Speaker
Description
LECTURERS’ AND STUDENTS’ PERCEPTIONS OF USING GEMINI IN THE WRITING 2 COURSE AT QUY NHON UNIVERSITY
Bui Thi Huong Giang1, Huynh Phan Gia Khanh2, Nguyen Thi Linh Chi3, Tran Le Thu Ha4, Ngo Le Minh Thao5
1 Faculty of Foreign Languages, Quy Nhon University, Gia Lai Province, Vietnam
2-5 English Language Teaching, Cohort 47, Faculty of Education, Quy Nhon University, Gia Lai Province, Vietnam
*Corresponding author: Bui Thi Huong Giang
Email: buithihuonggiang@qnu.edu.vn
Abstract
The integration of generative artificial intelligence (AI) into higher education raises questions about how tools such as Gemini affect foreign language writing and learners’ autonomy. Recent researches on AI-assisted writing has mainly examined learners’ perceived advantages and disadvantages; however, little attention has been paid to the aligned and diverging views of students and lecturers on the pedagogical, ethical, and assessment implications of using Gemini in an English as a Foreign Language (EFL) writing course. This study investigates the perceptions of lecturers and second-year English majors regarding the use of Gemini in Writing 2 at Quy Nhon University (QNU). Specifically, it examines students' Perceived Usefulness (PU), Perceived Ease of Use (PEOU), attitude, Subjective Norms (SN), Perceived Behavioural Control (PBC) and Behavioural Intention (BI), and explores lecturers' perspectives on the benefits, risks, and regulatory needs associated with Gemini-supported writing. A mixed-methods using explanatory sequential was adopted, combining survey data from 200 students and 3 lecturers with semi-structured interviews with 5 participants. Quantitative data were analyzed through Partial Least Squares Structural Equation Modeling (PLS-SEM), while qualitative data were examined using thematic analysis. The findings suggest that surveyed students generally hold favorable views of Gemini in Writing 2 and express a strong intention to continue using it. Lecturers likewise acknowledge Gemini as a valuable support tool, yet they raise concerns about overreliance and diminished independent thinking. The study argues that Gemini should be used selectively as a writing aid, with explicit guidance, training, and clear course policies to ensure responsible AI use and reflective, ethically informed writing practices.
Keywords: generative AI, Gemini, EFL writing, higher education, student perceptions, lecturer perspectives.
- Introduction
In recent years, AI has become increasingly common in education, particularly in foreign language learning. AI-powered tools can provide personalized learning support and immediate feedback, allowing learners to practice beyond the classroom (Pham, 2024). This is especially relevant to English writing, which is often considered a challenging skill. Learners need to generate and organize ideas, use appropriate vocabulary and grammar, and produce coherent texts. They may also experience difficulties related to motivation, confidence, and writing anxiety (Baharudin et al., 2023). Therefore, AI tools have the potential to support learners throughout different stages of the writing process.
Among current generative AI tools, Google Gemini has attracted attention for its ability to generate and revise texts, suggest ideas, improve language use, and provide support for different writing tasks. Previous research has indicated that Gemini can support learners in areas such as content organization, language refinement, and idea development (Kartika, 2024). However, the use of AI in writing also raises concerns. Excessive reliance on AI may limit learners’ independent thinking and creativity, while other issues such as academic integrity, authorial voice, and data privacy also need to be considered (Pham, 2024). Thus, understanding how AI is perceived by both learners and teachers is important for its effective and responsible integration into education.
Although research on AI-assisted language learning has developed rapidly, there is still a need for studies examining the use of specific AI tools in particular educational contexts. In particular, limited research has explored both students’ and lecturers’ perceptions of using Gemini in university-level English writing courses. Students and lecturers may have different perspectives on the usefulness, challenges, and appropriate use of AI in writing instruction. Therefore, examining both perspectives can provide a more comprehensive understanding of the role of Gemini in English writing education.
Therefore, this study, entitled “Lecturers’ and Students’ Perceptions of Using Gemini in the Writing 2 course at Quy Nhon University,” investigates the perceptions of second-year English-major students and lecturers from the Faculty of Education and the Faculty of Foreign Languages at QNU regarding the use of Gemini in the Writing 2 course. The study examines students’ PU, PEOU, SN, PBC, ATT, and BI, while also exploring lecturers’ perceptions of the potential benefits and challenges of integrating Gemini into writing instruction. Based on the findings, the study proposes recommendations for using Gemini as a supporting tool while supporting learners’ independence in developing their writing skills.
The study addresses the following research questions: - How do EFL students perceive the usefulness, ease of use, social influences, perceived control, attitudes, and behavioural intention associated with Gemini use in a Writing 2 course?
- How do students and writing lecturers perceive the opportunities, risks, and agency-related challenges of Gemini use in EFL writing?
- What recommendations can be suggested to optimize the use of Gemini in teaching and learning the Writing 2 course?
- Literature review
2.1 Foundational concepts
2.1.1 Definition of Writing
Writing is a complex cognitive activity that requires writers to control various aspects, including content, organization, sentence structure, vocabulary, punctuation, and spelling (Bell & Burnaby, 1984). According to Knoch et al. (2016), writing is the ability to express thoughts through meaningful written messages and communicate ideas effectively. From these definitions, it can be inferred that writing involves using language to convey ideas clearly and coherently by managing content, organization, and linguistic features.
2.1.2 AI
AI is described as a subfield of computer science concerned with developing systems capable of exhibiting human-like intelligence, including language understanding, learning, reasoning, and problem-solving abilities (Barr & Feigenbaum, 1981). Kaplan and Haenlein (2019) further define AI as a system that can interpret external data, learn from it, and flexibly adapt to achieve specific goals and tasks.
In education, AI uses technologies such as machine learning, natural language processing, and predictive analytics to improve learning through personalized instruction, adaptive materials, automated feedback, and increased student engagement (Alneyadi et al., 2023). These applications demonstrate the potential of AI to improve instructional effectiveness and support teaching and learning practices.
2.1.3 Generative AI
Generative AI refers to AI technologies capable of generating new content based on patterns learned from large-scale datasets, producing original outputs that closely resemble human-created content (Bommasani et al., 2021). In education, Generative AI supports personalized learning, real-time feedback, and academic writing through automated feedback, alternative expressions, and idea generation (Qian, 2025; Zhang et al., 2025).
2.1.4 Gemini
Gemini is a multimodal AI model developed by Google DeepMind that can process and generate information across multiple modalities, including text, images, audio, and video (Pichai & Hassabis, 2023). In addition, Gemini provides students with free access to advanced features through the Google One AI Premium plan, broadening AI accessibility for academic use (Google, 2025).
Using Gemini in student writing presents notable pedagogical and ethical challenges. Overreliance on AI can undermine independent thinking, creativity, and self-editing skills, especially when students substitute AI outputs for their own cognitive effort (Imran & Almusharraf, 2024; Cung et al., 2025). Instructors also voice concerns regarding content accuracy and the potential erosion of problem-solving abilities, highlighting the need for strategic guidance that frames AI strictly as a supplementary aid (Qian, 2025; Yavich, 2025). Furthermore, AI-generated work sparks ethical issues surrounding plagiarism, data privacy, security, and the lack of robust regulatory frameworks for responsible educational use (Rane, 2024; Perera & Lankathilake, 2023).
2.1.5 Perception
Perception refers to an individual’s ability to interpret and make sense of information from the environment (Goldstein, 2014; Hoffman, 2008). It is an active interpretive process shaped by individuals’ prior knowledge, beliefs, and contextual factors. In educational research, perception is therefore important for understanding how learners and educators evaluate instructional tools, teaching practices, and learning environments (Han & Xu, 2025).
Student perceptions refer to learners’ views and beliefs regarding their learning environment and instructional tools (Fraser, 1998). In this study, student perceptions specifically refer to how learners evaluate the use of Gemini in academic writing, particularly in terms of its usefulness, reliability, and relevance.
Lecturers’ perceptions involve their beliefs and evaluations of instructional tools and teaching practices, which can influence their instructional decisions (Borg, 2001; Pajares, 1992). In the context of AI-assisted writing, lecturers’ perceptions of Google Gemini are important in understanding how the technology is integrated into writing instruction, particularly regarding its accuracy, ethical concerns, and potential student dependency.
2.2 Theoretical framework
2.2.1 Technology Acceptance Model (TAM)
TAM, developed by Davis in 1989, is one of the most popular theoretical frameworks that are used for explaining individuals’ acceptance and use of new technology. Derived from the Theory of Reasoned Action, TAM emphasizes two main ideas: PU, defined as the extent to which a person believes that using a system improves task performance, and PEOU, which is how easy people think using the system will be. According to TAM, PU and PEOU jointly influence users’ attitude toward technology. This also ultimately leads to actual system usage (Davis, 1989), while PEOU also directly affects PU. In this study, TAM provides a theoretical foundation for examining students’ perceptions of Google Gemini in terms of its usefulness and ease of use in supporting academic writing, and how these perceptions influence their attitude and intention to adopt Gemini as a writing aid in the Writing 2 course.
2.2.2 Theory of Planned Behavior (TPB)
TPB, proposed by Ajzen in 1991, extends the Theory of Reasoned Action by incorporating PBC as an additional determinant of BI (Ajzen, 1991). According to TPB, an individual’s intention to perform a behavior is influenced by three factors: Attitude toward the behavior, SN, and PBC, which reflects the perceived ease or difficulty of performing the behavior based on prior experience and anticipated constraints (Ajzen, 1991).
Attitude refers to an individual’s positive or negative evaluation of performing a behavior, while SN involve perceived social pressure from significant others, such as teachers, peers, or institutions. PBC reflects individuals’ beliefs about their capability and available resources to perform the behavior (Ajzen, 1991). In this study, TPB is employed to explain how lecturers’ and students’ attitude toward using Gemini, perceived social influences (e.g., lecturers’ expectations or peer norms), and perceived control over using AI tools affect their intention to integrate Gemini into their academic writing practices. - Methodology
3.1 Instruments
This study employed an explanatory sequential mixed-methods design to investigate students’ and lecturers’ perceptions of using Gemini in the Writing 2 course at QNU. The study consisted of two sequential phases, beginning with a quantitative phase to provide an overall picture of students’ perceptions, attitude, and behavioural intentions towards using Gemini, followed by a qualitative phase to further explain and develop the quantitative findings. Data were collected through a student questionnaire, semi-structured interviews with students, and an open-ended questionnaire for lecturers.
The student questionnaire was designed to investigate students’ perception, attitude, and BI towards using Gemini in the Writing 2 course. It included mainly closed-ended items measured on a five-point Likert scale. The questionnaire was composed of 30 observed variables across six constructs: PU, PEOU, SN, ATT, PBC, and BI. These constructs were used to examine students’ perceptions of Gemini and the factors associated with their attitude and BI toward its use in writing. In addition to the closed-ended items, the student questionnaire included one open-ended question asking participants to provide practical recommendations and suggestions regarding the use of Gemini in the Writing 2 course. This question provided qualitative information that complemented the quantitative results.
The second instrument was a semi-structured interview conducted with five students. The interview questions focused on students’ perceptions, attitude, and BI toward using Gemini. The semi-structured format allowed the researcher to follow prepared questions while also exploring participants’ views in greater detail when relevant.
The third instrument was an online open-ended questionnaire for lecturers. It was designed to collect lecturers’ perspectives on students’ use of Gemini and its effects on teaching and learning writing. The questionnaire consisted of six sections: demographic information; awareness and experience with Gemini; attitude toward students’ use of Gemini; perceived impact on teaching and learning writing; risks, limitations, and challenges; and future orientation and recommendations. The open-ended format allowed lecturers to describe their experiences, concerns, and suggestions in detail.
3.2 Participants and Sampling
The participants were second-year students majoring in English Language and English Teacher Education at QNU who had completed the Writing 2 course. As Writing 2 primarily focuses on letter writing, including both formal and informal correspondence, these students were considered appropriate for investigating their perceptions and experiences of using Gemini in writing.
The sample size was determined based on guidelines for Exploratory Factor Analysis (EFA) and multiple regression analysis. According to Hair et al. (2013), the minimum sample size for EFA should be five times the number of observed variables. With 30 observed variables, at least 150 participants were required. Tabachnick and Fidell (1996) also suggested a minimum sample size of 50 + 8m, resulting in 82 participants for the four independent variables in this study. Therefore, 200 valid responses were considered sufficient for the quantitative phase.
For the quantitative component, convenience sampling was employed. The questionnaire was distributed online via Google Forms through social media, class chat groups, and personal networks. Of the 227 responses collected, 200 valid responses were included in the analysis.
For the qualitative component, five second-year students from the two target majors participated in semi-structured interviews. They were selected based on their willingness to participate and their ability to provide relevant experiences with Gemini in Writing 2. In addition, three lecturers teaching writing-related courses at the Faculty of Education and the Faculty of Foreign Languages at QNU completed an online open-ended questionnaire. They were selected based on their experience in teaching writing-related courses.
3.3 Data collection and analysis procedure
The study employed an explanatory sequential mixed-methods design, in which the quantitative phase was conducted first, followed by a qualitative phase to explain and complement the quantitative findings. Data collection and analysis therefore proceeded in two main phases.
For the quantitative phase, the student questionnaire was distributed online via Google Forms. The 200 valid responses were analyzed using SmartPLS 4.0 for Cronbach’s Alpha and to assess the measurement and structural models through PLS-SEM. The analysis examined factor loadings, Composite Reliability (CR), Average Variance Extracted (AVE), discriminant validity, Variance Inflation Factor (VIF), path coefficients, R², and p-values.
The qualitative phase was conducted after the quantitative analysis to provide deeper explanations of the survey findings. Each student interview lasted approximately 15 minutes, was audio-recorded with the participants’ consent, and transcribed verbatim. The semi-structured interview guide explored students’ perceptions, attitudes, and behavioural intentions regarding the use of Gemini in Writing 2. The interviews were conducted in English. In addition, the three lecturers completed an online open-ended questionnaire addressing the perceived benefits, challenges, attitudes, and recommendations concerning the use of Gemini in Writing 2. The qualitative data were analyzed thematically and interpreted alongside the quantitative findings to provide further insights into students’ experiences and lecturers’ perspectives on the use of Gemini in writing instruction.
3.4 Integrated TAM-TPB-inspired model
Based on TAM and TPB, this study recommends an integrated research model to examine student perceptions of using Google Gemini for writing skills learning, especially in the Writing 2 course.
Figure 3.1 Integrated TAM-TPB-inspired model
Based on TAM, PU and PEOU are assumed to influence users’ attitude toward Gemini by reflecting its effectiveness and ease of use in academic writing. From TPB, SN and PBC are incorporated to examine the effects of social influence and users’ perceived ability and resources on their attitude toward Gemini use. Attitude was positioned as an intervening construct linking students’ perceptions of Gemini with their behavioural intention to use it, which represents users’ willingness and intention to use Gemini for writing activities. Overall, the model explains how PU, PEOU, social influence, and perceived control shape attitude and subsequently influence students’ BI to use Google Gemini in EFL writing.
3.5 Hypotheses
H1: PU positively influences students’ attitude toward using Gemini in the Writing 2 course.
H2: PEOU positively influences students’ attitude toward using Gemini in the Writing 2 course.
H3: SN positively influences students’ attitude to use Gemini in the Writing 2 course.
H4: PBC positively influences students’ attitude toward using Gemini in the Writing 2 course.
H5: Students’ attitude toward using Gemini positively influences their behavioural intention to use Gemini in the Writing 2 course.
4. Result
4.1. Result: Students’ perspective
4.1.1. Quantitative Findings
4.1.1.1. Reliability and Validity
The measurement model demonstrated satisfactory reliability and validity. As shown in Table 4.1, none of the factor loadings are below 0.708; therefore, no variables need to be excluded (Hair et al., 2019). The results also showed that Cronbach’s Alpha and CR values for all constructs were above 0.70, indicating satisfactory internal consistency. In addition, AVE values ranged from 0.671 to 0.752, above the 0.50 threshold, confirming convergent validity (Fornell & Larcker, 1981).
Table 4.1 Measurement Model Reliability and Convergent Validity
Constructs
Cronbach’s Alpha
CR
AVE
PU
0.885
0.916
0.684
PEOU
0.886
0.912
0.676
SN
0.879
0.911
0.671
Attitude (ATT)
0.900
0.926
0.716
PBC
0.894
0.921
0.700
BI
0.917
0.938
0.752
4.1.1.2. Structural model Analysis
The analysis of the structural equation model showed that all VIF values in this study are below 5, suggesting no problematic collinearity (Hair et al., 2019). Moreover, in table 4.2, all five hypothesized relationships were supported. Among the hypotheses, ATT → BI showed the strongest path coefficient (β = 0.712, p < 0.001), followed by PU → ATT (β = 0.431, p < 0.001), PBC → ATT (β = 0.394, p < 0.001), PEOU → ATT (β = 0.250, p < 0.001), and SN → ATT (β = 0.237, p < 0.001). Finally,the model showed 49.2% of the variance in ATT and 50.5% of the variance in BI toward the use of Gemini in the Writing 2 course, showing a moderate predictive capability for both constructs.
Table 4.2 Structural Model Analysis
Hypothesis
Relationship
Original sample (O)
Sample mean (M)
Std.
T Values
P Values
Result
H1
PU ➝ ATT
0.431
0.430
0.050
8.673
< 0.001
Confirmed
H2
PEOU ➝ ATT
0.250
0.254
0.048
5.184
< 0.001
Confirmed
H3
SN ➝ ATT
0.237
0.238
0.052
4.556
< 0.001
Confirmed
H4
PBC ➝ ATT
0.394
0.393
0.050
7.943
< 0.001
Confirmed
H5
ATT ➝ BI
0.712
0.713
0.032
22.246
< 0.001
Confirmed
4.1.2. Qualitative Findings
4.1.2.1 Perceived Benefits and Ease of Use
Writing efficiency was one of the most frequently mentioned benefits of using Gemini. All five students (A1-A5) reported that Gemini helped them complete writing tasks more quickly. A2 and A3 emphasized that it helps them “finish assignments faster.” Similarly, its support for idea generation was highlighted by all participants. A4 described Gemini as a “brainstorming partner,” while other participants emphasized its usefulness in organizing ideas and refining vocabulary. Four students (A1-A4) noted that Gemini supported vocabulary development by suggesting alternative or more advanced words. Overall, Students valued the immediate practical benefits of the tool, particularly in overcoming difficulties during the writing process.
Regarding PEOU, students generally found Gemini easy to access and operate at a basic technical level, with none reporting significant technical difficulties. However, using the tool effectively was more challenging, particularly in constructing and refining effective prompts. A4 described prompt construction for complex academic tasks as a “distinct skill” requiring practice, while A2 reported having to revise prompts repeatedly because Gemini sometimes misunderstood the intended meaning. Participants also noted that some responses could be general, lengthy, or irrelevant responses, requiring further prompt adjustment. Although Gemini was technically accessible, its effective use required students to develop prompting skills and exercise judgment when evaluating its responses.
4.1.2.2 Social Influence and PBC
Based on the five interviews, SN were primarily shaped by peers, while lecturers played a relatively minor role. All five participants reported that Gemini was commonly used among their peers and was becoming increasingly normalized in academic work. This influence operated through two main effects: some students felt pressure to keep up with others, while others were encouraged by observing successful uses of Gemini. Peers also shaped students’ use by sharing prompting strategies and AI-generated outputs. A1 noted that “Classmates influence me by sharing tips and experiences,” In contrast, lecturers had a more limited and indirect influence, as several participants reported receiving little explicit guidance on AI use.
Students’ PBC was strongly dependent on the circumstances in which Gemini was used. Participants generally felt more in control when they had their own ideas or drafts and actively evaluated, selected, or rejected AI-generated suggestions. However, this sense of control became weaker when they faced a blank page, heavy workloads, stress, or low confidence. A5 noted that having an initial draft made it easier to direct Gemini, whereas having no ideas increased the likelihood of following its suggestions. Similarly, A3 and A4 reported greater reliance on Gemini under stress or time pressure.
4.1.2.3 Attitude and Intention to Continue Using Gemini
An analysis of the interviews reveals that students reported both positive perceptions and significant reservations toward using Gemini for their writing tasks. Almost all students express a very positive attitude toward the psychological comfort and stress relief that the tool provides. As A2 noted, “It makes writing faster and much less stressful.” Students also valued Gemini for providing ideas, different perspectives, and support in identifying mistakes.
On the other hand, these benefits were accompanied by feelings of guilt and concern about overdependence on AI. A2 and A4 specifically noted that they described discomfort with passively accepting AI-generated responses. Additionally, several students worried that relying too heavily on Gemini could reduce their confidence in their true abilities , and hinder the development of independent thinking and writing skills.
All five participants (A1-A5) expressed an intention to continue using Gemini in their future writing practices, although the extent and purposes of use varied. A1 and A2 described Gemini as “useful”, while A3 valued its convenience. A4 and A5 intended to use it more selectively, particularly for research support, editing, and reflection. Importantly, students did not view continued use as unrestricted reliance. They emphasized careful and controlled use, including avoiding excessive dependence, rewriting AI-generated content in their own words, and using Gemini after developing an initial draft. A4 and A5 further described Gemini as a “research assistant” and a tool for “editing and reflection.” Students also supported allowing Gemini in Writing 2 with certain restrictions or guidelines, suggesting that its use should support rather than replace their own thinking and writing.
4.2. Result: Lecturers’ perspective
4.2.1 Lecturers’ Perception
4.2.1.1 Perceived Benefits of Gemini
Lecturers (B1-B3) indicated that Gemini offers several pedagogical benefits in writing instruction. Three main themes emerged: idea generation, language improvement, and support for teaching methods. First, lecturers reported that Gemini helps students overcome writer’s block, generate ideas, and organize their thoughts more efficiently. Second, it supported language and structural refinement by identifying grammatical errors and improving clarity and tone. However, lecturers emphasized that Gemini should be used to refine ideas rather than generate complete final products. Third, its integration encouraged lecturers to adopt more process-oriented teaching and assessment, with greater emphasis on students’ thinking, revision, and critical evaluation of AI-generated outputs. Overall, Gemini was viewed as a useful instructional support tool, provided that lecturers guide students to use it responsibly without replacing independent thinking.
4.2.1.2 Perceived Drawbacks of Gemini in Writing Teaching
Lecturers (B1-B3) identified several potential drawbacks of integrating Gemini into writing courses, with students’ overreliance on AI emerging as the most prominent concern. B1 noted that excessive use of Gemini may prevent students from “struggling through an idea” independently, while B2 and B3 similarly pointed to dependence on AI for idea generation and essay organization, potentially resulting in superficial learning. Another major concern was the possible decline in critical thinking. B2 highlighted “reduced critical thinking,” and B3 warned of the “potential erosion” of these skills when students rely on ready-made ideas and language. Lecturers also expressed concerns about students’ authentic voice and creativity, as frequent use of AI-generated expressions may weaken individual writing styles (B2, B3). In addition, B2 raised academic integrity concerns when AI is used without clear guidelines. Overall, the lecturers viewed Gemini as useful but cautioned that uncontrolled use could undermine independent thinking, critical thinking, creativity, and writing identity, underscoring the need for clear pedagogical guidance.
4.2.2 Lecturers’ Attitude
4.2.2.1 Lecturers’ Attitude toward Students’ Use of Gemini in the Writing 2 course
Although lecturers acknowledged several risks, they did not consider Gemini inherently harmful. Instead, their responses reflected a cautious and conditional acceptance of the tool. B1 described Gemini as a “double-edged sword,” while B2 recognized its usefulness for brainstorming and checking grammar or tone, provided that students did not use it to produce entire texts. However, lecturers consistently expressed concerns about excessive reliance on AI. B1 warned that repeated use could weaken students’ critical thinking, and B3 highlighted “diminished independent thinking,” suggesting a risk of cognitive offloading. Concerns were also raised about students’ writing identity, as B1 and B3 noted that overuse could weaken individual voice, creativity, and personal argumentation. More broadly, the lecturers emphasized potential long-term effects on academic writing development in the EFL context, rather than focusing solely on immediate performance or plagiarism. Overall, their perceptions reflected conditional acceptance, with Gemini viewed as a useful support tool only when its use is carefully guided and controlled.
4.2.2.2 Lecturers’ Recommendations
The findings indicated that lecturers generally supported the integration of Gemini into writing instruction, but emphasized that it should serve as a support tool rather than a replacement for students’ writing efforts. B1 recommended a “Human-First framework,” while B2 suggested using Gemini for brainstorming, language support, and revision instead of generating complete texts. Similarly, B3 proposed using AI as a “style coach” rather than a “content generator.” Lecturers also stressed the need for clear guidelines and teacher supervision. B1 highlighted that students should use Gemini under teachers’ guidance, while B3 suggested activities that require students to critically compare AI-generated versions with their own drafts. In addition, transparency and ethical responsibility were emphasized. B2 recommended that students acknowledge AI use, and B3 suggested submitting both original and AI-assisted versions along with a brief reflection. Overall, lecturers advocated a guided and transparent approach that preserves human agency and ensures that AI enhances, rather than replaces, students’ learning and writing development.
4.3 Lecturers’ and Students’ Views on Gemini in the Writing 2 course
The findings indicated that both students and lecturers favored neither a complete ban nor unrestricted use of Gemini in the Writing 2 course. Instead, both groups expressed a preference for conditional and controlled use, with the tool being used within clear boundaries to support learning.
Both students and lecturers recognized several benefits of Gemini in the writing process. They noted its usefulness for generating ideas, organizing writing, and improving language quality, as well as for saving time and supporting draft revision. At the same time, both groups expressed concerns about overreliance on AI. They believed that excessive use of Gemini could reduce students’ independent thinking, limit the development of their own ideas, and negatively affect writing development.
Despite these similarities, the two groups differed in their perspectives and priorities. Students tended to focus more on the practical and immediate benefits of Gemini, particularly its convenience, efficiency, and ability to support assignment completion. Although they intended to continue using the tool, they were also aware of potential risks such as reduced independent thinking and loss of individuality in their writing. Lecturers, in contrast, adopted a more cautious and long-term perspective. Their concerns extended beyond immediate writing performance to the potential decline in critical thinking, loss of students’ individual writing voices, and difficulties in accurately assessing students’ actual writing abilities.
5. Discussion
These results show that students generally had a favorable but conditional attitude toward using Gemini in the Writing 2 course. The findings were consistent with the proposed model, with PU, PEOU, SN, and PBC significantly influencing ATT, while ATT showed the strongest effect on BI (β = 0.712, p < 0.001). PU had a significant positive effect on ATT (β = 0.431, p < 0.001), followed by PBC (β = 0.394, p < 0.001). The qualitative findings helped explain these relationships, as students valued Gemini for saving time, generating ideas, organizing writing, and improving language. For example, A4 described Gemini as a “brainstorming partner,” while A3 reported that it was helpful when they had “no ideas at all.” Meanwhile, students’ sense of control depended on having their own ideas or drafts and being able to evaluate AI-generated suggestions.
PEOU (β = 0.250, p < 0.001) and SN (β = 0.237, p < 0.001) had weaker effects. Although students found Gemini technically easy to use, they experienced difficulties in writing effective prompts and evaluating irrelevant or overly general responses. Peer use also influenced students’ adoption, whereas lecturers appeared to have a more limited influence due to limited explicit guidance.
Most notably, ATT had the strongest effect on BI (β = 0.712, p < 0.001). Consistently, all five students intended to continue using Gemini, particularly for brainstorming, editing, and research support. However, they also expressed concerns about overreliance, reduced independent thinking, and loss of individual writing voice. Lecturers shared these concerns but emphasized the longer-term effects on writing development and assessment. Overall, both groups favored controlled and purposeful use rather than either a complete ban or unrestricted reliance on Gemini.
6. Conclusion and Implications
6.1 The Major Findings of the Study
The findings suggest that students generally hold favorable perceptions of Gemini in the Writing 2 course and have a strong intention to continue using it. Students perceive Gemini as useful and easy to use, especially in developing ideas, organizing letter structure, and improving vocabulary. Interviewees reported that Gemini also makes the writing process more convenient and less stressful, which helps increase students’ confidence and motivation. Lecturers likewise recognize Gemini as a valuable support tool for brainstorming, organizing ideas, overcoming writer’s block, and improving grammar, clarity, and tone. However, they express concerns about students’ overreliance on AI, which may reduce independent thinking, creativity, and authentic writing voice. Assessment is also challenging due to the difficulty of distinguishing students’ original work from AI-assisted writing. In addition, the study argues that Gemini should be used selectively as a writing aid, with explicit guidance, training, and clear course policies to ensure responsible AI use and reflective, ethically informed writing practices. Overall, both students and lecturers recognize Gemini as a useful tool that can support different stages of the writing process, while also acknowledging that its benefits can only be fully achieved when the tool is used appropriately and with clear instruction.
6.2 Implications
6.2.1 Theoretical Implications
This study contributes to clearer understanding of students’ perception of Gemini in supporting the writing process. Students’ perceptions are defined through PU, PEOU, SN, and PBC, with PU having the strongest effect on students’ attitude. This suggests that students value and prioritise the benefits that Gemini brings in improving writing skills. In addition, attitude is confirmed as an important mediating role in the relationship between cognitive factors and BIs. This shows that students' intention to use Gemini is mainly formed through their positive evaluation of the tool. This result also provides evidence of support for the applicability of TPB theory and the TAM in explaining technology acceptance behavior in the learning context.
6.2.2 Practical Implications
The research suggests that responsible use of Gemini in the Writing 2 course requires clear boundaries rather than either unrestricted use or a complete ban. Students are encouraged to use Gemini as a support tool rather than rely entirely on it. They can utilize Gemini to develop ideas, improve sentence structure, check grammar and vocabulary, and edit drafts, while actively evaluating and adjusting the AI-generated content to enhance their writing skills and critical thinking.
For lecturers, instead of simply allowing students to use the tool, instructors should provide clear guidance and design activities that require students to compare and evaluate AI-generated content to develop critical thinking. Furthermore, lecturers should offer guidance on writing effective prompts, verifying information, and adjusting AI-generated content to improve learners’ skills in using the tool. For letter-writing tasks, lecturers could encourage students to produce an initial draft before consulting Gemini and reflect on how they revised AI-generated suggestions. Workshops or training sessions can also help students make appropriate commands and better evaluate the quality of AI-generated content.
6.3 Limitations and Recommendations
6.3.1 Limitations
This study has several limitations that should be considered. First, the research focuses only on students in the Writing 2 course, so the findings may not fully reflect other courses or educational contexts. Another limitation is the use of self-reported data, as participants’ responses may be affected by personal opinions and social desirability. Moreover, the sample only includes second-year students and a limited number of lecturers, making the findings less representative of the wider population and affecting the scope of the study. Finally, the study does not address all gaps identified in previous research but focuses on those most relevant to its research objectives.
6.3.2 Recommendations
Based on the study results, several recommendations for future research can be suggested. First, future studies should examine which Gemini functions, such as idea generation, grammar correction, and writing structure improvement, have the greatest impact on learning outcomes. Researchers should also investigate students’ actual usage of Gemini, including how often they use it at different stages of writing and the strategies they apply. In addition, longitudinal studies should explore how AI use affects students’ writing skills, critical thinking, and personal writing styles over time. Finally, further research could explore effective ways to integrate AI into writing classes while establishing clear guidelines for its appropriate use.
7. References
Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50(2), 179-211. https://doi.org/https://doi.org/10.1016/0749-5978(91)90020-T
Alneyadi, S., Wardat, Y., Alshannag, Q., & Abu-Al-Aish, A. (2023). The effect of using smart e-learning app on the academic achievement of eighth-grade students. Eurasia Journal of Mathematics, Science and Technology Education, 19(4), 22-48. https://doi.org/10.29333/ejmste/13067
Baharudin, F., Laily Ramli, N. H., Mohd Habali, A. H., Azlina Azmi, A., & Hanim Rahmat, N. (2023). Process of writing: The challenges in writing skill among ESL learners. International Journal of Academic Research in Business and Social Sciences, 13(10), 32-50. https://doi.org/10.6007/ijarbss/v13-i10/18649
Barr, A., & Feigenbaum, E. A. (1981). The handbook of Artificial Intelligence vol. 1 / ed. by Avron Barr and Edward A. Feigenbaum. HeurisTech Press.
Bell, J., & Burnaby, B. (1984). A handbook for ESL literacy. OISE Press.
Borg, M. (2001). Teachers’ beliefs. ELT Journal, 55(2), 186-188. https://doi.org/10.1093/elt/55.2.186
Cung, L. T., Hoang, U. P. T., Dinh, A. N., & Bui, T. H. (2025). University students’ perceptions of AI application in writing skills in Vietnam: A systematic review. Vietnam Journal of Education, 9(2), 325-334. https://doi.org/10.52296/vje.2025.533
Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly, 13(3), 319-340. https://doi.org/https://doi.org/10.2307/249008
E Bruce Goldstein. (2016). Sensation and perception. Access And Diversity, Crane Library, University Of British Columbia.
Fornell, C., & Larcker, D. F. (1981). Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research, 18(1), 39-50. https://doi.org/https://doi.org/10.2307/3151312
Fraser, B. J. (1998). Classroom Environment Instruments: Development, Validity and Applications. Learning Environments Research, 1(1), 7-34. https://doi.org/10.1023/a:1009932514731
Hair, J., Black, W., Babin, B., & Anderson, R. (2019). Multivariate Data Analysis 8th edition.
Hair, J. F., Risher, J. J., Sarstedt, M., & Ringle, C. M. (2019). When to use and how to report the results of PLS-SEM. European Business Review, 31(1), 2-24. https://doi.org/https://doi.org/10.1108/EBR-11-2018-0203
Han, T., & Xu, G. (2025). The relationship between learning environment perception, achievement goals, and the undergraduate deep learning approach: A longitudinal mediation model. Journal of Intelligence, 13(2), 19. https://doi.org/10.3390/jintelligence13020019
Hoffman, D. D. (1998). Visual intelligence: How we create what we see. W W Norton & Co.
Imran, M., & Almusharraf, N. (2024). Google gemini as a next generation AI educational tool: A review of emerging educational technology. Smart Learning Environments, 11(1), 1-8. https://doi.org/10.1186/s40561-024-00310-z
Kaplan, A., & Haenlein, M. (2019). Siri, Siri, in my hand: Who’s the fairest in the land? On the interpretations, illustrations, and implications of artificial intelligence. Business Horizons, 62(1), 15-25. https://doi.org/10.1016/j.bushor.2018.08.004
Knoch, U., May, L., Macqueen, S., Pill, J., & Storch, N. (2016). Transitioning from university to the workplace: Stakeholder perceptions of academic and professional writing demands. IELTS Research Reports Online Series.
Pajares, M. F. (1992). Teachers’ beliefs and educational research: Cleaning up a messy construct. Review of Educational Research, 62(3), 307-332. https://doi.org/https://doi.org/10.2307/1170741
Perera, P., & Lankathilake, M. (2023). Preparing to revolutionize education with the multi-model GenAI tool Google gemini? A journey towards effective policy making. Journal of Advances in Education and Philosophy, 7(8), 246-253. https://doi.org/10.36348/jaep.2023.v07i08.001
Phạm, T. T. T. (2024). Ứng dụng AI trong dạy học ngoại ngữ tiềm năng và thách thức. Journal of Educational Equipment: Applied Research, 2(319), 23-24.
Qian, Y. (2025). Pedagogical applications of generative AI in higher education: A systematic review of the field. TechTrends, 69(5), 1105-1120. https://doi.org/10.1007/s11528-025-01100-1
Rane, N. (2024). Enhancing the quality of teaching and learning through gemini, ChatGPT, and similar generative artificial intelligence: Challenges, future prospects, and ethical considerations in education. TESOL and Technology Studies, 5(1), 1-6. https://doi.org/10.48185/tts.v5i1.1000
Rishi, B., Hudson, D. A., Ehsan, A., Altman, R., Arora, S., von Arx, S., Bernstein, M. S., Bohg, J., Bosselut, A., Brunskill, E., Brynjolfsson, E., Buch, S., Card, D., Castellon, R., Chatterji, N., Chen, A., Creel, K., Davis, J. Q., Demszky, D., … Liang, P. (2021). On the opportunities and risks of foundation models (2108.07258V3). arXiv. https://arxiv.org/abs/2108.07258v3
S, K. (2024). Enhancing writing proficiency through AI-Powered feedback: A quasi-experimental study using google gemini. LinguaEducare: Journal of English and Linguistic Studies, 1(2), 83-96. https://doi.org/10.63324/h6q1ak58
Sundar Pichai. (2023, December 6). Introducing gemini: Our largest and most capable AI model. Google. https://blog.google/innovation-and-ai/technology/ai/google-gemini-ai/#sundar-note
Tabachnick, B. G., & Fidell, L. S. (1996). Using multivariate statistics. Harpercollins College Division.
Woodward, J. (2025, April 17). College students in the U.S. are now eligible for the best of Google AI - And 2 TB storage - For free. Google. https://blog.google/products-and-platforms/products/gemini/google-one-ai-premium-students-free/
Yavich, R. (2025). Will the use of AI undermine students independent thinking? Education Sciences, 15(6), 669-669. https://doi.org/https://doi.org/10.3390/educsci15060669
Zhang, Z., Aubrey, S., Huang, X., & Chiu, T. K. F. (2025). The role of generative AI and hybrid feedback in improving L2 writing skills: A comparative study. Innovation in Language Learning and Teaching, 1-19. https://doi.org/10.1080/17501229.2025.2503890
