Abstract:

This study presents an educational innovation implemented in the bachelor’s degree in Maritime and Port Operations at the Technological University of Panama, aimed at evaluating the impact of generative AI-GPT tools on student learning and performance in the course Port Machinery and Equipment. A quasi-experimental comparative design was applied between two cohorts: a 2023 control group without AI support and a 2024 experimental group granted full autonomy to use GPT-based tools with technical documents as contextual inputs. Students interacted freely with AI systems to prepare for a summative evaluation, constructing their own prompts and validation strategies. Results show that while the median score remained comparable between years, the 2024 cohort exhibited greater score dispersion, reflecting highly individualized learning pathways. This variability appears linked to differences in students’ ability to interpret context, formulate prompts, and critically assess AI-generated responses. The experience suggests that autonomous AI use fosters metacognitive skills, self-regulated learning, and inferential reasoning, yet also introduces inequities when the instructional design does not standardize the AI interaction process. The findings highlight the need for guided AI scenarios, teacher-designed GPT tutors, and structured prompt-based scaffolding to ensure equitable and pedagogically consistent outcomes in AI-supported technical education.

Keywords:

Generative AI, GPT tools, educational innovation, autonomous learning, maritime-port education, Moodle.