The urgency of using AI in the educational environment as a medium for optimizing student learning personalization in synchronous and asynchronous learning needs to be done to ensure that students experience improved attitudes, motivation, and learning satisfaction, which is reflected in student competencies. This study aimed to investigate the impact of ChatGPT use frequency on students' attitudes, satisfaction levels, and competence in higher education learning goals. The type of research employed was non-experimental quantitative research, specifically ex-post facto research. The sample size was 257 people, which was determined using the criteria of Issac and Michael (1971) and sampling distribution with a proportional random sampling technique. Data collection techniques that utilize questionnaires are effective in obtaining quantitative data. The PLS SEM analysis was used to determine the effect of exogenous latent variables on endogenous latent variables. The study revealed that reliability and satisfaction with ChatGPT positively impact learning outcomes, while the perceived impact on competence does not significantly enhance learning objectives. Increased usage frequency moderates these effects, diminishing the positive influence of perceptions, reliability, and satisfaction and shifting the impact on competence to negative and insignificant. These findings highlight the complexity of integrating ChatGPT into education, suggesting that initial positive perceptions may not be sustained with frequent use.