Problem solving and responsible GenAI use in assessment: students’ reported experiences in open and distance learning

Abstract

Generative artificial intelligence (GenAI) is increasingly used by students to obtain explanations, generate ideas, improve written responses and prepare assessment tasks. However, limited evidence exists on how students in open-distance alternative-access programmes experience GenAI in mathematics assessment. This qualitative study examined how 27 Higher Certificate students at a South African open and distance learning institution described the role, benefits and limitations of GenAI while preparing fraction-related portfolio activities. Data were collected via a single open-ended questionnaire item, completed voluntarily after all assessment activities had been submitted. The responses were analysed using reflexive thematic analysis. Four overlapping themes were identified: support for understanding and approaching tasks, support for constructing written responses, conditional trust and evaluative engagement, and non-use, resistance and ambiguous engagement. Students reported that GenAI assisted with clarification, examples, organisation, and language, but they also described outputs they perceived as inaccurate, overly general, or insufficiently explained. Ten participants expressed concerns about reliability or described checking, adapting, limiting or rejecting GenAI use, although the study did not verify the quality of these practices. The findings suggest that the educational value of GenAI is shaped by students' capacity to exercise judgement and retain responsibility for the reasoning represented in their work. Verification is therefore positioned as a potentially important epistemic practice in GenAI-supported mathematics assessment. The study recommends explicit discipline-specific guidance and assessment designs that make students' mathematical reasoning visible.