Promising Applications of AI in Education Research

---

The Use of AI in Education Research Interview

The following interview features Seth Van Doren, a doctoral student at the University of California Irvine, and June Ahn, a professor at the University of California Irvine, talking about how AI can facilitate the research process.

---

“Which students will be most in need of tutoring next week?” “Which classroom is showing the most signs of social and emotional stress?” To answer such questions, school district staff and education researchers turn to AI-powered learning analytics to help identify and understand patterns in student behavior. You’ll see terms like “predictive analytics” or “educational data mining” used to describe this work. These researchers use data to:

• Predict student performance (including enrollment and persistence)
• Identify students at risk of dropout or course failure
• Predict student engagement and student satisfaction
• Offer general real-time warning systems and select students for socio-emotional or academic interventions

Much of this work relies on longitudinal administrative data or more granular data from student engagement with learning management systems. Learning analytics empowers you to make more effective use of “Big Data” sources such as the data created by learning management systems or online course software (which can contain tens of thousands of data points for each student).

However, recent scholarship also highlights limitations to the use of predictive analytics technologies, including a lack of transparency in algorithmic predictions, ethical concerns with applying risk scores to students, and the potential that learning analytics may supplant instructors’ judgement and expertise.

---

“What themes emerge in open-ended questions about our school’s response to COVID?” “How many school district discipline policies define conditions under which restraint can be used?” To answer questions like these, you can turn to AI in analysis of qualitative data. AI support for qualitative education research may include:

• Automated transcription
• Assistance with content analysis
• Automated annotation
• Identification of representative quotes
• Thematic analysis

Most current research does not use AI to replace human qualitative data analysis, but instead to support human researchers. And while AI may make qualitative research more efficient, there are risks in terms of data privacy, transparency, and research quality. This is of special concern for education researchers who need to consider student privacy from both an ethical and a legal perspective.

---

Yet another promising contribution AI has made to the field of education research is the development of tools to support literature review and meta-analysis. Generative AI tools can be used to help identify relevant literature, extract information from large databases of papers, and summarize findings.

---

You can use AI in your research operations for help with tasks like brainstorming, extracting information from research papers, scientific writing and editing, conducting peer review, and formatting citations.

Researchers use both general purpose chatbots (such as ChatGPT, Gemini, or Copilot) and tools designed specifically for research and writing (such as Grammarly or Jenni). These uses of AI to help researchers plan, think about, and conduct research can improve research efficiency and reduce burden. However, as with any use of AI, you must carefully consider concerns about accuracy, privacy, and plagiarism when choosing to use these tools.

While it is clear that AI holds substantial promise in educational research, you must also be aware of potential risks to data privacy, accuracy, and transparency. To make effective use of AI tools, we suggest three important best practices:

• Understand what your chosen tool or method does. Understanding how machine learning and AI models work is key to implementing them effectively and identifying possible sources of bias or error.
• Benchmark or check the results of AI-generated analysis with manually generated comparison data. By verifying the results of AI-generated conclusions, you can have greater confidence in the accuracy of your research.
• Think about AI risk systematically by using an AI risk framework. Before using AI tools (in both research and practice), teams should consider applying a risk framework to help identify and mitigate AI-related risks.

---

Footnote

[1] Note that there are also many applications of AI to pedagogy, but we focus here on AI as a research tool.