AI Agent to Prepare Students for Final Assessments: A Strategy to Strengthen and Maximize Student Performance in Engineering Programs
Academic Article in Scopus
At Tecnologico de Monterrey, engineering students are required to take a final examination, the CENEVAL EGEL, a widely accepted option for certifying essential disciplinary knowledge and skills in Mexico. At this institution of higher education, the completion of this examination is significant for our future graduates, as it is a mandatory component of the graduation requirements. The utilization of generative artificial intelligence (Gen AI) holds considerable potential in the development of customized and effective practice exams, thereby facilitating enhanced student preparation for the CENEVAL EGEL examination. These tools have been identified as instrumental in improving student success and confidence. The advantages of this approach include the ability to customize content, access to a range of practice questions, immediate feedback, and the enhancement of confidence derived from prior experience with a similar examination. This study explores the impact of a personalized Gen AI agent that has been specifically trained in the knowledge areas of the CENEVAL exam. The artificial intelligence (AI) agent provides customized, practice-based feedback, with an emphasis on questions, with the objective of minimizing unsatisfactory results and maximizing outstanding grades. The experimental design involves a comparison between a control group (n=43) that did not utilize the AI agent and a pilot group (n=41) that did employ the tool. Each student in the experimental group completed over 160 questions, which encompassed four primary subject areas delineated by CENEVAL. The analysis showed consistent improvements in the experimental group after using the AI agent, with increases in performance on practical assessments and a reduction in the gap compared to the control group. Although the latter had a higher GPA, both groups achieved similar rates of satisfactory results on the final exam. Furthermore, students who used the agent reported a positive experience, highlighting its usefulness, quality of feedback, and effect on confidence. These findings suggest that generative AI can be an effective tool to support students' preparation for standardized assessments, especially for those starting at lower academic levels.
