Using Generative AI for Microcontroller Programming:

Effects on Experienced and Inexperienced Programmers

Abstract

With the increasing need for agricultural college graduates to understand applications of microcontrollers while having little programming experience, this study examined the effects of a generative AI microcontroller programming activity on undergraduate agriculture students’ attitudes and self-efficacy in relation to their level of programming experience. For the treatment in the one-group pretest-posttest research design, students prompted ChatGPT to write an Arduino program and observed the successful operation of a circuit. Results indicated the instructional treatment produced statistically significant posttest increases for both attitude and self-efficacy for students without prior programming experience, and significant increases in self-efficacy for students with programming experience. With all students successfully programming their Arduino, we concluded that ChatGPT is an effective tool for microcontroller programming for students with and without previous programming experience. When posttest scores were adjusted based on pretest scores, no significant effect of the instructional treatment on attitude or self-efficacy between students with and without previous programming experience was identified, making the instructional treatment equally effective for students with and without programming experience. Based on the results, we believe teaching with the use of generative AI for microcontroller programming has the potential to increase the number of agriculture students and graduates capable of developing and using these powerful embedded computing systems in their academic and career pursuits. Agricultural systems technology programs not currently teaching microcontroller courses because of a lack of student or instructor programming expertise should consider offering instruction by incorporating effective generative AI prompt engineering into these courses.