The impact of ethical
generative AI use on academic integrity and writing performance among EFL
university students
El
impacto del uso ético de la IA generativa en la integridad académica y el
rendimiento en la redacción de los estudiantes universitarios de inglés como
lengua extranjera
Piedad Rosario Guijarro
Paguay*
Marco Antonio Aquino Rojas*
Introduction
Digital technologies are now widely recognized for
their transformative role in higher education, as they facilitate the process
of teaching and learning foreign languages (Fuentes et al., 2024;
Zawacki-Richter et al., 2019). For a long time, the teaching of English as a
foreign language (EFL) has utilized automated tools such as automated writing
assessment (AWE), learning management systems (LMS), and machine translation
platforms, all of which have served as complementary resources for acquiring
vocabulary and improving grammatical structures (Shabir, 2025).
According to authors such as Chicaíza
et al. (2023), Dahlan (2026), and Kasneci et al.
(2023), the rapid growth of generative artificial intelligence, such as
ChatGPT, has marked a paradigm shift in language teaching. Unlike traditional
platforms, these tools not only correct mechanical or superficial errors but
also function as interactive platforms that.
We can create complex co-compositions in real time
that achieve a high degree of discursive fluency (Dahlan, 2026; Shabir, 2025).
As such, generative AI offers a wide range of
possibilities for writing in EFL, as it operates continuously and can act as a
personalized tutor (Junaid et al., 2024; Zakaria et al., 2026). It is important
to note that academic writing poses a major challenge in second-language
learning, as students must be able to master grammatical accuracy, vocabulary,
textual cohesion, and coherence in their arguments (Aljasser,
2025; Junaid et al., 2024). In this regard, the use of artificial intelligence
enables instant feedback and interaction, in addition to providing
contextualized suggestions and adjusting the tone of the writing according to
the purpose of the text (Aljasser, 2025; Huerta et
al., 2024; Nelson et al., 2025).
According to Aminah and Aly (2026) and Zakaria et al.
(2026), the use of artificial intelligence helps reduce the stress and anxiety
associated with writing, while strengthening students’ motivation and
confidence. In some cases, engaging in reflective and comparative activities
between one’s own written texts and those generated by AI enhances cognitive
and metalinguistic autonomy (Dahlan, 2026; Md Nawi et al., 2025).
Having outlined the opportunities presented by the use
of AI, it is important to highlight the ethical and pedagogical concerns that
are emerging in education worldwide (Cotton et al., 2023; Dwivedi et al.,
2023). Given the ease of providing brief instructions (prompts), the debate is
intensifying over what limits should be set to verify authorship and ensure
that students’ work is original (Gallent-Torres et
al., 2023; Lund et al., 2025). Failing to set limits or writing irresponsibly
leads to plagiarism, the misappropriation of intellectual effort, and even
cognitive offloading, where students fail to develop critical thinking,
analyze, or articulate their own positions and criteria, leaving everything in
the hands of an algorithm (Aminah & Aly, 2026; Dahlan, 2026; Shabir, 2025).
Added to this problem are the limitations of language
models, such as the generation of false or inaccurate information and
fictitious bibliographic references phenomena known as “hallucinations” which
pose a clear risk when generated content is accepted without a reasoned
verification process (Dahlan, 2026; Huang et al., 2025).
Given that traditional methods for assessing writing
are flawed, authors such as Lund et al. (2025) and Nelson et al. (2025) point
out that a total ban on the use of AI platforms is ineffective and unfeasible
in the context of today’s higher education. The use of automated AI detectors
is also not the solution, due to their inaccuracy and the bias they exhibit
toward non-native English writers (Dahlan, 2026; Gallent-Torres
et al., 2023).
Therefore, current literature promotes the ethical use
of AI, guiding the design of assessment tasks toward the writing process and
self-regulation (Dahlan, 2026; Md Nawi et al., 2025; Paniagua Urbáez et al., 2025). Studies on behavioral models and
academic integrity show that knowledge of institutional norms and rules is not
sufficient to prevent dishonesty; rather, beliefs about ethics, moral
responsibility, and information literacy skills are the determining factors
that lead students to use the tools at their disposal in a transparent and
legitimate manner (Huang et al., 2025; Lund et al., 2025).
Despite the surge in publications and research on the
use of artificial intelligence at universities worldwide, there are no
quantitative national studies on the subject. The knowledge gap lies in
determining the correlation that might exist between these two variables within
Ecuador’s public higher education system. Thus, this study seeks to directly
analyze how the responsible and ethical use of AI relates to academic integrity
and the true quality of writing by Spanish-speaking students learning English
(EFL).
The study population consists of sixth-semester
students in the School of Business Administration at the Escuela Superior
Politécnica de Chimborazo (ESPOCH) enrolled during the academic period from
March to July 2026. This population is the most appropriate because, in this
specific field, academic proficiency in the English language is considered a
professional competency of great strategic value.
Consequently, the main objective of this research is
to analyze the impact of the responsible and ethical use of generative
artificial intelligence on academic integrity and writing performance among EFL
students at the university, by examining usage practices and written products,
with the aim of aligning the use of artificial intelligence with the ethics of
higher education.
The research poses the following question: How does
the responsible and ethical use of generative artificial intelligence affect
academic integrity and writing performance among university students studying
English as a foreign language?
The core premise of the research is that the ethical
and transparent adoption of AI tools is positively correlated with academic
integrity and leads to significant improvements in the analytical and
linguistic quality of university students’ writing in English.
From a methodological perspective, this study was
quantitative and applied, employing a non-experimental, cross-sectional,
correlational design using a census sample comprising all sixth-semester
students in the March–July 2026 term of the Business Administration program at
ESPOCH.
The research aims to provide solid empirical evidence
to support the responsible incorporation of artificial intelligence into
English instruction in higher education, through structured questionnaires and
rubrics for evaluating analytical writing.
This literature review examines the impact of
generative artificial intelligence on written production and academic integrity
in English as a Foreign Language (EFL) context.
Cognitive models of written composition in EFL and AI
as scaffolding
Academic writing is based on a highly complex
cognitive and communicative process that requires writers to have extensive
knowledge of the subject matter, critical thinking skills, and mastery of
formal discourse. In EFL, there are additional demands: in addition to
overcoming the challenges of writing in a second language, writers must
organize their work consistently, recognizing that writing is a multi-level
activity that involves the use of various cognitive mechanisms.
According to Flower and Hayes (1981), writing is
related to the cognitive process, and they argue that writing is a recursive
activity, that is, it involves a constant back-and-forth. In summary, when
writing, one goes through three phases: planning, which involves generating and
organizing ideas; translation, which is the conversion of ideas or thoughts
into text; and revision, which is the evaluation and editing of the content.
The English as a Foreign Language (EFL) students in this study experience a
cognitive load during the translation phase that involves translation, grammar,
vocabulary, and argumentative structure in the second language.
The tools that existed before the rise of AI were more
mechanical and on a smaller scale, whereas the current use of AI and the advent
of LLMs have transformed the ecosystem into a cognitive scaffold. According to
authors such as Dahlan (2026) and Md Nawi et al. (2025), AI not only intervenes
in the revision phase through stylistic corrections but also becomes involved
recursively during the writing phase through the generation of outlines; in the
translation phase, through lexical suggestions and discursive rephrasing.
However, experimental studies by Junaid et al. (2024)
and Shabir (2025) show that the effectiveness of this scaffolding depends on
the student’s level of engagement. In other words, if the student engages in a
collaborative and iterative dialogue with the tool, it can help the student
achieve discursive fluency and can beneficially reduce the student’s
difficulties in writing; conversely, if the student adopts a passive attitude,
critical thinking is supplanted, leading to “cognitive offloading.”
Psychosocial determinants of academic integrity: The
Theory of Planned Behavior (TPB)
According to Ajzen (1985, 1991), the intention to
engage in a behavior is determined by three constructs attitude toward the
behavior, subjective norms, and perceived behavioral control from the Theory of
Planned Behavior (TPB). This would explain students’ choices regarding the
ethical or unethical use of AI.
Along the same lines, Huang et al. (2025) expanded the
TPB model by incorporating moral obligations and information literacy to
predict the intention to engage in academic dishonesty. Their findings revealed
that high perceived behavioral control (
) and the subjective standards of peers (
) significantly increase the intention to use AI
unethically. Conversely, internalized moral obligations (
) and information literacy (
) have a direct negative influence on the intention to
commit academic misconduct, acting as protective factors for academic
integrity.
For its part, the study by Lund et al. (2025) revealed
a critical gap between institutional policies and actual behavior: knowledge of
institutional policies on AI does not significantly predict student behavior (
). Rather, the determining factor is the internalized
ethical conviction regarding whether the use of AI constitutes cheating. In
other words, these theoretical models show that regulating academic integrity
in the context of AI is not about punitive measures, but about strengthening
individuals’ moral judgment and students’ ethical education.
Critical Literacy in AI and Process-Oriented
Assessment
AI detection systems perform poorly due to
well-documented biases that lead them to falsely classify texts by non-native
authors as AI-generated texts (Liang et al., 2023; Gallent-Torres
et al., 2023); therefore, one proposal for self-determination is “critical
literacy in generative AI.”
Dahlan (2026) conceptualizes this literacy through the
“GenAI Critical Writing Cycle.” This cycle is organized into six recursive
stages. The first stage is orientation based on the limitations of the
discursive genre before using GenAI. A second element to consider is strategic
guidelines that clearly indicate the role, tone, and boundaries. This is to
enable critical evaluation and the identification of biases or ambiguities.
Next, the output is checked, and statements are verified against bibliographic
references and actual academic literature to help avoid “hallucinations.” The
transformation process consists of rewriting the text with a human touch,
according to the author.
Finally, disclosure is expected to be transparent,
specifying the tools used and the percentage or level of assistance employed.
According to authors such as Perkins et al. (2024) and
Md Nawi et al. (2025), the AI Assessment Scale (AIAS) has been established to
classify the level of the tool’s involvement, ranging from Level 1 (no AI used)
to Level 5 (full generation). This methodology shifts the focus from the
typical verification of the final product to the verification of evidence of
the process. Thus, it enables a transparent, author-centered, and ethical
verification of writings in English as a foreign language.
Materials
and Methods
This study employed a quantitative approach to collect and analyze
numerical data. In terms of purpose, the research is applied in nature, as it
seeks to generate empirical evidence to inform pedagogical decision-making and
policy development regarding the responsible use of GenAI in higher education
(Hernández-Sampieri & Mendoza, 2018).
Regarding variable manipulation, a non-experimental design was adopted
because the study variables ethical use of AI, academic integrity, and EFL
writing performance were observed in their natural context without active
intervention or deliberate manipulation by the researchers (Hernández-Sampieri
& Mendoza, 2018). Consequently, the study is not quasi-experimental, as no
experimental treatment, control group, or pre-test/post-test manipulation was
introduced.
The study was cross-sectional because data collection took place at a
single point in time. Furthermore, it was correlational in scope, as it
assessed the degree of association, direction, and statistical significance
between the ethical use of AI, academic integrity, and student writing quality.
The population consisted of the 26 students in their sixth semester at the
School of Business Administration of the Chimborazo Polytechnic Institute who
were regularly enrolled in the English course during during
the academic period of March–July 2026.
Since this was an accessible group, the sample was non-probabilistic and
census-based, with n=26. This decision eliminated the sampling errors
associated with random selection.
The data obtained accurately reflect the dynamics of the group under
evaluation. The inclusion criteria established were that students be officially
enrolled at the institution and that they use AI tools such as ChatGPT, Jasper.
·
ChatGPT, Gemini, and Claude to
help them with their English writing assignments.
·
The students accepted the
digital informed consent form; participation was voluntary.
·
The data remained completely
anonymous and confidential.
With regard to techniques and instruments, two previously validated
instruments were used. The first instrument was a structured digital
questionnaire that collected information on general demographic data, habits
related to the use of AI tools, the frequency of adopting ethical practices in
their use such as checking grammar, verifying references, or openly declaring
that AI has been used as well as attitudes toward dishonesty and perceptions of
authorship. A 5-point Likert scale was used.
The second instrument consisted of a standardized analytical rubric for an
academic argumentative essay in English. The rubric covers four essential
dimensions: grammar, vocabulary, coherence and cohesion, as well as
macro-structure and academic tone. Each element was rated on a scale of 1 to 5.
This assessment was conducted by two trained instructors in the field to
ensure consistency in the analysis.
The collected data were consolidated into a master database and processed
using SPSS Statistics software; all data processing and analysis were carried
out in four sequential phases:
First, data cleaning was performed to verify numerical integrity and check
for missing values or numerical anomalies in the study population. Second,
descriptive statistics—specifically the sample mean (M) and sample standard
deviation (SD)—were calculated to characterize behavioral patterns regarding
the ethical use of Generative Artificial Intelligence, as well as academic
integrity and scores on the English as a Foreign Language (EFL) writing
performance dimensions. Third, given the small sample size (n=26<50), the
Shapiro-Wilk normality test was applied to assess whether the variables
followed a parametric distribution.
Finally, a bivariate correlation analysis was conducted using Pearson’s
correlation coefficient (r) with a two-tailed significance level of α=0.05 to test the research
hypothesis.
Results
This
section presents the results obtained from the census sample (N=26) of
sixth-semester Business Administration students at ESPOCH during the March–July
2026 academic period. This constitutes a systematic evaluation of the proposed
hypothesis; the results will be presented in three sections. The first section
presents descriptive statistics characterizing the dimensions of the ethical
use of AI, academic integrity, and English writing performance. The next
section verifies the assumption of normality using the Shapiro-Wilk test; and
third, it presents the inferential correlation analysis using Pearson’s
correlation coefficient (r).
Descriptive
statistics for the study variables
To
characterize the behavior, central tendency, and dispersion of the variables in
the census sample (
),
we applied the mathematical models of the sample mean (
)
and the sample standard deviation (
),
formally defined as:
Where
represents the observed individual score, 
is the cumulative sum of the group's scores, is the population size, and 
corresponds to the degrees of freedom of the
sample.
Ethical
use of Artificial Intelligence
The
information gathered from the online survey, rated with a Likert scale,
indicated that students mainly utilize AI for grammar correction and feedback.
Feedback
and Grammar Correction: With a cumulative total of
scores of
and
a sum of quadratic residues of
:
Verification of factual claims: With
a total score of
and a sum of quadratic residues of
:
Transparent statement of AI assistance: With
a total score of
and a sum of quadratic residues of
:
Overall
consolidated score for the variable: When
calculating the overall average for the dimension within the group, the total
sum of the composite responses was
and
the total sum of squared deviations was
:
The
overall average score (
)
reflects frequent adoption of ethical practices, while the low standard
deviation (
)
indicates a high degree of homogeneity in the class’s self-regulated behaviors.
The
overall average score (M = 3.89) reflects frequent adoption of ethical
Academic
Integrity
For
the academic integrity variable, the cumulative sum of the questionnaire
responses was
with
a sum of the squares of the residuals of
:
The
students demonstrated a strong commitment to taking responsibility for their
own work (
,
)
and to rejecting uncredited plagiarism (
,
).
With regard to perceived clarity regarding institutional policies on AI, they
showed greater variability (
,
).
Performance
in academic writing in English (EFL)
The
argumentative essays evaluated using the analytical rubric had an overall sum
of mean scores of 
and a sum of squared residuals of 
:
Table
1 presents the descriptive statistics for each dimension of written
performance:
Table
1. Descriptive Statistics for EFL Writing
Performance Dimensions ()
|
Dimension
of writing |
Sum
(∑Xi) |
Mean
(M) |
Standard
Deviation (SD) |
Performance
Level |
|
Grammatical
Accuracy |
106.08 |
4.08 |
0.56 |
High |
|
Coherence
& Cohesion |
100.88 |
3.88 |
0.62 |
Moderate-High |
|
Lexical
Resource |
98.02 |
3.77 |
0.58 |
Moderate-High |
|
Academic
Structure & Tone |
92.04 |
3.54 |
0.65 |
Moderate |
|
Overall
Writing Performance (Average) |
99.32 |
Assessment
of assumptions using the Shapiro-Wilk Normality test
Because
the sample size is small (
),
the distribution of the data was assessed using the Shapiro-Wilk W statistic
formula:
Where
represents the data sorted from smallest to
largest, 
are the tabulated coefficients derived for 
,
and the denominator 
corresponds to the sum of squared residuals
obtained in the descriptive analysis (
for ethical use, 
for integrity y 
for writing).
Table
2 shows the results of the normality test
Table
2. Shapiro-Wilk Normality Test for Main Variables
()
|
Variable |
Sum of Squares ∑(Xi−M)2 |
Statistic
(W) |
df |
p-value
(Sig.) |
|
Ethical
GenAI Use |
8.410 |
0.961 |
26 |
0.412 |
|
Academic
Integrity |
6.500 |
0.954 |
26 |
0.284 |
|
EFL
Writing Performance |
6.002 |
0.968 |
26 |
0.568 |
Note.
Significance level set at 
.
Since
all calculated p-values exceed the significance level of ,
the null hypothesis of normality is not rejected (
).
It is concluded that the data follows a normal parametric distribution, making
the use of Pearson’s correlation coefficient (
)
technically appropriate.
Correlational
Inferential Analysis
Development
of calculus between ethical use (X) and academic integrity (Y):
Sum of the cross-products of deviations:
- Sum of Squares for
Ethical Use
(
):
- Sum of Squares for Academic Integrity
(
):
Substituting
the values into the formula:
Table
3 shows the paired correlation matrix obtained for the three study variables:
Table
3
Pearson
Correlation Matrix Among Study Variables ()
|
Variables |
(1)
Ethical GenAI Use |
(2)
Academic Integrity |
(3)
Writing Performance |
|
(1)
Ethical GenAI use |
1.000 |
0.624** |
0.581** |
|
(2)
Academic integrity |
0.624** |
1.000 |
0.492* |
|
(3)
EFL writing performance |
0.581** |
0.492* |
1.000 |
*
Correlation is significant at the 0.05 level (2-tailed).
Correlation
is significant at the 0.01 level (2-tailed).
Interpretation
of results and conclusions regarding the hypothesis
The value 
indicates a moderate-to-high, statistically
significant positive correlation. It shows that students who use the IAG with
transparent and self-regulated practices demonstrate a greater commitment to
academic honesty.
A direct and
significant positive correlation has been confirmed. This indicates that the
use of AI as a language support tool and for formative feedback leads to higher
grades on argumentative essays written in English.
A moderate positive
association is also observed between students' ethical attitudes and the
quality of their written work.
(p < 0.05), the research hypothesis is
accepted. It is concluded that the ethical and transparent adoption of
generative artificial intelligence is directly and significantly related to
academic integrity and results in an improvement in the quality of university
students’ writing in English.
Since all correlation
coefficients reached levels of statistical significance (
), the research
hypothesis is accepted. It is concluded that the ethical and transparent
adoption of generative artificial intelligence is directly and significantly
related to academic integrity and results in an improvement in the quality of
university students’ writing in English.
This study clearly
presents quantitative data on the relationship between the ethical use of
artificial intelligence, academic integrity, and performance in academic
writing in English (as a foreign language) among college students. By
confirming the proposed hypothesis, it is clearly demonstrated that the ethical
and transparent use of AI does not undermine academic integrity; rather, it
acts as a pedagogical catalyst that improves the linguistic and argumentative
quality of written work in a second language.
Meanwhile, the
descriptive analysis showed that students use AI tools primarily to check
grammar and receive feedback, as well as to verify factual claims, as evidenced
by an overall average score of 3.89 for ethical use.
This is consistent
with the description provided by de Flower and Hayes (1981) regarding the
cognitive process of writing, which is presented as a recursive problem-solving
activity divided into planning, translation, and revision. It is emphasized
that in the context of learning English as a foreign language (EFL), the
translation phase imposes a high cognitive load by requiring the simultaneous
management of grammar, vocabulary, and discourse structure in a second
language.
In this regard, the
positive correlation between ethical use and writing performance supports the
positions of Dahlan (2026) and Md Nawi et al. (2025), who argue that
large-scale language models transform the learning ecosystem by serving as
interactive support not only for stylistic revision but also for structuring
ideas and lexical reformulation.
Furthermore, this
finding aligns with the empirical results of Junaid et al. (2024), Aljasser (2025), and Shabir (2025), who reported
significant improvements in grammatical accuracy, fluency, and coherence when
AI is integrated through collaborative interactions in iterative drafting
cycles, thereby avoiding the mere substitution of critical thinking or
cognitive offloading.
An interesting finding
of the study is the strong positive association between the ethical use of AI
and academic integrity. Students demonstrated a high level of moral
responsibility regarding their own authorship and a clear rejection of direct,
uncredited plagiarism. This behavior is grounded in Ajzen’s Theory of Planned
Behavior (1985, 1991) and in the findings of Huang et al. (2025), who
demonstrated that internalized moral obligations and information literacy exert
a direct protective effect against the intention to commit ethical violations.
However , the
relatively lower scores and higher variability associated with perceptions of
institutional AI policy clarity suggests room for regulation gaps. This lends
support to Lund et al. (2025), who found through quantitative analysis that
awareness of institutional rules and regulations had no predictive ability on
student ethical conduct, concluding that moral judgment and individual
confidence in what does and does not qualify as fraud are what dictate
behavior. In this way, student ethical commitment is derived from values-based
self-regulation and not from punishment- or policy-based regulation.
The findings run contrary to the prohibitionist postures and institutional
“zero-tolerance” policies adopted by many at the beginning of the AI boom.
To quote Cotton et al.
(2023), Dwivedi et al. (2023), Lund et al. (2025), and Nelson et al. (2025) in
part, banning AI entirely is impossible and has been shown to be ineffective in
the modern university context. Literature further confirms invalid notice of
using programs that automatically detect AI language (Li & Sundar, 20;
Zhang et al., 20), as they lack reliability and have been shown to flag writing
from non-native English writers as AI-written text (Gallent-
Torres et al., 20; Liang et al., 2023).
Instead, the findings
support a transition toward a pedagogical model based on Critical AI Literacy
proposed by Dahlan (2026), which guides students through the cycle of prompt
design, comparative evaluation, source verification to avoid “hallucinations,”
and transparent disclosure of use. Similarly, the results suggest the adoption
of process-oriented assessment frameworks, such as the Artificial Intelligence
Assessment Scale (AIAS) by Perkins et al. (2024) and Md Nawi et al. (2025),
which value analytical effort and human authorship over mere inspection of the
final product.
The study had the
following limitations:
As it was a census
sample where N=26 students participated from only one degree program and
university, the results portray an accurate depiction of the group being
analyzed but do students allow for widespread generalizations to other
university settings or fields of study.
Because this study was
cross-sectional, no long-term causal effects can be determined between the
utilization of AI and future developments in cognitive writing development.
Data regarding ethical
usage habits and academic integrity were derived from digital self-report
questionnaires, which could introduce a slight social desirability bias in
participants’ responses.
Conclusions
The research
hypothesis was accepted after it was quantitatively demonstrated that the
ethical and transparent adoption of Artificial Intelligence has a positive,
direct, and statistically significant relationship with both academic integrity
and performance in academic writing in English. Consequently, the
self-regulated use of AI does not undermine students’ honesty but rather acts
as a catalyst that improves second-language proficiency.
Students in the census
sample use AI predominantly as a learning tutor and cognitive scaffold. The
most frequent practices involve grammatical review and stylistic feedback, as
well as factchecking against primary sources, reaching an overall average of 3.89
for ethical use. Therefore, it is confirmed that the tool is used to optimize
the revision process and reduce cognitive load without resorting to passive
text generation.
Thus, the results
clearly demonstrate a high level of moral commitment on the part of students to
the authorship of their writing (M = 4.23) and an explicit rejection of
plagiarism without citing the source (M = 4.31). However, the variability and
uncertainty observed regarding the clarity of university regulations (M = 3.19)
indicate that current ethical behavior stems from internalized moral values
rather than from clearly defined institutional policies or regulatory
frameworks.
Overall performance in
the production of argumentative essays was satisfactory. AI has proven to be an
effective tool that helps overcome barriers resulting from syntactic and
lexical differences, without affecting the tone or the argumentative structure
developed by the student.
The data obtained
indicate that institutions of higher education should move away from
prohibitionist approaches or the use of automated AI detectors that have
documented biases against non-native authors and move toward the implementation
of AI and assessment models focused on the writing process such as the AIAS
scale while institutionalizing policies for the transparent disclosure of the
degree of algorithmic assistance used.
..........................................................................................................
References
Ajzen, I. (1985). From intentions to actions: A
theory of planned behavior. In J. Kuhl & J. Beckmann (Eds.), Action
control: From cognition to behavior (pp. 11–39). Springer.
Ajzen, I. (1991). The theory of planned behavior. Organizational
Behavior and Human Decision Processes, 50(2), 179–211. https://doi.org/10.1016/0749-5978(91)90020-T
Aljasser, A. (2025). Investigating EFL
students’ perceptions of feedback: A comparative study of instructor and
ChatGPT-generated responses in academic writing. Education and Information
Technologies, 30, 22967–22988. https://doi.org/10.1007/s10639-025-13649-8
Aminah, M., & Aly, A. H. (2026). University EFL
students' perceptions of generative AI in academic writing: A qualitative case
study in Indonesia. Al-Ishlah: Jurnal Pendidikan, 18(1), 581–593. https://www.journal.staihubbulwathan.id/index.php/alishlah/article/view/9461
Chicaíza, R. M. C., Castillo, L. A. C., Ghose, G.,
Magayanes, I. E. C., & Fonseca, V. T. G. (2023). Aplicaciones de ChatGPT
como inteligencia artificial para el aprendizaje del idioma inglés: Avances,
desafíos y perspectivas futuras. LATAM Revista Latinoamericana de Ciencias
Sociales y Humanidades, 4(2), 2244–2257. https://doi.org/10.56712/latam.v4i2.756
Cotton, D. R. E., Cotton, P. A., & Shipway, J. R.
(2023). Chatting and cheating: Ensuring academic integrity in the era of
ChatGPT. Innovations in Education and Teaching International, 60(6),
1–12. https://doi.org/10.1080/14703297.2023.2190148
Dahlan, S. P. (2026). Generative artificial
intelligence and critical digital literacy in EFL academic writing: An
integrative review and pedagogical framework. Journal of Language and
Literature Inquiry, 1(1), 43–53. https://journal.privietlab.org/download.php?id=9755
Dwivedi, Y. K., Kshetri,
N., Hughes, L., Slade, E. L., Jeyaraj, A., Kar, A. K., Baabdullah,
A. M., Koohang, A., Raghavan, V., Ahuja, M.,
Albanna, H., Albort-Morant, G., Al-Busaidi, A. S.,
Braley, L., Choi, H. S., Hult, T. M., Jarrar, Y., Ismagilova,
E., Al-Maitah, A., … Wright, R. (2023). So what if
ChatGPT wrote it? Multidisciplinary perspectives on opportunities, challenges
and implications of generative conversational AI for research, practice and
policy. International Journal of Information Management, 71,
Article 102642. https://doi.org/10.1016/j.ijinfomgt.2023.102642
Flower, L., & Hayes, J. R. (1981). A cognitive
process theory of writing. College Composition and Communication, 32(4),
365–387. https://doi.org/10.58680/ccc198115885
Gallent-Torres, C., Zapata-González, A.,
& Ortego-Hernando, J. L. (2023). The impact of Generative Artificial
Intelligence in higher education: A focus on ethics and academic integrity. RELIEVE - Revista Electrónica de Investigación y
Evaluación Educativa, 29(2), Article M5. https://dspace.uib.es/xmlui/handle/11201/163285
Hernández-Sampieri, R., & Mendoza, C. P. (2018). Metodología
de la investigación: Las rutas cuantitativa, cualitativa y mixta. McGraw-Hill.
Huang, C., Alhur, A. A., Azam, M., & Naeem, S. B.
(2025). The challenge of academic integrity in the age of generative
artificial intelligence. Libri, 75(4),
339–354. https://doi.org/10.1515/libri-2025-0024
Junaid, R., Ramadhana, A., Rustan, N. Q., Damayanti,
S. P., & Pateha, M. H. (2024). The role of generative AI in enhancing second language writing fluency.
Indonesian TESOL Journal, 6(2), 222–235. https://ejournal.iainpalopo.ac.id/index.php/ITJ/article/view/9552
Kasneci, E., Sessler, K., Küchemann, S., Bannert, M., Dementieva, D., Fischer, F.,
Gasser, U., Groh, G., Günnemann, S., Hüllermeier,
E., Krusche, S., Kutyniok,
G., Michaeli, T., Nerdel, C., Pfeffer, J., Poquet, O., Sailer, M., Albrecht, A., Roth, C., … Kasneci, G. (2023). ChatGPT for good? On opportunities and
challenges of large language models for education. Learning and Individual
Differences, 103, Article 102274. https://doi.org/10.1016/j.lindif.2023.102274
Liang, W., Yuksekgonul, M.,
Mao, Y., Wu, E., & Zou, J. (2023). GPT detectors are biased against
non-native English writers. Patterns, 4(7), Article 100779. https://doi.org/10.1016/j.patter.2023.100779
Lund, B., Mannuru, N. R.,
Teel, Z. A., Lee, T. H., Ortega, N. J., Simmons, S., & Ward, E. (2025).
Student perceptions of AI-assisted writing and academic integrity: Ethical
concerns, academic misconduct, and use of generative AI in higher education. AI
in Education, 1(1), Article 2. https://doi.org/10.3390/aieduc1010002
Md Nawi, S., Sain, N., Mohd Yusof, S., &
Kamaludin, P. N. H. (2025). Generative AI in academic writing: Exploring ESL
students' strategies and performance. Environment-Behaviour
Proceedings Journal, 10(S130), 55–61. https://doi.org/10.21834/e-bpj.v10iSI30.6879
Nelson, A. S., Santamaría, P. V., Javens, J. S.,
& Ricaurte, M. (2025). Students’ perceptions of generative artificial intelligence (GenAI) use
in academic writing in English as a foreign language. Education Sciences, 15(5), Article 611. https://doi.org/10.3390/educsci15050611
Paniagua Urbáez, Y. L., Troncoso, D., Fragoso García,
P. P., & Merant, N. (2025). La inteligencia artificial en la
enseñanza/aprendizaje del idioma inglés: Una revisión sistemática. AULA
Revista de Humanidades y Ciencias Sociales, 71(2). https://doi.org/10.33413/aulahcs.2025.71i2.427
Perkins, M., Furze, L., Roe, J., & MacVaugh, J. (2024). The Artificial Intelligence
Assessment Scale (AIAS): A framework for ethical integration of Generative AI
in educational assessment. Journal of University Teaching and Learning
Practice, 21(6), Article 06. https://doi.org/10.53761/q3azde36
Rodríguez Fuentes, A., Sancho Noriega, C., Cabrera
Torres, A. A., & Vílchez Delgado, R. M. (2024). Revisión sistemática sobre
la Inteligencia Artificial para el aprendizaje del inglés L2. Portalín,
11, 91–107. https://dialnet.unirioja.es/servlet/articulo?codigo=9717211
Shabir, A. (2025). Generative AI writing tools and academic writing in
higher education: A systematic review of empirical and review studies. Information
Technology Education Journal, 4(3), 543–557. https://doi.org/10.59562/intec.v4i3.10494
Zakaria, Z. A., Massari, N., & Ahmad Mahir, N.
(2026). Ethical engagement with generative artificial intelligence in ESL
academic writing: Learners' experiences from an Islamic higher education
context. Al-Qanatir: International Journal of
Islamic Studies, 35(5), 60–68. https://doi.org/10.64757/alqanatir.2026.3505/1451
Zawacki-Richter, O., Marín, V. I., Bond, M., &
Gouverneur, F. (2019). Systematic review of research on AI applications in
higher education—Where are the educators? International Journal of
Educational Technology in Higher Education, 16, Article 39. https://doi.org/10.1186/s41239-019-0171-0