Background This study discusses the effectiveness of artificial intelligence in one-to-one psychological intervention system for middle school students. The purpose of this study is to determine whether middle school students have enough trust in AI machines, to explore their internal psychology, and to evaluate the intervention effect of AI in middle school students' one-to-one psychological intervention system. Subjects and Methods 76 eighth grade students from two parallel classes in the same grade were randomly assigned to the same teacher. The research tools used include two sets of autonomic nerve detectors produced by South Korean medicore company and two sets of self-developed artificial intelligence one-to-one psychological intervention systems, including two artificial intelligence robots Xiaoxin and two emotional bracelets. 36 students in the intervention group received one-on-one intervention of artificial intelligence robot Xiaoxin, and t-test was conducted on physical stress, psychological stress, stress index, sympathetic nerve activity, stress resistance and average heart rate. Test participants can effectively alleviate psychological stress and prevent depression. The data showed that the DASS-21 comprehensive score of subjects participating in the test was 1.7 times that of subjects not participating in the test. In contrast, without participation, they are more likely to show depression, anxiety and stress, and tend to be passive in simple coping styles. The improved method can effectively alleviate the psychological pressure of employees and improve the ability to actively deal with pressure. Results Except for the four indexes of nerve stress and sympathetic nerve activity, P < 0.001. After the intervention, the average heart rate, physical stress and stress index decreased significantly, while the anti stress ability increased significantly. Statistical analysis shows that the one-to-one intervention of artificial intelligence robot Xiaoxin can effectively alleviate students' physical stress and stress index, and improve students' stress resistance. The intermediary effect analysis of emotional arousal found that the direct effect of anger on MCT fluency was significant, B = 3.80, P < 0.001, CI = [1.99, 5.62], and the indirect effect of arousal on MCT fluency was significant, B = 0.58, CI = [0.08, 1.66]; The direct effect of anger on MCT anger was significant, B = 6.08, P < 0.001, CI = [3.26, 8.89], and the indirect effect of arousal on MCT anger was significant, B = 0.80, CI = [0.08, 2.41]; The direct effect of anger on MCT harmfulness was significant, B = 0.42, P = 0.004, CI = [0.14, 0.69], and the indirect effect of arousal on MCT harmfulness was not significant, B = 0.03, CI = [− 0.03, 0.18]. The direct effect of anger on aut fluency was significant, B = 2.58, P = 0.032, CI = [0.23, 4.92], and the indirect effect of arousal on aut fluency was significant, B = 1.22, CI = [0.27, 2.62]); The direct effect of anger on aut novelty was significant, B = 4.60, P = 0.008, CI = [1.23, 7.96], and the indirect effect of arousal on aut novelty was significant, B = 1.79, CI = [0.42, 3.91] (see Figure 4D). The above results show that emotional arousal plays a partial mediating role in the influence of anger on the fluency and novelty of malicious creativity and general creativity, but it has no mediating role in the influence of anger on the harmfulness of malicious creativity. Conclusion Artificial intelligence one-to-one psychological intervention system is effective. Among the 38 students in the intervention group, this shows that most students have a high sense of trust in the artificial intelligence robot Xiaoxin. This sense of trust is enough to make the artificial intelligence one-to-one psychological intervention system play the role of psychological intervention. We should fully understand the important position of artificial intelligence robot in emotional factors, actively improve its role psychological ability, pay special attention to its impact on life attitude and healthy behavior, and be a healthy self-manager, coordinator and collaborator. According to the psychological characteristics of these groups, we should actively and flexibly organize colorful activities, strive to create a relaxed and harmonious living atmosphere, create a friendly and mutually beneficial emotional artificial intelligence relationship, encourage and guide these groups, adjust their psychology to the best state, and promote the healthy development of their good cognition and life.