Nirmal Ghimire Advances Teacher Education with AI

Dr. Nirmal Ghimire, assistant professor of Teaching English as a Second Language at UNCW’s Watson College of Education, was recognized at UNCW’s Research and Innovation Celebration in September for his outstanding record of peer-reviewed journal publications. His work explores how artificial intelligence (AI) and machine learning can be used as research tools to analyze educational data and uncover insights that inform teacher education and help close student achievement gaps.

“The big-picture goal of my research is to fundamentally rethink how we prepare teachers by using data to uncover what truly works in teacher education,” Dr. Ghimire said. “What drives me is a curiosity about how we can better prepare teachers to make a measurable difference in students’ learning, grounded in evidence. I am motivated by the idea that decades of educational research and large-scale data already exist; our task is to connect and interpret them meaningfully.”

Through his Artificial Intelligence for Data-Enhanced Teacher Preparation (AIDE) initiative, supported by a UNCW start-up grant, Dr. Ghimire is developing AI-driven tools to analyze educational research. Working with a graduate assistant from Computer Science, he created a database of more than 2,000 peer-reviewed studies on effective teaching strategies for multilingual learners. The project lays the groundwork for identifying evidence-based patterns that can inform teacher preparation and professional development. Grant proposals to advance this research to its next phase are currently under review.

In a recent study published in Large-Scale Assessments in Education, Dr. Ghimire and co-author Dr. Kouider Mokhtari analyzed data from more than 600,000 students in 79 countries participating in PISA 2018. Using Random Forest modeling, they examined how students’ metacognitive reading strategies—such as summarizing, identifying key ideas, and evaluating information—relate to reading achievement. They found that students who use deeper learning strategies tend to perform better, while surface-level approaches, such as copying text, are linked to lower scores.

The study has gained international attention, ranking among the top 10% of research outputs tracked by Altmetric and leading its field in large-scale educational assessment.

His most recent publication, a book chapter co-authored with Dr. Mokhtari, Thinking with Machines: Leveraging Artificial Intelligence (AI) to Foster Metacognitive Reading Comprehension, explores how teachers and student can use AI productively and ethically in the classroom. At a time when many educators are grappling with students’ unsupervised use of AI tools, the chapter provides models for using AI as a scaffold for deep reading and reflective learning.

As a multilingual educator, Dr. Ghimire says he’s motivated by a desire to bridge research, technology, and classroom practice. He chose a career in Teaching English to Speakers of Other Languages (TESOL) because he wanted to prepare teachers who not only understand the linguistic and cognitive challenges multilingual learners face, but who also empathize with them and teach in ways that recognize their strengths and potential.

“TESOL is, at its core, about language and how multilingual learners make sense of text,” he explained. “Reading comprehension is central to that process. Many standardized tests are heavily text-based and use complex academic language, so they often measure language proficiency as much as content knowledge. This can make multilingual learners appear to struggle academically when, in fact, they are facing linguistic barriers to demonstrating what they know.”

While Dr. Ghimire’s early research centered on TESOL and reading achievement, his work has expanded to explore how linguistic, cognitive, and instructional factors interact to shape student success across disciplines and populations.

“Educational data are becoming increasingly complex, multidimensional, and voluminous,” he said. “Traditional statistical models often assume linearity, but real-world learning doesn’t always work that way. Machine learning allows us to see patterns that traditional tools can’t capture, offering a deeper understanding of how teaching and learning really happen. For me, that’s the next step in improving educational research and teacher preparation.”

Looking ahead, Dr. Ghimire sees his work through the Artificial Intelligence for Data-Enhanced Teacher Preparation (AIDE) initiative as a lifelong pursuit. “It’s a big, complex, and long-term goal—one I expect to work on for the rest of my professional life,” he said. “If I can help us take even the first few steps in that direction and generate meaningful insights along the way, I’ll consider that a success.”