The relationship between adad (numbers) and madud (counted nouns) in Arabic is governed by complex and strict grammatical rules. Inaccuracies in applying these rules, such as gender agreement (masculine or feminine), numerical forms, and sentence structure (i‘rab), can affect meaning and communication accuracy, especially in formal or religious texts. Google Translate, as one of the most popular automated translation tools, faces significant challenges in handling grammatical structures like adad-madud. Its limitations lie in recognizing broader contexts and applying deep grammatical rules. This study aims to explore how Google Translate can improve translation accuracy in the context of adad-madud and understand the weaknesses that impact its translations. The research focuses on identifying errors, such as gender mismatches between numbers and nouns, incorrect numerical forms, and errors in sentence structure application. As an initial step, this study compares Google Translate's outputs with the grammatical rules established in Arabic. The research employs a qualitative descriptive method, using data collection techniques through translation trials. Various combinations of numbers and nouns in Arabic were input into Google Translate to analyze the results. The research results indicate that Google Translate still frequently makes errors in translating adad-madud, particularly in determining gender, numerical forms, and the grammatical position of adad and madud in the context of the sentence. This research addresses a gap in previous studies, which focused more on general, idiomatic, or religious text translations. By specifically examining adad-madud, this study provides new insights into the challenges and limitations of automated translation tools and offers recommendations for developing more accurate translation algorithms in the future.