Generative AI is beginning to change how people usually translate text, and nowadays college students use it daily to finish assignments. The present study compares ChatGPT to Google Translate in academic settings. In collecting data for this research, we used 161 undergraduate students who attend at the university where this research was conducted and taking translation course. Three genres were chosen: religious, media, and literary. Five parameters were used to measure the outputs including punctuation/diacritics, use of connective words, sentence structure, overall accuracy and culturally grounded metaphors. Each translation was graded by subject matter experts and students stated their preference through an official questionnaire. ChatGPT scored higher than Google Translate on all parameters. The averages for ChatGPT were 4.25 for media texts, 4.28 for religious texts, and 4.15 for literary texts. The averages for Google Translate were lower at 3.78, 3.42, and 3.59 respectively. Paired-samples t-tests indicated that these differences were significant (p < .001) with a large effect size (η² = .44 - .72). Student-reported preferences leaned toward ChatGPT 87% of the time. This was predominantly due to ChatGPT producing more coherent sentences that maintained the register of the source text, especially when it came to religious text. Google Translate consistently provided literal translations and failed to translate the intent of metaphors. We conclude that generative AI has the potential to be a helpful tool for multilingual accessibility and education if used responsibly and under certain conditions so that students can gain translation skills. Longitudinal studies about how this technology will affect education should be explored (pp.42-51).