The Impact of Using DeepL Artificial Intelligence on Students’ Writing Skills in an Indonesian Senior Highschool Context

This study looks at how eleventh-grade students at SMA Negeri 2 Singaraja's writing abilities are affected by the use of DeepL artificial intelligence. The study uses mixed methods with an explanatory sequential design; for quantitative analysis, a quasi-experimental non-equivalent control group design is used, and for qualitative insights, in-depth interviews utilizing the RASE model are conducted. Purposive random sampling was used to choose the samples, and two classes participated in the investigation. While the control group received traditional instruction, the experimental group received training via DeepL-based AI. Descriptive and inferential data analysis showed that the experimental group's mean score (81.36) was higher than the control group's (76.26). Furthermore, the inferential analysis demonstrated a significant difference in writing competency between students taught with DeepL AI and those without (Sig. (2-tailed) = 0.000, p < 0.05), supporting the alternative hypothesis that DeepL AI-enhanced instruction leads to superior outcomes. Interview findings further underscored DeepL's efficacy in enhancing writing skills by facilitating vocabulary expansion, offering grammar corrections, and suggesting alternative word choices for language learners. Additionally, inferential analysis showed a significant difference in writing competency between students who received DeepL-based AI instruction and those who did not (Sig.(2-tailed) = 0.000, p < 0.05), confirming the alternative hypothesis that better results are obtained from DeepL AI-enhanced instruction. The results of the interviews provided more evidence of DeepL's effectiveness in improving writing abilities by helping language learners expand their vocabulary, suggest other words, and provide grammatical corrections.