1. Introduction
AI, a branch of computer science, focuses on developing systems capable of performing tasks typically requiring human intelligence, such as learning, problem-solving, and decision-making (Malik, 2023) has become the driving force behind recent rapid technological advancements (Datta, 2023; French et al., 2023). Furthermore, according to Bristol et al. (2023), it occupies the top position in a hierarchy of fourth industrial revolution (4IR) technologies, acting as a ‘conductor’ orchestrating their collective impact. The release of ChatGPT in November 2022 marked what is arguably the most significant breakthrough in AI to date. As an advanced chatbot, it can, amongst a number of other tasks, generate detailed responses, perform multilingual translations, debug code, and write stories. (Wu et al., 2023). Its rapid adoption, 100 million users within two months, earned it the title of the fastest-growing application in history. With its user-friendly interface and versatility, ChatGPT has significantly expanded AI’s capabilities and applications, representing a pivotal moment in AI’s evolution (Malik, 2023). This rapid development has and is predicted to further impact numerous fields, including education. ChatGPT’s ability to produce human-like responses almost instantly raises questions about its implications, particularly in second-language education such as English as Foreign Language (EFL) classes. Such courses are commonly included in Korean university curricula and aim to improve students’ general English language skills, often focusing on speaking and writing. Students taking these classes will most likely be required to submit homework assignments, to be completed outside of regular lesson time and away from the supervision of the instructor. According to Songsirisak and Jitpran (2021), completing these assignments helps to reinforce that habit of learning outside the classroom and improves overall academic performance and achievement. They continue by discussing how students in the current education climate have two resources for homework assistance, their peers and the internet. These assignments are commonly scored and factored and ultimately contribute towards a student’s grade point average (GPA). Given the power of ChatGPT to quickly produce large chunks of text, translate from one language to another almost instantaneously, and identify errors or inconsistencies with language usage, there is a growing concern students are using this technology to complete homework assignments consequently affecting grades and influencing academic performance. This poses a number of pedagogical issues for teachers relating to what type of homework should be assigned, how homework is assessed / graded, and class policies regarding the use of this technology. Hence, there is scope for a better understanding of how prevalent the use of ChatGPT is for English homework assignments, how teachers can deal with its ever-increasing influence, and students’ perceptions of it. A case study was designed to investigate these areas further with the hope of helping teachers be better prepared when planning, assigning, and grading homework assignments.
2. Literature Review
2.1. Prevalence of ChatGPT Usage
Despite it only being approximately 2 years since ChatGPT was released there is a body of research which examines its influence on the field of language learning and specifically how prevalent it is being used to prepare and complete EFL homework assignments. A review of this literature reveals two contrasting trends. The first of which is students being aware of the existence and potential for using ChatGPT for homework assignments yet choosing not to do so. One study (Krecar, Kolega & Jurcec, 2024) concluded that 350 university students in Croatia generally did not use ChatGPT for their homework assignments and indicated neutral feelings towards both its effectiveness and ease of use. In a different study (Singh et al., 2023) used a 12-point questionnaire on 430 students in the United Kingdom. The results indicated that students do not choose to use ChatGPT for homework assignments with potential for misuse being listed as the primary reason.
However, several different studies have found that an ever-increasing number of learners are opting to use ChatGPT for homework assignments and are acutely aware of the benefits and drawbacks of doing so. Črček and Patekar (2023) sent an online questionnaire to 201 students attending both public and private universities in Croatia These authors found over half of the participants used ChatGPT to either partially or fully complete written homework assignments despite significant ethical reservations. In this study, the participants indicated the most common use of ChatGPT was to generate ideas for writing although it was also used to summarize, paraphrase and to generate chunks of texts to be included in their essays. Two studies done in Saudi Arabia mirror these findings to some extent. The first of which (Almanea, 2024) found 135 university EFL students to have generally positive views on using ChatGPT; however, they would conceal its use and would use it to complete homework assignments when possible. The second study (Almulla & Ali, 2024) found 60% of 473 undergraduate students reported using ChatGPT to assist with their EFL homework assignments with slightly less than half of them doing so on a weekly basis. Once again, generating ideas and research was one of the primary uses of ChatGPT, but it was also used as a way of completing assignments. In Morocco, a survey of 115 high school students done by Tamimi, Addichane, and Madani (2024) concluded that students are heavily reliant on Chat GPT to complete English assignments with the perceived primary benefits being getting immediate feedback, better understanding of materials, and ease of use. Interestingly, just under 20% of these students selected ‘does everything for you’ as a perceived benefit indicating a reliance on ChatGPT and potential for overuse resulting in reduced ability to generate ideas and being able to write independently. Two studies done in Vietnam (Ngo, 2023; Le & Tran, 2024) using a questionnaire and interviews found students were positive about using ChatGPT for homework assignments due to its efficiency and effectiveness. There were also indications that it is going to be used more in the future when approaching homework assignments. Both groups of participants also demonstrated an awareness of the pitfalls of using this technology including reliability of information, overreliance, and ethical concerns.
The studies discussed above show university students in varying countries and contexts are either choosing to ignore ChatGPT as a tool for helping with homework assignments or adopting it whilst being aware of both the advantages and disadvantages.
2.2. Role of GPT and Implications
It is logical to assume the influence of ChatGPT will continue to grow; therefore, the second part of this study attempted to uncover how Korean university EFL learners view its role moving forward. Specifically, what is their position on it and how teachers should control the use of this technology for completing homework assignments and the consequent pedagogical implications.
A review of previous research shows a number of different perspectives regarding this issue. A study by Famaye et. al (2024) surveyed over 600 high school students from varying countries regarding the integration or banning of ChatGPT. This study concluded that ChatGPT provides students with an easier and quicker way to cheat or plagiarize. However, there is also an acknowledgement that this technology is here to stay and as such there is a need for clearer guidelines for both teachers and students on how to use it in a more ethical way. A different study by the same author (2023) on banning or adapting ChatGPT in US schools revealed students perceive it to be a valuable learning tool. However, they also have significant reservations in relation to fairness and the potential for cheating which influences their choice to use it. Baek et al, (2024) conducted a study of over 1000 university students enrolled at a variety of different higher education institutions in the US. The results showed that non-native English speakers are more likely to use ChatGPT but also have a greater fear of punishment as a result. There were also indications of a desire for clear guidelines on its use for both in-class work and for homework assignments. Almost half of the respondents said they were unsure of such policies and 20% saying their institution has no such guidelines. This highlights “conflict between traditional academic values and the disruptive nature of AI technology like ChatGPT” (p. 5). Interestingly, a significant number of participants in this study stated they had no concerns about using ChatGPT for assignments and simply see it as a complement to human effort rather than replacement. A study (Xu et al. 2024) of 187 Chinese university students enrolled in various disciplines revealed similar concerns regarding greater need for ethical guidance on using ChatGPT outside of the classroom. This work also affirmed the notion that significant change in the approach by administration and teachers is needed in order to adapt to the changing education climate.
In terms of literature related to the role of ChatGPT in English language-based courses, a selection of studies was reviewed. 235 Omani students enrolled in EFL English courses responded to a questionnaire in a study conducted by Dehdary and Meschi (2024). When asked whether ChatGPT increases the possibility of cheating on homework assignments, the results were divided with just over half believing it does, 35.8% being unsure, and 13.2% believing it does not. Despite this, almost 94% of respondents indicated they disagreed with the idea of banning the use of ChatGPT for any type of work, including homework, and punishing those who did so. Once again, the need for clearer guidelines on how students can correctly use it was highlighted along with a change in approach on behalf of teachers to include rather than exclude this technology moving forward. In South Korea, Van Horn (2023) also highlighted the need for increased levels of support and flexibility in integrating ChatGPT into all elements of conversation-based courses at a university. Richards and Jones (2024) studied Korean university student opinions of using ChatGPT in mandatory college EFL writing class highlighting a positive perception in terms of helping with grammar, vocabulary and brainstorming. One interesting finding was there was very little correlation between students’ English level and how often ChatGPT was used. However, this research was solely focused on in-person class time rather than how ChatGPT might be used to complete homework assignments
This current body of research provides a valuable insight to what extent ChatGPT is being used for EFL homework assignments and how students view its role. There is some scope to expand on this and it is hoped this study can build on previous work. Consequently, this study will attempt to answer the following research questions:
3. Method
3.1. Research Context and Participants
In order to address the above research questions, a case study was designed and implemented in December 2024. 261 people who had all taken at least one university course taught in English at a private university located in the greater Seoul area responded to a survey questionnaire. 54.1% were male and 45.9% were female. 82.7% of respondents stated they were in their freshman year, 9.8% were sophomores, 2.7% and 4.7% were juniors and seniors respectively. The data came from students of varying English ability with 16.5% of respondents saying they took basic level classes, 69% took intermediate classes, and 14.5% took advanced classes. These levels were pre-determined by the university administration prior to the student’s enrollment. This was done using a pre-existing level test created and provided by a 3rd party in which the research team had no control or input. The majority of respondents were either attached to the colleges of economics (26.7%), software engineering (24.7%) or engineering (18.8%). The remaining respondents were studying a variety of other areas, including but not limited to humanities, law, social sciences, art, and music. Regarding the number of university courses taught in English, 25.1% had taken one, 54.5% had taken two, 12.5% had taken 3 with the remaining 7.9% having taken 4 or more.
3.2. Data Collection
Data collection occurred between December 15th and December 30th, 2024, initially through the use of a survey questionnaire (Appendix A) which was designed and written by the researchers. As suggested by Evans and Mather (2005), this method offers a number of benefits. Firstly, the flexibility of a questionnaire means collection is not limited by time and location constraints. In addition, it aided the collection of a large dataset while also allowing the inclusion of a number of differing question types to gather a wider variety of responses. Moreover, since survey questionnaires are highly effective in assessing perceptions on particular topics or situations (Young, 2016), the researchers concluded that this method was the best approach for this project.
A QR code and hyperlink to the survey was created and posted to the online learning management system (LMS) for classes taught by both researchers and six other colleagues. Students of these classes were requested to log into the LMS and complete the survey in their own time. They were under no obligation to do so and were informed that any responses would in no way impact any assignment scores, course grades, GPA, or overall academic integrity.
Following on from the initial demographic questions, the remaining survey items were a combination of different question types including long and short answer items, multiple choice items and 5-point Likert scale questions. They focused on students’ general experience of using ChatGPT for assignments given in courses taught in English, the perceived benefits and drawbacks, perceptions on the role of using ChatGPT for homework assignments, and subsequent pedagogical implications. A pilot test was conducted prior to survey distribution to assess its validity, in addition, a Cronbach Alpha test applied to the 9 Likert scale items revealed a score of 0.81 indicating good internal consistency. To ensure clarity survey questions were provided in two languages. Questions were originally written in English, and a reverse translation approach was used to convert them into Korean. Consistency between both languages was ensured by asking a colleague of the research team who is fluent in both to double check accuracy.
The penultimate survey question asked if respondents would be willing to be contacted by the researchers to discuss their response in more detail. Those who responded positively were asked to leave their email address and informed that this data would be permanently deleted at the conclusion of this project. A number of these respondents were contacted in early January for further discussions. They either took place on an email thread or if convenient, via a Zoom call which was recorded and transcribed. The final survey item requested permission to use responses as part of this research. 6 students answered negatively to this request; therefore, their answers were discarded leaving data from a total of 255 students to take to the analysis stage.
3.3. Data Analysis
This was done in two phases. Initially, using a descriptive statistical approach, the researchers organized and examined the Likert scale and multiple-choice items to highlight important themes and patterns. Following on from this, the short and long answer questions, which generally provide a deeper insight into the participants’ perceptions and opinions, each researcher utilized a general analysis of qualitative research. Recurring themes and patterns were identified. The researchers then convened several times to compare findings and resolve any discrepancies. At this point the second phase began, any comments/answers to these questions that the research team felt worthy of following up on were identified and, if permission had been granted, exploratory emails were sent to each participant requesting further discussion regarding their answers. In total, 30 were contacted with 4 of those not responding to requests, leaving follow-up data from a total of 26 students. These were examined in detail by the researchers probing for thoughts that could be used to expand on the themes identified in phase one. All Korean responses were back translated into English using a reverse translation method. A bilingual colleague verified consistency and accuracy between both languages.
4. Results and Discussion
4.1. Prevalence of ChatGPT Usage
Research question 1 attempts to develop an understanding on how prevalent ChatGPT usage is for homework assignments in courses taught in English. Survey question 7 asked the participants if they had ever done so. 171 (67.1%) answered ‘yes’ and their responses were compared with other demographic data points such as age, major, and how many English courses they had taken, but there were no clear identifiable trends. However, there was one interesting data point found when examining students’ English level and response to question 7. As shown in Table 1, students from intermediate level courses appear to be using ChatGPT for home assignments more often than students from the other two levels. Although the reasons behind this were not clear, it might be possible to suggest lower-level students find it more difficult to apply ChatGPT to their homework assignments whereas higher level learners are skilled enough to not need to use it.
<Table 1>
Percentage of students who use ChatGPT for homework based on English level
| English Level | n | No | Yes |
|---|---|---|---|
| Beginner | 42 | 19 (45.2%) | 23 (54.7%) |
|
|
|||
| Intermediate | 176 | 50 (27.8%) | 127 (72.1%) |
|
|
|||
| Advanced | 37 | 16 (43.2%) | 21 (56.7%) |
Those who answered ‘yes’ to question 7 were directed to a section of the survey designed to further understand how they used ChatGPT and any perceived benefits / drawbacks. The first part of this section (survey question 8) asked the extent to which the participants found ChatGPT helpful for homework assignments. Slightly over 40% of the 171 students indicated they found it either very helpful or extremely helpful. A similar number (45%) said they found it somewhat helpful with only 11.7% saying it was slightly helpful. Only 1 student selected the “not very helpful” option. Based on the findings indicated in Table 1, data from question 8 was also compared to students’ English level. Once again, there are clear indications (Table 2) that students who took intermediate classes are not only using ChatGPT more often but also find it to be the most helpful.
<Table 2>
Percentage of students who find ChatGPT most helpful based on English level
Survey question 9 asked the students to explain why or why not. Analysis of responses reveal there were a number of recurring themes and patterns including ChatGPT helping with grammar and sentence structure, assistance with brainstorming and developing ideas, ChatGPT being better than other available technology, and ChatGPT reducing time and increasing efficiency when completing assignments. A summary of these, along with a selection of student responses can be seen in Table 3.
<Table 3>
Selection of comments from 171 participants on why ChatGPT was helpful
In an email follow-up, one participant offered a brief summary of all these benefits while also commenting on how they feel ChatGPT cannot replace the human input needed to complete English homework assignments. However, they also felt it has helped them to improve their overall ability to complete English homework assignments:
Since ChatGPT is generative AI, it doesn’t help me create original sentences, but it’s useful for checking words and flow based on large language models (LLM). It doesn’t replace human creativity but helps with structure and flow. It quickly turns ideas into actual results with just a command. It helps me realize errors in my thinking, and sometimes what I thought was a good idea isn’t actually good. My overall improvement has been faster since using GPT. (Participant A, male, age 20, intermediate level).
Despite all the above listed benefits, there was one area in which the students were unhappy about ChatGPT performance when completing assignments. They felt it is not always accurate and provides incorrect information / responses to prompts. There were a number of comments relating to how students used ChatGPT to help with an assignment, but they felt the information was either incorrect or not suitable, leading to students using their own knowledge of English to complete it. A summary of such comments can be seen in Table 4. This not only demonstrates awareness of the limitations of this technology but the fact the students knew ChatGPT was wrong and were able to apply their own knowledge to rectify the mistakes.
<Table 4>
Selection of comments from 171 participants on why ChatGPT was unhelpful
The survey continued by asking the participants what types of assignments they use ChatGPT for and why. In survey question 10, respondents were presented with a list of commonly assigned types of English homework. They were asked to choose what type of assignment they used ChatGPT for and multiple selections were possible. There was also the option of selecting ‘other’ and entering any other types of assignment not covered in the initial list. The data from this question can be seen in Table 5.
<Table 5>
Types of assignment that students most commonly used ChatGPT for
Given that previous questions have revealed some interesting finds when comparing question data to English level, the same principle was applied to this data. It showed that students who took intermediate classes were more likely to use ChatGPT for a wider variety of assignments. Students from the intermediate level generally selected a greater number of assignment types compared to students from other levels suggesting they are more likely to leverage ChatGPT for different types of tasks. Advanced students tend to use ChatGPT for more complex combinations of assignments, such as essay writing combined with grammar or speaking assignments. Basic level students primarily use ChatGPT for basic grammar or vocabulary assignments.
Survey question 11 asked the participants the reasons behind why they opted to use ChatGPT for those particular types of assignments. General observations from the data indicate ChatGPT is particularly helpful for tasks that require extensive writing and organization of ideas, such as essays, paragraphs and preparing presentations. There were indications that it was somewhat helpful for vocabulary and grammar assignments, yet its use for speaking assignments was variable. More specifically, there were clear trends that emerged when examining each type of specific assignment as shown in Table 6.
<Table 6>
Selection of reasons from 171 participants for using ChatGPT based on assignment type
Aside from the themes and comments listed above, there was one other notable theme found in the responses to question 11 which relates back to an idea discussed earlier. ChatGPT now appears to be the go-to form of technology used for home assignments with students seemingly abandoning previously used resources such as Google Translate and Naver’s Papago. One participant stated: “I felt that using Papago for translation had its limitations, so I wanted to use more advanced English offered by ChatGPT”. Another commented: “It is difficult to get the correct grammar using Google or Naver, but ChatGPT is better and more natural”. A different participant elaborated: “Because it is the most widely used program among students because it generates excellent English results directly without needing a separate translation process”. This indicates a shift in approach to how students are using technology to assist with homework assignments.
This section concluded with question 12, which asked the students if they felt using ChatGPT helped them get a higher score on the homework assignments. Almost 60% of respondents selected agree or strongly agree. Interestingly, only 12 (7.1%) out of the 171 answered negatively and felt it did not help them achieve higher scores. The data for this question can be seen in Figure 1.
A more detailed analysis of the question once again reveals students who took intermediate classes were more likely to agree as shown in Table 7. This could be due to ChatGPT’s ability to provide better vocabulary, correct grammar, and enhance sentence structure, which are crucial for students who have a general level of understanding but need assistance with more complex tasks. Basic level students might not feel as strongly that ChatGPT helps them achieve higher scores because they are still developing foundational language skills and might struggle to understand the suggestions provided by ChatGPT. Advanced learners might agree to a lesser extent, as they are more confident in their language abilities and may feel ChatGPT is less helpful than students in other levels 84 participants (32.9%) stated they had never used ChatGPT to help with English homework assignments. Thus, were not required to respond to any of the survey items discussed above. Instead, they were directed to question 13 asking them to explain their reasons for not doing so. Analysis of these responses has uncovered two clear reasons which could be considered as drawbacks to this technology.
<Table 7>
Percentage of students by level who feel ChatGPT helped with homework scores
The most frequently mentioned reason was a strong desire to be able complete assignments independently of ChatGPT and therefore maintaining ownership of the process of learning English. Many of these students appear to value being able to complete and submit assignments without having to rely on AI based technology. Furthermore, they feel that doing so allows them to develop as students and learn the language at a pace which is comfortable for them. One survey response stated: “Submitting ChatGPT’s answers that I didn’t create is no benefit to me.” Another elaborated: “I have to think for myself. If I use GPT for assignments, my English skills won’t improve”. Another commented: “It ruins my English learning a bit”. Several students were contacted to discuss this idea in more detail. In an email thread, one stated:
I believe using ChatGPT means missing out on opportunities to grow and study on my own. While it might get you points by doing your assignments, as a student who wants to learn, I think it’s more important to study and work on assignments myself at my own speed. I don’t plan on using it in the future, and I think we should stop students who use ChatGPT instead of doing their own work. (Participant B, female, age 20, intermediate level).
This idea was also somewhat reflected in conversation with a different participant:
My English class was within my ability, and I thought doing it myself would benefit my academic achievement more. It’s easier for me to do my English assignments in my own words and easier to memorize that way. I could do it on my own, so there was no particular reason to use ChatGPT. I just didn’t feel the need to use it. The tasks weren’t difficult, and I heard most classes prohibit GPT”. (Participant C, male, age 21, advanced level).
This comment from a Zoom call mirrored these sentiments while also implying that using ChatGPT is a waste tuition fee:
Using ChatGPT in a class meant to help me learn language skills is not beneficial at all. We pay tuition, and if ChatGPT does all the assignments, the efficiency of learning goes down, and it defeats the purpose. I don’t use it because I don’t want to waste my tuition money. (Participant D, male, age 20, intermediate level).
The second reason for not using ChatGPT for homework assignments relates to authenticity and accuracy of content produced. Students appear to place high value on submitting work that is original and created by themselves. Furthermore, there are high levels of distrust with the answers produced by Chat GPT. A selection of student comments elaborating on this can be seen in Table 8.
<Table 8>
Selection of comments from 84 participants relating dissatisfaction with GPT’s content
While not as apparent in the data as the reasons discussed above, there was also one other theme evident in the responses to question 13. It relates to concerns over plagiarism and a fear of punishment by their professors for using ChatGPT for homework assignments. Several students commented they were worried about getting into trouble for submitting content they could not verify and could be viewed as plagiarized. One student stated: “I heard that using ChatGPT in e-learning systems can cause originality issues. I don’t use ChatGPT to avoid that problem”. Another said: “I was worried about plagiarism issues so decided to do all my work by myself and it felt good”. When discussing possible reprisals for using ChatGPT, one comment read: “I heard most classes prohibit GPT”. Another said: “My friends told me not to use ChatGPT because they heard about one student who got punished for using it”. Both of these comments imply that they had not been directly told not to use ChatGPT, rather heard it second-hand which affected their decision.
4.2. The Future Role of ChatGPT
Research question 2 asked students how they feel about the role of ChatGPT for completing English homework assignments and its pedagogical implications. Survey question 14 asked the participants if they should be allowed to use ChatGPT for their homeworks, assignments and quizzes. A combined total of 109 (42.7%) of the participants answered favorably by strongly agreeing and agreeing to this question, whereas only 57 (22. 4%) participants answered strongly disagree and disagree. 89 (34.9%) of the participants answered neutrally. Inspecting the data more closely and contrary to previous questions, it indicates a very weak positive correlation between the English level of students and their opinion on allowing ChatGPT usage in English-taught classes. The data for this question can be seen in Figure 2.
Following this, question 15 asked for more in-depth responses. Participants who were in favor of allowing ChatGPT to aid them with out of class assignments noted comments such as: “By using ChatGPT’s assistance, one could achieve better grades”.; “Using ChatGPT can improve one’s insufficient English skills, so if used correctly, it can be helpful”. and “ChatGPT provides significant assistance and makes things easier to understand, so it is very helpful”. To receive a more detailed response, several participants were contacted via email. One participant stated:
In this rapidly changing digital age, it seems inevitable to use these tools, especially when needed. Because it helps quickly with tasks and if you don’t use it, the gap in skill levels becomes too large. (Participant E, female, age 22, intermediate level).
Another participant responded:
It simplifies understanding. Since language is based on speaking and writing, it helps with grammar and things that may not show up on Google. It aids in accumulating knowledge and therefore supports learning. (Participant F, female, age 20, beginner level).
Conversely, participants who disagree with the use of ChatGPT for homeworks and out of class assignments made comments such as; “It may help with their vocabulary and grammar, but if they just copy and paste, it could be detrimental”.; “It can make one’s brain dull, so I’m not sure about it”.; “It could cause losing the ability to think for oneself, making the mind stagnant”. and “I think using ChatGPT has no positive effect on learning”. Participants who answered neutrally, understood the help ChatGPT gives but they shouldn’t be too reliant on its use noted the following comments; “I think there’s a middle ground with pros and cons to the choice”.; “It can be ambiguous because depending on the student, they could use GPT well or abuse it”. and “I think it’s good to get some help, but I don’t think I should be too dependent on it”.
Survey question 16 asked the participants if their professors in any of the classes taught in English had told them not to use ChatGPT for homework assignments. 199 (78%) answered “no” and 56 (22%) answered “yes” with responses to the follow up question (Question 17) being somewhat mixed
A number of students who answered negatively seemed to favor this approach with several appearing to appreciate the freedom to explore the use of new technology for English homework assignments. Comments included: “I thought allowing students to use new tools to do homework is a good thing”; “I think ChatGPT is a useful learning tool, so it should be allowed”.; “Since it’s the trend of the times, banning it seems strange”.; “The ability to use ChatGPT well is a skill itself. If used properly, it’s an excellent tool”.) Others appeared to be grateful to the professor for not restricting its use “I’m glad we can use ChatGPT”.; “I was relieved that I could use it”.; “I liked that the professor let us make our own judgments”. However, there were also those who felt indifferent noting that allowing its use did not alter their approach to homework assignments. “I had no strong opinions about AI policies”.; “I didn’t think much about it”.; “I had no particular thoughts”; “Not really bothered by it. I can do it on my own”.
Data from those who responded with ‘yes’ to question 16 also revealed very mixed feelings. There were those who agreed with the restriction and understood the rationale behind it, emphasizing the importance of using English independently of ChatGPT for homework assignments to help them learn by themselves. “Since the purpose is to improve one’s English skills, I agree with the restriction on ChatGPT use”.; “To improve one’s English learning, limiting ChatGPT use is necessary”.; “It fosters critical thinking skills, so I viewed it positively”.; “I think not using it would be more helpful for learning, as it provides answers too easily”. Others expressed feelings of indifference indicating that it did not significantly impact how they chose to complete homework assignments. Interestingly, there were also students who appeared to criticize the restriction of using ChatGPT highlighting how useful it can be for those who struggle with English or are not confident in their ability. “For those who are not good at English, it can be difficult and burdensome without GPT”.; “I thought it might be necessary when using a foreign language”.
In addition, regardless of whether students were told they were allowed to use ChatGPT or not, one common response to question 17 implied that they were willing to follow the guidelines set by their professors. If ChatGPT usage was permitted they would make use of it, likewise, would not use it if told not to. (“If the professor says not to use it, I won’t”; “I agree with the professor”.; “The professor didn’t mention it, so I thought it was okay to use”; “I follow the professor’s rules”.
Overall, the analysis reveals that the majority of students who were not restricted from using ChatGPT felt positively or were indifferent about it. This suggests a general acceptance and appreciation for the freedom to use AI tools in their academic work. However, the lack of clarity or detail to some responses highlights the need for more detailed feedback to understand students’ perspectives better.
Since this technology is relatively new and both students and teachers need time to adapt to its influence, one area that could be important to consider is in what ways teachers allow students to use it for homework assignments. In previous years, due to perceptions that it could encourage plagiarism and cheating, the default approach would have been telling students not to use it. Survey question 18 asked the participants if they felt teachers of courses taught in English need to be more flexible in their position on using ChatGPT for homework assignments. The data showed a strong preference for more leniency with almost 60% saying they agreed or strongly agreed. One very interesting point can be seen in Figure 3 with only 9.8% selecting either disagree or strongly disagree.
[Figure 3]
Should professor be more flexible in their approach to using ChatGPT be used for homework / assignments / quizzes
The reasons behind these responses expanded on in question 19, which asked the participants to explain why they felt this way. Several key themes emerged from this data. Firstly, the idea that we are now living in a new era in which there is no option but to adopt new technologies such as ChatGPT was one of the most frequent comments. This implies the students feel teachers need to adapt their approach to homework assignments accordingly. Secondly, the benefits in terms of giving students access to more information and content which they can apply to their assignment and consequently learn more was also mentioned. Finally, there were strong indications that students would like to see ChatGPT appropriately integrated into lessons and homework so it can be used as a positive aid to learning rather than being viewed as a threat. A summary of these themes and a selection of responses to question 19 can be seen in Table 9.
<Table 9>
Selection of comments from 173 participants on why teachers should be more flexible in their approach to ChatGPT
Finally, the last question (20) asked the participants if they feel the university administration should provide clearer guidelines for students on if and how they can use ChatGPT. Almost two thirds (65.1%) of the participants were in favor of the university administration department doing so, with 101 (39.6%) and 65 (25.5%) participants selecting agree and strongly agree respectively. Only 6 (2.4%) participants selected strongly disagree, 16 (6.3%) chose to disagree, while 67 participants (26.3%) remained neutral. This illustrates the participants’ belief that the university administration should establish more explicit and comprehensive guidelines regarding the permissible use of ChatGPT for homework assignments. These guidelines should clarify whether students are allowed to utilize the tool and, if so, outline the specific contexts, limitations, and ethical considerations associated with its use in academic work.
5. Conclusion and Implications
This study aimed to explore the extent to which Korean university EFL students use ChatGPT for completing English homework assignments, along with their perceived benefits, drawbacks, and attitudes toward its pedagogical implications. To address these questions, a case study was conducted involving 255 students from a private university in the greater Seoul area, all of whom had taken at least one English-taught course. Data were collected through a bilingual survey questionnaire comprising various question types and select follow-up interviews were conducted for deeper insight. This mixed-method approach enabled a comprehensive understanding of students’ experiences and perceptions regarding ChatGPT in an academic EFL context.
A number of key implications can be drawn from this study. Firstly, the prevalence of ChatGPT for homework assignments in this particular context is reasonably high despite a level of awareness of potential pitfalls associated with its use, mirroring several previously discussed studies (Almanea, 2024; Almulla & Ali, 2024; Tamimi, Addichane, and Madani, 2024). While students of all levels appear to be using this technology, those of intermediate level appear to be relying on it more, which contradicts the results of Jones & Richards (2024). Furthermore, this study showed that ChatGPT was particularly beneficial for writing essays and composing paragraphs which is consistent with the findings of Črček and Patekar (2023). The primary reasons for its usefulness included improvements in grammar and sentence structure, assistance with brainstorming, and a reduction in the time required to complete assignments. Additionally, students noted that the use of ChatGPT contributed to higher assignment scores, which, in turn, could positively impact their overall course grades. Moreover, ChatGPT emerged as the preferred AI tool among respondents, surpassing alternatives such as Google Translate and Naver Papago. However, almost 30% of students are not using this technology for homework assignments due to concerns regarding potential risk of being accused of plagiarism and facing academic penalties. Some participants also highlighted that the information generated by ChatGPT was occasionally inaccurate or overly complex, raising concerns about its reliability. These findings align with several studies discussed in the literature review (Krecar, Kolega & Jurcec, 2024; Singh et al., 2023). This widespread use of ChatGPT among students indicates a growing reliance on ChatGPT for homework assignments. There are clear indications students feel it not only improves the quality of their work but also their efficiency. Consequently, educators should consider exploring how to allow students to use it as a resource for assignments without fear of repercussions.
Secondly, in line with the findings of Van Horn (2023) and Dehdary and Meschi (2024), there needs to be greater flexibility in approaches to using ChatGPT for homework assignments. Specifically, this data reaffirms the finding of Famaye et al. (2024) in that the education landscape has changed indefinitely and allowing the use of ChatGPT for assignments is needed to help students adapt to it. In addition, as discussed by Famaye (2003) and Beak et al. (2024) ChatGPT should be embraced as a learning tool without being overly reliant on it. It is important to balance its use with traditional teaching methods. This ensures that students develop critical thinking skills, problem-solving skills, and without them becoming overly dependent on AI.
Finally, as discussed by Baek et al. (2024) and Xu et al. (2024), this study confirms the need for university administrations to establish clear guidelines on the appropriate and ethical use of ChatGPT for out of class assignments. As AI technologies such as ChatGPT continue to evolve, institutions must proactively develop strategies to manage their impact on education. Collaboration among university administrators, faculty, and students will be essential in ensuring that AI tools are used responsibly and effectively. To this end, universities should consider implementing educational initiatives to raise awareness of both the benefits and limitations of AI, equipping students and faculty with the necessary knowledge to navigate this rapidly advancing technological landscape.
This study highlights the growing reliance on ChatGPT for homework assignments, particularly among intermediate-level students, due to its perceived benefits in improving grammar, aiding brainstorming, and enhancing overall efficiency and grades. Despite these advantages, concerns about plagiarism, academic penalties, and occasional inaccuracies persist, deterring some students from using the tool. The findings emphasize the need for flexible, balanced integration of AI in education, where students are encouraged to use ChatGPT as a supportive resource without becoming overly dependent. Moreover, the study calls for clear institutional guidelines and educational initiatives to ensure the ethical and effective use of AI in academic settings.
6. Limitations
Despite the wealth of valuable data points regarding the use of ChatGPT for out of class assignments among students in EFL classes at a Korean university, this study is subject to several limitations. First, while every effort was made to obtain a comprehensive dataset, the final sample consisted of 261 participants, a number deemed satisfactory by the research team. However, a larger sample size would have been preferable. Additionally, 82.7% of respondents were first-year students, limiting the study’s ability to capture perspectives from upper-year students with greater experience in university settings. Including data from more senior students may have yielded different insights.
Moreover, this case study was conducted at a single university in South Korea, which limits the generalizability of the findings. Students from other Korean universities or international institutions may hold distinctly differing views on this topic. A broader study incorporating a more diverse range of educational contexts and a larger sample size would likely provide a more comprehensive understanding.
Furthermore, the majority of data collection relied on multiple-choice and short-answer questions, which inherently limit the depth of responses. A more robust qualitative approach, incorporating methodologies such as interviews, focus groups, and direct observations, could uncover additional findings not evident through the methods employed in this study.
Finally, since the research team had no control over the levels assigned to students, analysis could only be done on courses which pre-determined levels. To ensure greater accuracy, using a purposely designed level test prior to distributing the survey may have been more effective.
PDF Links
PubReader
ePub Link
Full text via DOI
Download Citation
