This article studies the application of large language models (LLMSs) such as ChatGPT, Claude AI in developing students' self-study and self-research capacity when studying the Statistics Probability course. Large language models (LLMSs) support solving exercises and questions and help students develop critical thinking and discover new knowledge through interacting with the system. By comparing the ability to solve different types of exercises, the article proposes effective self-study processes, from preparing before class to doing exercises after studying. Research results show that LLMSs not only help improve self-study ability but also support students in correcting mistakes during the math problem solving process, contributing to improving students' learning efficiency in this module.