<![CDATA[This study examined Self-Regulated Learning (SRL) behavior among EFL students in Pontianak, Indonesia, based on the frequency of use of Artificial Intelligence (AI) tools in independent learning outside the classroom. Using a comparative survey design, 235 participants were divided into three groups: low (n=44), medium (n=124), and high (n=67) frequency users. Data were collected using a closed-ended Likert-type questionnaire with a 4-point scale ranging from never to always to indicate SRL behavior. Data were then analyzed using One-Way ANOVA and Tukey's HSD post hoc test. The analysis revealed a significant difference in total SRL between groups (F(2,232)=16.11; p<0.001)), with high frequency users achieving the highest score (M=3.81, followed by high users (M=3.48), and low users (M=3.35). Per-phase analysis revealed a consistent inverted U-shaped pattern across phases. In the planning (forward thinking) phase a significant difference was found (F(2,232)=13.73; p<0.001) with moderate users (M=3.49) outperforming low (M=3.11) and high users (M=3.05). Moderate users consistently showed high scores. Post-hoc analysis confirmed that the main difference lies between moderate users and low and high users across all SRL phases. These findings indicate a curvilinear relationship between frequency of AI use and SRL, with AI playing its most optimal role at moderate levels. This study concludes that moderate intensity of AI use is associated with the best self-regulation behaviors, with implications for the need for balanced integration of AI in self-directed learning outside the classroom context.]]>
