more accurate outputs), more access to voice mode, and access to o1 pro mode, the most powerful version of GPT. These features were designed to meet the needs of researchers and professionals who require high-level AI capabilities.

Teams of two or more can select a flexible plan for $30 per person per month billed monthly or $25 per person per month billed annually. Organizations can subscribe to an “Enterprise” plan, which includes unlimited GPT-4o, enterprise-grade security for data protection, advanced data analysis tools, and administrative controls for team management. Pricing information for the Enterprise plan is not publicly available.

Related Products

Several AI tools and platforms compete with ChatGPT, each offering various features and capabilities. Anthropic’s Claude is an AI tool that emphasizes safe and ethical AI interactions, catering to users concerned about bias and harm in AI outputs. Microsoft Copilot, in Office 365, excels at automating tasks within Word, Excel, and PowerPoint. There are also library-specific tools, such as EBSCO’s AI-powered search and Gale’s Digital Scholar Lab, that are tailored to enhance research and resource discovery in an academic setting. Perplexity AI is a conversational AI search assistant that pulls information from the web to provide users real-time updates.

Many of these platforms have a free plan with limited features, similar to ChatGPT.

Recommendation

As AI tools such as ChatGPT continue to evolve, their integration within library services presents both immense potential and complex challenges. To fully utilize ChatGPT’s benefits while mitigating its risks, libraries should approach implementation strategically. This includes investing in staff training, promoting AI literacy among patrons, and maintaining a balance between automation and human expertise. I recommend libraries take a proactive role in integrating ChatGPT by establishing clear policies and ethical guidelines, creating opportunities for collaboration with IT and other departments, focusing on accessibility for all, and continuing to educate librarians and patrons about the strengths and limitations of AI tools.

ChatGPT is not meant to be a replacement for the expertise and critical thinking of librarians, but rather a tool to complement their work. By implementing thoughtful strategies, libraries can make AI an ally in providing equitable access to information and fostering lifelong learning in the digital world.

References

Cox, C., & Tzoc, E. (2023). ChatGPT: Implications for academic libraries. College & Research Libraries News, 84(3), Article 3. https://doi.org/10.5860/crln.84.3.99

George, A. S., A.S. Hovan George, & A.S. Gabrio Martin. (2023). The Environmental Impact of AI: A Case Study of Water Consumption by Chat GPT. https://doi.org/10.5281/ZENODO.7855594

Li, P., Yang, J., Islam, M. A., & Ren, S. (2023). Making AI Less “Thirsty”: Uncovering and Addressing the Secret Water Footprint of AI Models (No. arXiv:2304.03271). arXiv. https://doi.org/10.48550/arXiv.2304.03271

Stahl, B. C., & Eke, D. (2024). The ethics of ChatGPT – Exploring the ethical issues of an emerging technology. International Journal of Information Management, 74, 102700. https://doi.org/10.1016/j.ijinfomgt.2023.102700

Yang, S. Q. (2024). ChatGPT: Unleashing the Power of Conversational AI for Library Reference Services. International Journal of Librarianship, 9(1), 109–115. https://doi.org/10.23974/ijol.2024.vol9.1.375