Gemini-Based Generative AI-Chatbot for Mathematical Problem-Solving

There remains a limited body of research on the utilization of generative artificial intelligence (AI), particularly large language models (LLMs), for mathematical problem-solving in elementary school. In this study, which aims to develop a generative AI chatbot for mathematical problem-solving, the focus is on exploring elementary students’ acceptance of a chatbot named RAISA (Responsive Artificial Intelligence for Smart Assistance). This development research was conducted using the ADDIE (Analyze, Design, Develop, Implement, and Evaluate) framework. Google DeepMind’s Gemini model was chosen as the basis for the chatbot development. The chatbot was designed not to directly provide the final solution but to guide students through the stages of problem-solving. A pilot test was conducted involving ninety-eight elementary school students who participated in four math learning sessions using the RAISA chatbot. To determine their acceptance, a questionnaire based on the technology acceptance model (TAM) was used. The results indicated that students had positive responses regarding TAM aspects such as perceived usefulness, perceived ease of use, attitudes toward using, and behavioral intention to use. These positive responses from participants show the great potential of generative AI chatbots for learning, including math problem-solving.