CROSS-LINGUISTIC EVALUATION OF AI-GENERATED TEXT DETECTION: A COMPARATIVE STUDY ON ENGLISH AND INDONESIAN USING PRECISION, RECALL AND F1 SCORE

In the age of generative AI, the line between human-written and machine-generated text is becoming increasingly blurred. This paper explores the performance of AI content detection systems across two linguistically and structurally diverse languages—English and Indonesian—through an empirical evaluation using 5,000 samples. The study evaluates detection outcomes using widely accepted performance metrics: precision, recall, and F1 score. Results reveal higher detection accuracy for English compared to Indonesian, due to linguistic complexities and dataset bias. This study underscores the growing importance of multilingual AI verification tools, especially in academic and regulatory environments.