<aside> <img src="/icons/flash_blue.svg" alt="/icons/flash_blue.svg" width="40px" />

Project snapshot

This project explored how students and faculty use AI tools in the academic research process, including what motivates their use, which tools are favored, and where AI fits within existing research workflows. Through remote contextual inquiry, semi-structured interviews with diverse participants, and supplementary survey data, the study identified key use cases, perceived benefits and risks, and opportunities for integrating AI support into Library Search.

If you’d like to learn more, please email [email protected] to request more documentation.

</aside>

Key documents

Presentation Slides: How Researchers Use and Feel About Generative AI

Findings Report: AI Use and Expectations by Students and Faculty

The challenge

This project was initiated to address uncertainty around how generative AI tools are impacting academic research practices and what role the library’s digital services should play in the changing academic research tool landscape. While previous research had identified common pain points in the research material discovery process in Library Search, we wanted to know how researchers were overcoming these challenges and whether AI tools were part of those strategies. The project aimed to understand the real-world contexts in which students, staff, and faculty use AI for research as well as their motivations and concerns.

Objectives

• Understand the contexts in which students and faculty are using AI tools, including frequency of use, specific research phases where it is used, and motivations for use
• Identify the perceived benefits and risks, specifically beneficial applications and users' concerns and limitations
• Inform AI strategy by understanding the key user problems to mitigate, desired features, and desired user experience

Our approach

This project primarily consisted of user interview/moderated contextual inquiry with the target audience. In addition, supplementary data from the 2025 Library Search Benchmarking Survey was used to quantify some of the findings from the interviews.

• User Interviews/Moderated Contextual Inquiry: I conducted 11 user interview sessions with students, staff, and faculty at the University of Michigan. These interview sessions consisted of semi-structured interview questions as well as an interaction portion, where I asked them to show me how they completed research tasks with or without AI.
• 2025 Library Search Benchmarking Survey: In the winter of 2025, a large-scale survey (n=756 responses) measuring trends in Library Search was conducted. While the survey primarily focused on research habits and user sentiment towards Library Search, a handful of questions relating to AI use and sentiment (Qs 28 - 32) were included.

Findings

1. AI is primarily used for efficiency and overcoming initial hurdles, not core content creation or research

   1. Participants largely used AI tools to quickly grasp unfamiliar topics, brainstorm ideas, and save time on repetitive or administrative tasks. However, they did not rely on AI for generating original research content or conducting in-depth analysis, instead turning to traditional research tools like Google Scholar and Library Search for those core activities.
   2. List of tasks people do with AI:
      1. Background Information Gathering & Summarization
      2. Brainstorming/Coming up with Ideas for Getting Started
      3. Coding & Technical Assistance (with notable limitations)
      4. Image Creation
      5. Studying & Conversational Practice Aids
      6. Writing Assistance (Outlining, editing, rewording, etc.)

2. Trust in AI-generated information is limited and requires human verification

   1. While participants found AI helpful for overviews and summaries, they expressed consistent skepticism about its accuracy and reported routinely verifying AI outputs against trusted sources. This behavior reflects a conditional trust where AI serves as a starting point, but not a reliable endpoint in the research process.

3. Traditional library resources remain crucial, primarily for full-text access to high-quality materials

   1. Despite experimenting with AI, all participants continued to rely on Library Search and Google Scholar for access to peer-reviewed, high-quality sources. These tools were seen as essential for deeper research stages, citation validation, and retrieving full texts, which AI tools were not equipped to handle reliably.

4. Ethical concerns are prominent among users

   1. Participants raised multiple ethical issues related to AI use, including misinformation, academic integrity, environmental impact, and exploitative labor practices behind AI models. Some also intentionally limited their AI use out of concern that it might compromise their learning or diminish critical thinking skills.