The relationship between the AI-assisted TYT Exam and the TYT Exam prepared by ÖSYM

The purpose of this study is to investigate the relationship between the Artificial Intelligence (AI)-assisted Basic Proficiency Test (TYT) administered in 2024 and the TYT examination prepared by the Assessment, Selection and Placement Center (ÖSYM). The growing prevalence of digitalization and AI technologies in contemporary educational environments necessitates a transformation in measurement and evaluation practices. Within this framework, determining the measurement capabilities of AI-assisted testing systems and comparing them to traditional assessment approaches represents a significant research problem. A correlational research design was employed in this study. The sample consisted of 233 students who took both the AI-assisted TYT and the official TYT administered by ÖSYM in 2024. The data collection instrument comprised a dataset of net scores obtained from the subtests of both examinations. For data analysis, several statistical techniques were utilized, including the Pearson Correlation Coefficient, the independent samples t-test, and Cohen's d effect size calculations. The findings indicate a significant and positively directed relationship between the AI-assisted examination and the ÖSYM TYT. Notably, high levels of correlation were observed in the Turkish, Basic Mathematics, and Science subtests. Descriptive analyses revealed that the mean Turkish subtest score in the ÖSYM examination was significantly higher than that of the AI-assisted test. Independent samples t-test results demonstrated that most of these differences were statistically significant (p < .001). Specifically, the difference observed in the Turkish subtest corresponds to a large effect size according to Cohen's d. For the Mathematics and Science subtests, the calculated effect sizes fell within the small to medium range. Although the difference in the Social Sciences subtest was statistically significant, its effect size was negative and small. Overall, this study suggests that AI-assisted assessment systems may serve as a potential alternative in higher education entrance examinations and offers insights into the ongoing technological transformation in educational assessment. The results provide valuable implications for both software developers engaged in educational technology design and policymakers responsible for shaping measurement and evaluation practices. Keywords: artificial intelligence, TYT, assessment and evaluation, AI-supported testing, correlation