The study aimed to evaluate the Artificial Intelligence-assisted Smart Counseling System in supporting the detection and classification of adolescent behavioral difficulties and to identify changes in students' behavioral conditions following the intervention. The study used a quantitative pretest–posttest control group design with 666 junior high school students in Medan City: 333 in the experimental group and 333 in the control group. Data were collected using the Strengths and Difficulties Questionnaire and analyzed by comparing pretest and posttest scores, as well as classifying them using the Support Vector Machine method. The results showed that the experimental group's mean Strengths and Difficulties Questionnaire score decreased from 21.00 (standard deviation = 12.469) at the pretest to 18.05 (standard deviation = 8.975) at the posttest. However, the analysis indicated that the difference between the groups at the posttest was not statistically significant, F(2,663) = 1.578, p = 0.207. Therefore, the evidence was insufficient to conclude that the Smart Counseling System was significantly more effective than the control condition. In contrast, the Support Vector Machine analysis showed strong classification performance, with accuracy increasing from 98.50% using the original features to 99.25% after feature engineering for classifying students into the Normal, Borderline, and Abnormal categories. These findings indicate that the Smart Counseling System has potential as a supporting system for the early detection and classification of adolescent behavioral difficulties; however, its effectiveness as an intervention requires further investigation through analyses of time-by-group interactions, effect sizes, and more comprehensive model validation. Therefore, the Smart Counseling System is better positioned as a decision-support tool for counselors rather than a replacement for counselors or a diagnostic instrument.