Artificial Intelligence and Assessment Practices in Mathematics: Views of Nepali Mathematics Teachers

Source abstract

Background: Artificial intelligence is rapidly entering classrooms worldwide, yet little is known about how mathematics teachers in low-resource settings such as Nepal view its use in student evaluation. Methods: This study used a convergent parallel mixed-methods approach within a cross-sectional survey design. Using purposive and convenience sampling, a structured questionnaire combining demographic items, closed questions, five-point Likert statements, and open-ended prompts was administered to 50 mathematics teachers in community (public) schools across the urban, semi-urban, and rural areas of Bhaktapur district in the Kathmandu Valley. The closed and Likert items were analyzed with descriptive statistics and the open-ended responses through thematic analysis, and the two strands were then merged against the study’s objectives. Results: All respondents had heard of these tools and 84 percent had used them, most commonly ChatGPT and Gemini. Teachers strongly agreed that these tools can help them to produce good quality questions, and strongly disagreed that they can help to identify struggling learners, but was weakest on the statement that there is a need for clear national guidelines. The majority of respondents thought the technology saves time, but were not certain that it is able to judge reasoning and proof. The most common barriers were lack of training, lack of devices, and unreliable internet while the most common ethical concerns were copying and over-reliance. Almost every teacher expressed a desire to use the technology, given the opportunity to be trained and provided resources. Conclusion: Nepali mathematics teachers' attitudes are cautiously positive, seeing the technology as a supportive assistant to the routine work of evaluating students and not as a substitute for professional judgement, provided they are provided with training, guidelines, and equal access. Novelty: The study provides the first district-level, survey-based picture of Nepali mathematics teachers' perspectives on AI-supported assessment, linking readiness, trust, and equity within a single low-resource South Asian setting (Bhaktapur, Kathmandu Valley).