PERCEIVED USEFULNESS AND EFFECTIVENESS OF AI TOOLS AND STUDENTS’ LEARNING OUTCOMES IN SCIENCE HIGH SCHOOL

Artificial intelligence (AI) tools are increasingly used by secondary students, yet localized evidence is needed to understand how students perceive their usefulness and effectiveness in relation to learning outcomes. This study determined the association between perceived usefulness and perceived effectiveness of AI tools and students perceived learning outcomes in knowledge, skills, and values among Grade 9 and Grade 10 students in selected public science high schools in the Division of Taguig City and Pateros, Philippines. A quantitative, descriptive, cross-sectional, non-experimental design was used. A researcher-developed questionnaire was administered to 290 students selected through stratified random sampling with proportional allocation. After screening, one persistent multivariate outlier was removed, leaving 289 cases for analysis. Descriptive statistics and multivariate analysis of variance (MANOVA) were employed, with Pillai’s Trace used because the homogeneity of covariance matrices assumption was violated. ChatGPT received the highest net agreement ratings for usefulness (94.12%) and perceived effect (88.58%), followed by Grammarly. Perceived learning outcomes were positive across all three domains, with knowledge receiving the highest net agreement rating (88.93%). MANOVA indicated significant multivariate differences associated with perceived effectiveness and perceived usefulness, and a significant interaction between the two perceptions. Students who simultaneously perceived AI tools as highly useful and highly effective reported the highest perceived learning outcomes. The findings support localized AI educational programs that integrate tool knowledge, practical skills, critical evaluation, and ethical responsibility rather than treating AI access as sufficient for learning.