The rapid integration of Artificial Intelligence (AI) into language education has introduced a paradigm shift from traditional computer-based tools to sophisticated, adaptive environments. This study investigates the relative efficacy of Computer-Assisted Language Learning (CALL) versus Mobile-Assisted Language Learning (MALL), specifically focusing on their impact on standardized testing outcomes for English as a Foreign Language (EFL) learners. Utilizing a randomized pretest-posttest experimental design, 120 participants were divided into three groups: an AI-integrated CALL group using desktop computers, an AI-integrated MALL group using smartphones (palm devices), and a traditional teacher-led control group. The intervention employed an AI-driven platform featuring diagnostic testing, score predictions, and personalized learning pathways. Over a full academic semester, the study measured performance gains in the TOEIC Listening and Reading tests. Statistical analysis via paired samples t-tests and one-way ANOVA revealed that while all groups demonstrated significant score increases, the AI CALL group significantly outperformed the control group in both sections and surpassed the AI MALL group specifically in reading comprehension. These findings suggest that while the portability of "palm" devices supports learner autonomy, the technical affordances of "desktop" computers such as larger screen size and interface stability—provide a superior environment for complex literacy tasks. The study concludes with pedagogical recommendations for integrating AI across different hardware to optimize standardized test preparation.