| dc.description.abstract | Scientific inquiry is a problem-oriented learning method, but teacher guidance is crucial for its effectiveness. Therefore, this study combines an online scientific inquiry platform with a Generative AI-based chatbot to establish an pedagogical agent called the ＂AI-Tutor,＂ which can guide students′ higher-order thinking. Acting in the role of a teacher, the AI-Tutor guides students through scientific inquiry, investigating the effects of its assistance on students′ answer performance, learning effectiveness, inquiry behavior, and self-efficacy.
This study recruited 62 high school students from northern Taiwan to participate in an inquiry activity on the CoSci online scientific inquiry platform. The activity focused on the ＂center of mass and momentum＂ unit in high school physics and included computer simulations and multiple inquiry questions that students needed to complete. The students were divided into two groups: the experimental group, which was guided by the AI-Tutor (33 students) , and the control group, which conducted the inquiry activities independently using related physics materials (29 students). This study collected pre- and post-test data on both physics conceptual test and self-efficacy questionnaire. Students′ answers on inquiry questions and operational behaviors during the inquiry activities were recorded. Additionally, the interactions and dialogues between students and the AI-Tutor were collected.
The results showed that both groups significantly improved their post-test scores in the physics conceptual test and their self-efficacy after the activity. However, students assisted by the AI-Tutor performed significantly better on inquiry activity questions than the control group and showed greater improvement in their post-test physics concept scores, indicating that the guidance of the AI-Tutor helped students construct scientific concepts. However, due to the short duration of the activity, the assistance of the AI-Tutor did not have a significant impact on self-efficacy. In terms of inquiry behaviors, students guided by the AI-Tutor participated more actively and answered inquiry questions significantly more frequently than the control group. They were also more strategic during the inquiry activities. Students in the control group were more directionless and switched between different inquiry questions significantly more frequently than the experimental group. Moreover, analysis of the dialogue interactions revealed that the feedback from the AI-Tutor provided support when students needed it, and its questions prompted students to think critically. In summary, in an online learning environment, the assistance of the AI-Tutor can effectively help students engage in scientific inquiry, providing timely instructional support to enhance their learning performance and understanding of physics concepts. | en_US |