Abstract:

The growing use of AI tools in higher education creates new learning opportunities. However, because AI systems can generate answers that appear correct but may be inaccurate or incomplete, prior knowledge is necessary to critically assess their reliability before accepting them as valid. This is particularly important in engineering disciplines, where terminology and concepts are highly specific.

In this study, 4th-year students enrolled in the Food Industries intensification of the Bachelor’s Degree in Agri-Food Engineering at the Universitat Politècnica de València completed several activities designed to examine the use of AI tools in learning and to promote critical thinking. First, working in pairs, students answered the same three true/false questions related to the course content at two different time points: during the first class of the semester, before the introduction of the hot-air drying unit operation, and during the last class, after all theoretical and practical concepts had been covered. In both cases, students were asked to justify their answers, compare them with those generated by an AI tool, and indicate whether the responses coincided, whether any part of the AI explanation was unclear, and whether they would change their answer if it differed from the AI output. At the beginning of the semester, students were also asked to identify the knowledge required to provide an accurate answer. In a second activity, students solved a calculation-based case study using traditional equations and tools related to the topic, and then asked an AI assistant to solve the same problem. They were required to report all prompts used until the AI produced the same result as their own calculation. Finally, students completed a five-question questionnaire administered via Google Forms to report their perceptions of how these activities contributed to the development of critical thinking and responsible AI use, as well as the usefulness of AI tools for learning the subject.

Students’ responses to the activities were not as satisfactory as expected. Many appeared to accept AI-generated answers without questioning them, often assuming they were correct even when they lacked sufficient knowledge of the subject. In addition, the overall level of subject mastery did not substantially increase from the beginning to the end of the semester. This may be explained by the fact that not all students kept up with the course content throughout the semester, and the final assessment takes place approximately one month after the end of classes. Despite these results, a large number of students reported having understood the importance of prior knowledge for formulating effective prompts and for critically evaluating the responses generated by generative AI tools. Many also indicated that the activities helped them become aware of the limitations of these tools and of the need to review and verify their outputs. Furthermore, students recognized that generative AI should be used as a complementary support rather than as a substitute for reliable sources or for the learning process itself. Overall, students perceived AI tools as more useful for addressing questions related to the theoretical content of the course than for solving practical exercises or case studies.

Keywords:

Students' background, blended learning tools, critical thinking.