This study investigates the impact of AI-assisted language teaching tools on students’ speaking performance compared to traditional textbook-based methods. The research employs a quasi-experimental design, involving both an experimental group and a control group of students. Initially, all participants underwent a pre-test using IELTS speaking questions, assessed on semanteme, pronunciation, and fluency under the assistance of AI speech evaluation technology, with results analyzed via SPSS 26.0 to ensure baseline parity. In the mid-research stage, the experimental group received instruction through an AI platform, while the control group used conventional textbook materials. The study’s findings reveal significant improvements in the experimental group’s post-test scores, indicating the efficacy of the AI-assisted approach. However, the research’s generalizability is limited due to the single-school sample and the lack of further validation for broader applicability. Additionally, the study did not delve into performance variations among students of different proficiency levels within the experimental group, an area suggested for future research.