In terms of perceived ease of use (Items 7–10), most of the participants (82%) agreed that the performance of ChatGPT was fast and the interaction of ChatGPT was clear and understandable (nearly 80%). Moreover, there were 75,8% and 72.7% students who had intention to use because it was easy to create effective prompts to enhance the quality of translation.

However, when examining attitude toward using ChatGPT (Items 11–14), the results revealed the drawbacks, as 62.1% and 71.9% of participants had an intention to feel lazy and over reliant on AI tools in case of utilizing ChatGPT for translation assignments. This indicates that while students recognize the benefits of ChatGPT, they also acknowledge the risks of dependence on AI-generated translations.

Finally, the findings related to behavioral intention to use (Items 15–17), there was 72% and 68.9% of participants who tended to agree that ChatGPT would be beneficial in their translation careers and they would recommend ChatGPT to their friends for translation tasks, indicating widespread utilization of its effectiveness. However, the percentage of item 13 was the lowest, with 6.1% strongly disagreed and 17.4% disagreed. Although it is a valuable tool in translation, students do not integrate it into their regular study routines.

Research question 2: What are students’ challenges while using ChatGPT in translation subjects?

The interview questions were designed based on a questionnaire to investigate the challenges of students while using ChatGPT in translation. These difficulties were divided into three main groups: Cognitive risks, linguistic risks, and technical risks. Each group is analyzed based on the answers from participants.

 

Cognitive risks

At the beginning, some students used ChatGPT as a supportive tool in translation courses and they used it all the time to complete translation assignments. This reduced the ability to learn independently, as students were no longer actively involved in the translation process. This change not only affected learning outcomes, but also had a long-term influence on the development of learners’ language skills. Furthermore, it made students lose initiative in learning. Some students admitted that their level of dependence:

“Now I depend on ChatGPT about 50%. Initially, I translated assignments completely on my own. But after 2-3 exercises, I started using AI tool. Finally, I would just review ChatGPT's translations and only edit translations when mistakes were found.” (P7)

In addition, deadline is a big problem that leads to overreliance. Participant 6 shared:

“Sometimes, I feel dependent on ChatGPT, especially when using it to meet urgent deadlines. It's become a solution for last-minute assignments."

Speed and efficiency are key factors in academic context, from that students easily choose the optimal solution – using AI tools instead of developing their own assignments. However, this will make them lose their learning ability without the help of ChatGPT.

Some students admitted that the utilization of ChatGPT reduced their motivation, memory and creativity:

“The use of ChatGPT limits creativity, I just think about using it and I do not want to learn new vocabularies.” (P5)

 

Linguistic risks

1 Grammatical errors: While ChatGPT generates text rapidly and with accuracy at a basic level, it often has grammatical errors when applied to academic settings. Some students revealed that they had to adjust grammar and structure to meet assignment requirements while utilizing ChatGPT in translation:

“ChatGPT can translate quickly, but I have to correct grammatical errors.” ( P8)

“While translating from Vietnamese to English, I need to correct grammatical errors.” (P4)

Students without a grammatical foundation had to face this challenge, as they struggled to identify and correct errors. Over-reliance on ChatGPT translations without the necessary grammatical skills could lead to a reduction of effectiveness in developing translation skills.

2 Accuracy: Students reported that the translations generated by ChatGPT were not close to the original meaning and it was not coherent:

“ChatGPT is effective in speed but not completely accurate, I have to double or triple check the translations.” (P1)

“Using AI too much will lead to unnatural sentences or follow its structure.” (P7)

Particularly, in assignments requiring creativity such as literary translation or news, students found that ChatGPT failed to fully convey the emotion and meaning.

3 Technical risks: Cultural and contextual inappropriacy. In translation, especially when translating documents with cultural elements, understanding the context is extremely important. However, ChatGPT just produced general answers without detailed context information:

“ChatGPT does not support culture and context, I have to consult different sources.” (P1)

The lack of deep context and culture affected translation quality and caused serious misunderstandings when translating cultural or historical documents. This forced learners to consult additional books, lectures or other trusted sources to understand and revise the translation.

4 Challenges in prompt engineering and output accuracy: Many students revealed that if prompts were not clear, the output translation would not meet the requirements:

“ChatGPT does not provide the expected output if the prompt is unclear.” (P8)

This limitation forced students to learn how to generate the prompt in a detailed and structured way. In addition, students also had to try many times to adjust the command to make output usable, which was time-consuming and easily frustrating when they did not get the expected results.

5 Overtranslation: Some students found that ChatGPT added information which was not relevant to the original text:

“Sometimes, ChatGPT adds some irrelevant information when translating the paragraph.”(P10)

This led to the reduction of accuracy and students had difficulties in finding and correcting mistakes because they could not identify the original meaning and added information.

 

Discussion

Students’ perceptions of utilizing ChatGPT in translation subjects

Overall, based on the results of the study, it is obvious that most students have explored ChatGPT helps them complete translation tasks faster and improve their learning effectiveness according to the TAM model. This is in line with the study of Almhasees et al. (2024) which showed the opinions of students that ChatGPT could enhance learning effectively and produced accurate translations. Especially, in the case of having fully background information and research context, ChatGPT proves the ability to translate complex context, which is more excellent than traditional machine translation.

Data from the study suggested that the perceived ease of use of ChatGPT impacts on the participants’ intention to utilize the tool. This is in line with the research of Alshurideh et al. (2024), who stated that the necessary of improving and designing AI tools with user-friendly and accessible interfaces to promote their widespread adoption and use in academic settings. Many students, who participated in this research believed that ChatGPT was a useful tool, this view is completely consistent with the above authors, who stated that students used it as a reliable resource in academic settings.

 

Students’ challenges while using ChatGPT in translation subjects

The answers of the participants showed there were agreements that using ChatGPT led to concerns and could be categorized into three major types of risks: cognitive, linguistic, and technical.

The over-reliance on ChatGPT has raised concerns regarding its bad influence on human cognition, particularly in critical thinking and memory. According to Zhang et al. (2024), using AI for a long time could reduce critical thinking skills of students, cause they tended to accept all the answers of AI. The authors also claimed that students who used ChatGPT for exam preparation had lower grades than those who studied traditional methods. This suggests that when students frequently use ChatGPT to find out information instead of recalling it, their memory is weakened day by day.

Beyond cognitive concerns, linguistic risks are another serious issue that needs to be considered in the use of ChatGPT for translation. Many students who participated in the study admitted that they no longer attempted to improve their language skills because they used ChatGPT in translation. This mindset led to a degradation in their linguistic proficiency. If these problems are not addressed clearly, the language skills of students may become weakened.

Although ChatGPT can provide fast translations, it still has many technical limitations, especially in structure and context. Stap and Araabi (2023) highlighted that ChatGPT was an unreliable tool in translation because it could not translate complex or unfamiliar language structures due to a lack of data. Sometimes, the translations of ChatGPT cannot reflect the author’s intention or lack precision in technical terminology.

 

Implications for teaching & learning

The use of ChatGPT in translation courses should not be viewed simply as the introduction of a new technological tool. From a pedagogical perspective, it offers meaningful opportunities to support students in developing analytical and critical thinking skills in translation. Based on students’ perspectives, ChatGPT is not only seen as a tool for producing translations but also as a resource that allows them to compare different versions and reflect on language choices. This indicates that when guided appropriately, ChatGPT can play a supportive role in translation learning rather than replacing students’ own effort.

According to Jurenka et al. (2024), the rapid advancement of artificial intelligence requires translation curricula to be updated on a regular basis so that students can keep pace with current learning trends and professional practices. Incorporating AI-related content into translation courses helps ensure that students are exposed to tools that are increasingly used in the translation industry. From the students’ point of view, learning how to work with tools such as ChatGPT enhances their readiness for the job market, which is more technology-driven compared to the past.

However, students also raised concerns about the potential drawbacks of using ChatGPT too frequently, particularly the risk of becoming overly dependent on AI and engaging less deeply with translation tasks. To address this issue, BlackDuke and Akosubo (2024) suggest a teaching approach in which students first complete translation tasks in class without the assistance of ChatGPT, and then redo the same tasks at home using the tool. The two versions are subsequently compared and discussed in class to highlight the advantages and limitations of both human and AI-assisted translations. This approach aligns well with students’ learning experiences, as it encourages them to actively evaluate ChatGPT’s output instead of accepting it without reflection. Through guided comparison, students become more aware of linguistic accuracy and contextual suitability—areas where AI-generated translations may still lack. More importantly, this teaching strategy helps reduce the negative impact of ChatGPT use while maximizing its value as a learning aid.

In addition, ChatGPT can work well as a scaffolding tool for less confident students, helping them notice linguistic patterns they might otherwise miss. But lecturers must treat it as a starting point rather than a final answer. Students need specific training to catch the robotic or inaccurate outputs that AI often generates. Besides, because AI is so accessible, traditional take-home assignments need a makeover. Moving toward in-class tasks or asking students to critique AI output is a much better way to see their progress. It is also vital to build AI ethics directly into the curriculum. Educators should be aware of students' professional responsibilities. If AI is implemented through structured, critical activities, students can grow without sacrificing academic integrity.

In summary, the findings suggest that the effectiveness of ChatGPT in translation education depends largely on how it is implemented. Rather than allowing unrestricted use, lecturers should design structured activities that promote critical comparison and reflective learning. In this way, ChatGPT can be integrated into translation courses in a manner that supports students’ skill development while preserving learner autonomy and academic integrity.

 

Conclusions and recommendations

With the development of technology, the application of ChatGPT in education is no longer restricted, particularly in translation subjects. Through this study, it can be seen that students have a positive attitude toward the utilization of ChatGPT in translation, especially in improving speed, quality of translation and learning effectiveness. However, it is also important to recognize three types of risks while using ChatGPT.

Several limitations of this study should be acknowledged. The sample size is limited in terms of academic level and population, and the study does not explore the challenges and solutions in depth, highlighting gaps in understanding the difficulties and solutions of students when using ChatGPT in translation subjects. Despite these limitations, the current study provides valuable insights into perceptions of students in using ChatGPT in translation subjects at Van Lang university.

Although this study provides useful information, there are some suggestions for further research that the utilization of ChatGPT in translation subjects should be expanded to clarify the role of it and in- depth perceptions of students in translation learning. The future studies can expand the sample size in educational institutions, academic levels, and translation programs to have a comprehensive view of students’ perceptions in using ChatGPT in translation learning. In conclusion, it is necessary to have more studies about integrating AI translation tools into academic settings.

 

References

Abidah, N. K. K., Hikmawati, R., & Erawanto, V. (2024). ChatGPT in Indonesia-Arabic Translation: A quality analysis. Abidah | Al-Arabi: Jurnal Bahasa Arab Dan Pengajarannya = Al-Arabi: Journal of Teaching Arabic as a Foreign Language. https://doi.org/10.17977/um056v8i1p33-52

Alammar, A., & Abdel-Reheem Amin, E. (2023). EFL Students’ Perception of Using AI Paraphrasing Tools in English Language Research Projects. Arab World English Journal, 14(3), 166–181. https://doi.org/10.24093/awej/vol14no3.11

Almhasees, Z., Rayyash, H. A., Olimat, S., & Mahadin, D. (2024). An Analytical study on student perceptions of using CHATGPT in Language translation. Language Value. https://doi.org/10.6035/languagev.7925

Alshurideh, M., Jdaitawi, A., Sukkari, L., Al-Gasaymeh, A., Alzoubi, H. M., Damra, Y., Yasin, S., Kurdi, B. A., & Alshurideh, H. (2024). Factors affecting ChatGPT use in education employing TAM: A Jordanian universities’ perspective. International Journal of Data and Network Science, 8(3), 1599–1606. https://doi.org/10.5267/j.ijdns.2024.3.007

Atlas, S. (2023). ChatGPT for higher education: Opportunities and challenges [Report]. College of Business, University of Illinois.

BlackDuke, S. E., & Akosubo, D. S. (2024). The Challenges of AI in the Translation Class: What Do the Learners Lose? Port Harcourt Journal Of History & Diplomatic Studies. https://phjhds.com/wp-content/uploads/2024/12/12.-The-Challenges-of-AI- in-the-Translation-Class.pdf

Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative

Research in Psychology, 3(2), 77–101. https://doi.org/10.1191/1478088706qp063oa

 Cochran, W. G. (1977). Sampling techniques (3rd ed.). New York: John Wiley & Sons

Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly, 13(3), 319. https://doi.org/10.2307/249008

Wang, R., & Hou, F. (2024). How to Design Translation Prompts for ChatGPT: An Empirical Study. In Proceedings of the 6th ACM International Conference on Multimedia in Asia Workshops (pp. 1-7). https://doi.org/10.1145/3700410.3702123

Halaweh, M. (2023). ChatGPT in education: Strategies for responsible implementation.

Contemporary Educational Technology, 15(2), ep421. https://doi.org/10.30935/cedtech/13036

Hamarashid, H., Karim, L., & Muhammed, D. (2023). ChatGPT and Large Language Models: Unraveling Multifaceted Applications, Hallucinations, and Knowledge Extraction. Indonesian Journal of Curriculum and Educational Technology Studies, 11(2), 60-70. https://doi.org/10.15294/ijcets.v11i2.

Hamdan, S. (2024). Exploring the Efficiency of ChatGPT vs. Google Translate in Translating Idioms and Idiomatic Expressions: “The Catcher in the Rye” as a Case Study. http://hdl.handle.net/20.500.14131/1795

Hatibu, H. (2024). Students perceptions on the utilization of ChatGPT and Google Translate in English language learning. Journal of Innovation Research and Knowledge, 4(7), 4611-4616. https://bajangjournal.com/index.php/JIRK/article/view/9047

Jain, S., & Angural, V. (2017). Use of Cronbach's alpha in dental research. Medico Research Chronicles, 4(03), 285-291. https://medrech.com/index.php/medrech/article/view/242

Jurenka, I., Kunesch, M., McKee, K. R., Gillick, D., Zhu, S., Wiltberger, S., Phal, S. M., Hermann, K., Kasenberg, D., Bhoopchand, A., Anand, A., Pîslar, M., Chan, S., Wang, L., She, J., Mahmoudieh, P., Rysbek, A., Ko, W., Huber, A., Ibrahim, L.

(2024). Towards Responsible Development of Generative AI for Education: An Evaluation-Driven Approach. arXiv.org. https://arxiv.org/abs/2407.1268

Karpinska, M., & Iyyer, M. (2023). Large language models effectively leverage document- level context for literary translation, but critical errors persist. Cornell University. https://doi.org/10.48550/arxiv.2304.03245

Klimova, B., Pikhart, M., & Al-Obaydi, L. H. (2024). Exploring the potential of ChatGPT for foreign language education at the university level. Frontiers in Psychology, 15. https://doi.org/10.3389/fpsyg.2024.1269319

Paton, B. (2024). Will ChatGPT change the game for language translation? Localize. https://localizejs.com/articles/chatgpt-for-translation

Ren, X. (2025). We want but we can’t: measuring EFL translation majors’ intention to use ChatGPT in their translation practice. Humanities and Social Sciences Communications, 12(1). https://doi.org/10.1057/s41599-025-04604-6

Tumbal, S., Liando, N. V., & Olii, S. T. (2022). Students’ perceptions toward the use of Google Translate in translating. Kompetensi, 1(02), 313–320. https://doi.org/10.53682/kompetensi.v1i02.1853

Venkatesh, V., & Davis, F. D. (2000). A theoretical extension of the technology acceptance model: four longitudinal field studies. Management Science, 46(2), 186–204. https://doi.org/10.1287/mnsc.46.2.186.11926

Zhang, S., Zhao, X., Zhou, T., & Kim, J. H. (2024). Do you have AI dependency? The roles of academic self-efficacy, academic stress, and performance expectations on problematic AI usage behavior. International Journal of Educational Technology in Higher Education, 21(1). https://doi.org/10.1186/s41239-024-00467-0