This study investigates the effectiveness of Artificial Intelligence (AI) learning using Teachable Machine in improving vocational high school (SMK) students’ understanding of classification concepts. The research employed a quasi-experimental design with a nonequivalent control group involving two classes: an experimental class that learned AI classification concepts through Teachable Machine and a control class that received conventional instruction. The learning process in the experimental group emphasized hands-on activities, including data collection, labeling, model training, testing, and evaluation, enabling students to directly experience the workflow of machine learning classification. Data were collected using pretest and posttest instruments designed to measure students’ conceptual understanding. The results of independent sample t-test analysis revealed a statistically significant difference in posttest scores between the two groups (p < 0.05), with the experimental group achieving higher mean scores and greater learning gains. These findings indicate that integrating Teachable Machine into AI instruction enhances students’ conceptual comprehension by providing interactive, visual, and experiential learning opportunities. The study concludes that AI-based learning using Teachable Machine is an effective instructional strategy to support the development of classification concept understanding in vocational education contexts