The limit concept, foundational to calculus, is frequently associated with persistent student misconceptions that can be effectively diagnosed and remediated through rigorous assessment and targeted feedback. This study investigates the advantages and limitations of employing artificial intelligence (AI)—specifically Grok—in evaluating first-year undergraduate mathematics students’ mastery of the limit concept. Conducted as an applied mixed-methods investigation, the research involved 37 students from the Mathematics Education Department at Farhangian University of Isfahan. Data were collected via a researcher-designed questionnaire comprising seven items on limits; responses were digitized and submitted to the AI for analysis. Evaluation adhered to the CIPP framework, with emphasis on the process dimension, encompassing answer correction, scoring, ranking, error classification (conceptual, procedural, computational), and individualized feedback for students and the instructor. Quantitative results revealed no statistically significant difference between AI-generated and researcher-assigned scores. Qualitative content analysis identified key strengths—rapid processing, scoring precision, analytical feedback, and resource recommendations—alongside notable weaknesses, including deficiencies in graphical interpretation, overemphasis on final answers at the expense of procedural reasoning, and dependency on grammatically precise input for accurate recognition of mathematical notation. Findings underscore AI’s viability as a supplementary assessment tool in mathematics education, contingent upon enhancements in visual processing and process-oriented evaluation.