This study determined whether creating science storybooks using generative artificial in telligence (AI) affects the computational thinking and self-efficacy of elementary school students. For this purpose, 41 sixth-grade students from elementary school A in Gwangju, South Korea, participated in six sessions of science storybook creation using generative AI. Although students were interested and actively participated in creating science storybooks using generative AI, they struggled with the coherence of the story and the consistency of the images. Pre- and post-tests were conducted before and after the science storybook class, and the following results were analyzed. The results were as follows. First, the computational thinking of the students positively changed in various areas such as abstraction, decomposition, algorithmic thinking, evaluation, and generalization. Second, the self-efficacy of the students positively changed in the areas of task difficulty preference, self-regulatory efficacy, and self-confidence. The significant improvement between these tests was most likely due to the feeling of accomplishment in the challenging tasks. Altogether, the findings indicate the need to introduce and improve AI-based learning programs in educational settings, while creating an environment in which students are actively engaged and self-directed in their learning. As a follow-up study, the researcher aims to compare the impact of using generative AI on the computational thinking and self-efficacy of students at different grade levels and examine the long-term effects of project-based learning through this innovative approach.