Partnering with AI in the Scholarship of Teaching and Learning (SoTL)

This guide explores how generative AI and AI-enabled tools can support instructors who want to study their teaching. It highlights practical uses of AI while emphasizing rigorous, invigorating, and responsible approaches to SoTL research.

Introduction

The Scholarship of Teaching and Learning (SoTL) invites instructors to approach their teaching with the same curiosity and rigor that guide their disciplinary scholarship and research. By turning questions about student learning into opportunities for inquiry, SoTL can (re)connect instructors with a deeper sense of purpose while opening new avenues for intellectual engagement, collaboration, and publication. It also brings an opportunity for renewal by introducing fresh perspectives and practices that can (re)invigorate the classroom. At its heart, SoTL has the potential to transform teaching from a routine responsibility into an ongoing process of discovery.

One of the challenges for instructors who want to conduct research on their teaching is the time commitment required. This evidence-based guide highlights practical ways to use AI to support SoTL work, identifying strengths and limitations of AI and emphasizing rigorous and ethical use. From conducting literature reviews to designing studies, analyzing data, and preparing manuscripts, AI tools can support instructors efforts to be time-efficient without compromising intellectual engagement in the research process and methodological rigor.

When using AI in your SoTL work, remember that, as a human being who is innovative, imaginative, and capable of self-reflection, you are the key to success. As Cooperman and Brandao (2024) point out, “While AI tools can assist in automating tasks and simplifying the writing process, it is important to remember that AI tools are not a substitute for human creativity and intuition. AI systems learn from pre-existing data sets and identify patterns, but they are not capable of ‘thinking outside of the box’ or making novel connections in the same way that humans can.” 

For further guidance on using AI for research, see also Northwestern Guidance on the Use of Generative AI.

Developing and Refining SoTL Research Questions

Having a clear, well-articulated research question is an important initial step in conducting research on teaching. This is because the study design and research methods that will be used in the study depend largely on the research question that is being asked. AI assistants can:

• Refine research questions and even suggest questions based on text and research literature provided.
• Identify strengths and weaknesses of a study design and make suggestions for overcoming the weaknesses.

Conducting Literature Reviews

SoTL research studies are always grounded in a thorough review of the literature. Literature reviews help instructors to:

• Develop a deep knowledge of the area that they are planning to research. 
• Understand whether and how well their research question(s) of interest have been addressed.
• Identify gaps in the literature and limitations of previous studies.

AI Tools to Identify Relevant Literature:

Several AI-powered search tools and assistants are available to make it easier to identify sources for literature reviews, identify gaps in the literature, and summarize key findings from sources. AI-powered search tools, such as Connected Papers, Consensus, Elicit, and ResearchRabbit, map existing research literature, identify patterns, and highlight research gaps.

AI Tools to Assist with Analyzing and Synthesizing Literature:

AI assistants, such as ChatGPT, Claude, and Google Gemini, can summarize articles, generate annotated bibliographies, and compare perspectives. Google NotebookLM can help summarize and synthesize key findings from sources by creating mind-maps and podcasts based on sources that the researcher uploads.

Designing Surveys, Focus Group Questions for SoTL Studies

Many SoTL studies use surveys, focus groups, and interviews to investigate students’ perceptions about innovative teaching approaches and courses. Like all research tools, they must be carefully designed and calibrated so that they will yield accurate, trustworthy data.

AI assistants can draft survey, focus group and interview questions. They can suggest questions based on a topic or description of the research project and study aims. However, providing the AI assistant with examples of questions that you have drafted will help it to design questions that align more closely with the goals of your project. In a comprehensive overview of the use of AI for survey development in health research Kuru (2025) explains how Large Language Models (LLM’s) and Natural Language Processing (NLP) can also be used to mine journals, reports and historical surveys to generate comprehensive pools of questions. 

Survey questions can be developed based on responses to open-ended questions that have already been collected. This process is often used when researchers want to design a survey to distribute to a larger number of participants to confirm interview/focus group findings. Learner (2024) reports that AI assistants are very effective in analyzing pre-existing data and suggesting questions based on the data.

AI-powered search tools can help identify published survey instruments that have been proven to be valid and reliable. When they align with your own research study aims, using questions from validated surveys is always the best option because validated instruments reduce measurement error in a study. Dozens of survey tools, such as Qualtrics, now incorporate AI and can suggest survey questions, survey designs and response scales.

Analyzing Data

Qualitative data:

Qualitative data analysis tools such as NVivo integrate AI in their software to code data, identify themes and patterns in data from interviews, focus groups and open-ended survey questions. Many also offer sentiment analysis, a computational technique that uses algorithms, which are often based on natural language processing and machine learning, to examine text data and identify patterns related to emotional tone, such as positivity, negativity, neutrality, or more specific affective states (e.g., confusion, enthusiasm, frustration). 

AI assistants, like ChatGPT, can suggest initial codes and emergent themes for analyzing focus groups, interviews, or open-ended survey responses based on the raw qualitative data. Alternatively, they can develop codes based on research papers that share frameworks and taxonomies that the researcher provides. AI assistants can also identify quotes from data to include in a paper and can create tables with themes and supporting quotes.

Zhang et. al (2025) emphasize that the results of qualitative analyses conducted by AI assistants, such as ChatGPT, depend on the quality of the prompt used. They summarize the characteristics of high-quality prompts and have developed a human-centric workflow/framework for prompt engineering for qualitative data analysis. Prompts include:

• Purpose of the study
• Desired outcomes
• Qualitative methodology that should be used
• Analytical process that should be used
• Format of the transcription inputs
• Description of how the output should be formatted
• The perspective that should be taken when doing the analysis

One of the criticisms of using GAI for qualitative data analysis is the lack of transparency in how it has conducted the analysis. To enhance transparency, Zhang et. al (2025) recommend strategies such as asking the AI Assistant to analyze each line of text independently, rather than doing an overall analysis of the input data.

Quantitative data:

• AI assistants can assist with data cleaning, such as identifying errors and inconsistencies in the data. 
• AI assistants can also provide support for statistical analysis by helping to choose statistical tests, testing statistical assumptions, and writing and explaining statistical code (R, Python, SPSS). 
• Journals increasingly require plain-language explanations of statistical analyses in the results section. AI assistants are an excellent tool for translating results of complex statistical analysis into plain language. 
• AI assistants can assist with the creation of data visualizations by suggesting the best graphs and other visualizations for data and by generating code to make graphs and plots.

Writing and Manuscript Preparation

One of the distinguishing features of SoTL is that the results of SoTL studies are shared publicly via journal articles, conference presentations and presentations for peers at department/institutional meetings etc. AI assistants can support all phases of the writing process but should not be used in place of a human author.

• AI assistants can help create outlines of papers and suggest structures.
• AI assistants can also translate abstracts into multiple languages or create lay summaries and edit and adapt content for different audiences (e.g., journals, conferences, workshops).
• AI assistants are helpful for providing feedback on writing clarity and suggesting improvements.

Formatting Manuscripts and References

Formatting manuscripts for submission and after they have been accepted can be one of the most time-consuming steps in the SoTL process. AI assistants can be great time savers in manuscript preparation as far as formatting text, tables and references according to manuscript guidelines.

Ethical Considerations for Using AI to Support SoTL Research

There are many ethical considerations in using AI to support SOTL research:

• Transparency: Document how AI tools were used in a SoTL project. Always consult journal guidelines for authors to determine if AI use is allowed in research and writing processes and how that use should be described in the manuscript.
• Integrity: In the guidelines for using AI tools, editors of Teaching and Learning Inquiry, the journal of the International Society for the Scholarship of Teaching and Learning emphasize the importance of integrity when using AI to prepare SoTL publications. They point out that AI tools cannot replace the author’s responsibility to produce original academic content, and that if an AI tool is used, the author(s) must ensure that the main arguments, research, and conclusions come from their own work and are not generated by AI. The guidelines include a very useful table outlining when it IS and IS NOT acceptable to use AI in every step of the SoTL process.
• Privacy: Most AI assistants and tools (unless in closed systems like CoPilot at Northwestern) use the data provided by users for training purposes. Always remove identifiers before uploading data to AI tools. When choosing an AI tool to support your research, “read the fine print” regarding how the data will be used by the company and who it will be shared with. Note that according to Northwestern Guidance on the Use of AI  University faculty, staff, students, and affiliates should not enter institutional data into any generative AI tools that have not been validated by the University for appropriate use and have explicit permission of the data provider.

• Use of text produced by GAI: Combrink (2024) urges users to “Use resources such as ChatGPT responsibly by rephrasing, repurposing, and reintegrating responses as they would information from other sources”.
• Reflexivity: Noting the importance of examining the researcher's role, assumptions, and influence on the research process in qualitative research, Kuru (2025) suggests that reflexivity becomes even more vital when integrating artificial intelligence (AI) methods into survey development. Kuru advocates for researchers to actively interrogate how their views on technology, efficiency, and innovation shape the design, analysis, and interpretation processes.