Review

ChatGPT Integration in Higher Education for Personalized

Learning, Academic Writing, and Coding Tasks:
A Systematic Review
Kaberi Naznin 1 , Abdullah Al Mahmud 2, * , Minh Thu Nguyen 1 and Caslon Chua 1

1 Faculty of Science, Engineering and Technology, Swinburne University of Technology,

Melbourne 3122, Australia
2 Centre for Design Innovation, School of Design and Architecture, Swinburne University of Technology,
Melbourne 3122, Australia
* Correspondence: [email protected]

Abstract: The emergence of ChatGPT in higher education has raised immense discussion
due to its versatility in performing tasks, including coding, personalized learning, human-
like conversations, and information retrieval. Despite the rapidly growing use of ChatGPT,
a dire need still exists for an overarching view regarding its role and implications in educa-
tional settings. Following the PRISMA guidelines, this study represents a systematic review
of 26 articles exploring the use of ChatGPT in academic writing, personalized learning, and
code generation. The relevant literature was identified through electronic databases, includ-
ing Scopus, ACM Digital Library, Education Research Complete, Computers & Applied
Sciences, Web of Science, and IEEE Xplore. Key details from each article were extracted and
synthesized narratively to provide insights into ChatGPT’s efficacy in academic writing,
personalized learning, and coding. The findings indicate that ChatGPT enhances tailored
learning by adapting delivery methods to individual needs, supports academic writing
through error detection and content refinement, and assists in coding by offering clarifica-
tions and reusable code snippets. However, there are concerns over its ethical implications,
including the impact on academic integrity, overreliance by students on AI, and privacy
Academic Editor: Antonio
concerns about data use. Based on these insights, this study proposes recommendations
Sarasa Cabezuelo for the ethical and responsible integration of ChatGPT into higher education, ensuring its
Received: 6 January 2025
utility while maintaining academic integrity. In addition, the results are discussed based
Revised: 1 February 2025 on the relevant learning theories to understand how students engage with, learn through,
Accepted: 4 February 2025 and adapt to AI technologies such as ChatGPT in educational contexts.
Published: 7 February 2025

Citation: Naznin, K.; Al Mahmud, A.; Keywords: ChatGPT; systematic review; PRISMA; personalized learning; academic writing;
Nguyen, M.T.; Chua, C. ChatGPT coding
Integration in Higher Education for
Personalized Learning, Academic
Writing, and Coding Tasks: A
Systematic Review. Computers 2025, 14,
1. Introduction
53. https://doi.org/10.3390/
computers14020053 ChatGPT has become a focal point of discussion in higher education in recent years.
Developed by OpenAI, this advanced language model has attracted attention because
Copyright: © 2025 by the authors.
Licensee MDPI, Basel, Switzerland.
of its diverse capabilities, including proficiency in essay writing, assistance with coding
This article is an open access article tasks, human-like conversation engagement, and seamless information retrieval [1]. These
distributed under the terms and capabilities raise a number of questions concerning the effectiveness of ChatGPT in an
conditions of the Creative Commons academic environment and the ethical consequences of its use. A student’s learning
Attribution (CC BY) license
experience quality impacts their academic performance level and satisfaction with their
(https://creativecommons.org/
educational journey. This will further enhance learning experiences for students by offering
licenses/by/4.0/).

Computers 2025, 14, 53 https://doi.org/10.3390/computers14020053

Computers 2025, 14, 53 2 of 36

them more personalized opportunities to learn through ChatGPT [2] as personalizing

education can improve learning outcomes [3]. This personalized approach may enrich
academic writing by offering tailored feedback and further support coding tasks with
customized help in code generation and solving programming challenges, improving
student learning outcomes in both areas.
Effective writing support is crucial for students to excel both academically and pro-
fessionally. Imran and Almusharraf [4] showed that ChatGPT provides immediate and
personalized assistance in writing tasks and delivers instant feedback to students [4]. With
ChatGPT, personalization allows feedback on everything from individual writing styles to
personal preferences. This may create a deficiency in contextual understanding and raise
questions about the accuracy of the responses this tool provides. This has been highlighted
in studies by Roumeliotis and Tselikas [5]. Moreover, it needs to be added that accidental
plagiarism continues to occur due to the nature of the training on the different datasets.
There can always be more over-reliability, which is the factor inhibiting one’s independent
writing improvement. According to Ray [6], this tool cannot generate original ideas as it is
supposed to do; it can also produce results that may sometimes be inconsistent and may
have trouble with complicated commands.
According to Dwivedi [7], ChatGPT can be used as a coding buddy for students in
generating code snippets. It also provides an avenue for students to get coding assistance at
their own pace, hence enhancing independent learning [8]. Tian’s [9] study raises concerns
regarding the limitations of AI in addressing complex programming challenges. Ethical
considerations include the potential for academic misconduct through the use of machine-
generated code. Attention must be paid to maintaining academic integrity, as indicated by
Azeem and Khan [10].
The reason why academic writing and coding were chosen as distinct types of learning
experiences in this paper is that they relate to different types of cognitive and practical
skills. First, academic writing is essentially a verbal and conceptual operation based on
substantial knowledge of language, structure, and reasoning [11]. Writing professionally
in an academic context has been supported to require a student to apply such cognitive
and metacognitive processes as critical thinking, organization, and reflection. Writing, in
turn, implies linguistic precision and the ability to express complex ideas [12]. On the other
hand, coding is a logical problem-solving ability to change abstract ideas into functional
solutions using programming languages. Learning to code has enhanced problem-solving
skills, whereby the learner breaks down tasks into smaller, manageable pieces and applies
algorithmic thinking [13]. Coding also involves much technical understanding and is,
therefore, iterative in nature, where errors and debugging form a part of learning to do
it correctly [14]. While both are quite different in their approach to learning, academic
writing and coding can shed much light on how these two modes of learning (i.e., one more
language-driven and the other more technical) shape our cognitive development and our
approach to problem-solving.

1.1. Theoretical Foundation: ChatGPT Within the Broader Context of Educational Technology
Personalized learning is an educational approach that tailors learning experiences to
individual students’ needs, preferences, and interests. Instead of a one-size-fits-all curricu-
lum, personalized learning looks to adaptive instruction, flexible pacing, and customized
content to optimize each learner’s path to mastery. According to the theory, learning cen-
tered on the learner’s prior knowledge, goals, and learning styles is the most effective [15].
Key components of personalized learning include using technology to collect student learn-
ing data to provide real-time adjustment, multiple pathways to the learning content, and
cultivating an autonomous and self-directed learner. Research supports that personalized
Computers 2025, 14, 53 3 of 36

learning can increase student engagement, motivation, and achievement by addressing

individual learning gaps and enabling learners to take ownership of their educational
journey. The approach often links back to adaptive learning technologies, which employ
data-driven algorithms that deliver a tailored learning experience in real-time [16].
However, personalized learning is more contentious than this suggests because
the educational community has debated its efficacy and practicability. For example,
Pashler et al. [17] felt that there was limited empirical evidence on the effectiveness of
personalized learning approaches, particularly related to the differential tailoring of instruc-
tion on individual learning styles. Moreover, Coffield et al. [18] analyzed several learning
style models and found that many of these approaches lack a scientific basis, which raises
concerns regarding their widespread usage in educational institutions.
The learning style theory postulates that individuals learn best when a material is
presented in their preferred mode, such as visual, auditory, or kinesthetic. While the
theory is well-practiced in education, it has come under considerable attack due to a lack of
empirical support. For example, Pashler et al. [17] and Coffield et al. [18] did not find any
substantial evidence linking learning styles to improved academic performance. Critics
suggest that learning styles grossly simplify cognitive diversity, and that someone only
learns through one modality that is not taken into context or is situational [19].
Despite these criticisms, the notion of personalized learning has value. Yet, it suggests
that, rather than learning styles being so rigid, a more fluid approach would be far more
advantageous. Education should draw upon an evidence-based approach: Hattie’s [20]
emphasis on visible learning, where feedback and adaptability lie at the core of success, is a
far more valued means forward. This nuanced perspective acknowledges that personalized
learning is useful yet should not be bound to the model of learning styles. Ultimately,
a more evidence-based and adaptive approach to personalization would allow diverse
instructional methods tailored to diverse learner needs without using unsupported theories
of fixed learning styles. ChatGPT may support such an adaptive approach to provide
personalized learning, as identified in this paper.
ChatGPT relates closely to the principles of personalized learning and is, therefore,
an effective tool to improve individualized education. Since it is an AI language model,
ChatGPT can provide personalized interactions with a learner based on his input, knowl-
edge level, and preference. The literature supports that AI technologies in education have
been shown to support personalized learning through real-time, adaptive feedback that
meets the individual needs of learners [21]. Immediate feedback, deeper explanations, and
additional resources help ChatGPT support learners in progressing at their own speed,
giving personalized guidance tailored to their distinctive learning needs. ChatGPT can also
evaluate student responses and guide them through the learning process by asking probing
questions, helping learners develop critical thinking and problem-solving skills—skills
central to personalized learning environments [22]. For example, if a learner struggles
with a particular concept in any subject, ChatGPT can paraphrase the explanations or offer
simpler examples and adjust responses according to the learner’s understanding. It also
aligns with research illustrating how AI dynamically adapts instructional content to meet
the learner’s needs while increasing engagement and deepening comprehension [23]. With
such a broad-ranging remit over subjects and topics, ChatGPT’s support could be flexible
to meet nearly any learning needs in many ways through visual, auditory, or text-based
means. Research has identified that adaptive technologies utilizing artificial intelligence
provide tailored support across various areas of content, thus ensuring that learners gain
swift access to fit-for-purpose education [24].
According to the socio-constructivist theory [25], knowledge is constructed through
interaction with others, and tools like ChatGPT could play a central role in fostering
Computers 2025, 14, 53 4 of 36

dialogue, peer collaboration, and scaffolding learning experiences. Research shows that
ChatGPT conversation engages students in language production and problem-solving,
especially in teaching and learning English [26]. While using ChatGPT, students can actively
negotiate meaning and co-construct knowledge, demonstrating active engagement [27]
and learner autonomy [28] that supports constructivist learning theory.
Similarly, human–technology interaction theory is based on the interaction between
users and technology [29]. It investigates how humans interact with technological tools and
how such interactions affect user experiences and learning outcomes [30,31]. This theory
also considers issues related to user agency, trust, and the potential of AI tools to empower
or limit students in their learning journey. Research shows that ChatGPT can mimic
human empathy and enhance user experience by providing sophisticated conversational
capabilities [32].

1.2. Existing Reviews and Their Limitations

Scholars have conducted several reviews on the use of ChatGPT in education. For
example, Lo [33] researched approximately 50 articles published from December 2022
to February 2023 and summarized how students and teachers benefit from ChatGPT.
Mai et al. [34] conducted a strengths, weaknesses, opportunities, and threats (SWOT) anal-
ysis of 51 articles to understand ChatGPT’s contribution to teaching and learning. Their
review suggests that it supports learning effectiveness despite the inaccuracy of information
and the generation of limited answers. In a SWOT analysis of 12 studies conducted by
Tur [35], the authors determined the utilization of ChatGPT in K-12 education. Tutors can
use the ChatGPT to help students learn better, but there are concerns about the reliability of
such sources [36]. Twelve articles were examined by Montenegro-Rueda et al. [37] for using
ChatGPT in education. Ultimately, it was found that while ChatGPT can improve learning
in the classroom, teachers must first undergo related training before they can assist learners.
Nevertheless, the reviews provided show how using ChatGPT may have advantages
and disadvantages despite the limitations of these studies. For example, the study of
ChatGPT is rapidly evolving, and most existing reviews have focused on a small number
of studies published between 2022 and 2023 [34,38]. Some of the included studies were not
peer-reviewed; for instance, Lo [33]. Montenegro-Rueda et al. included only 12 studies
between May and June 2023. To the best of our knowledge, no published review has
investigated the use of ChatGPT for personalized learning, academic writing, and coding
tasks. Although Zhang and Tur [35] reported in their review that ChatGPT could be
utilized for personalized learning, their review did not elaborate on how it has been
utilized. Another study by Baig and Yadegaridehkordi [39] emphasized that ChatGPT
offers a cost-effective alternative to traditional resources. However, it is still in the early
stages and requires further investigation to understand this tool fully in the context of
higher education. The ChatGPT’s support for writing is especially critical for coding
tasks in higher education. It is now advisable to create a comprehensive summary of the
available studies regarding ChatGPT’s capacity to improve individual learning, scholarly
writing, and programming in higher education institutions. As ChatGPT came into being
in November 2022, no review has directly targeted the above-mentioned areas; hence, this
study aims to address four main research questions.
RQ1: How can ChatGPT be used for personalized learning?
RQ2: How does ChatGPT effectively enhance academic writing skills?
RQ3: How can ChatGPT be used for improved coding tasks?
RQ4: What are the challenges in the integration of ChatGPT in higher education?
Computers 2025, 14, 53 5 of 36

2. Methods
The literature review was conducted considering the Preferred Reporting Items for
Systematic Reviews and Meta-Analyses (PRISMA) guidelines [40].

2.1. Search Strategy and Inclusion Criteria

We searched six electronic databases to ensure that sources were comprehensive and
diverse for our study. These were selected based on their relevance to the research topic,
academic credibility, and coverage of the specific fields of interest. These criteria thus led
us to choose databases that, in their coverage, include a wide variety of peer-reviewed
journals, conference papers, and academic articles. The targeted databases were Scopus, for
extensive multidisciplinary coverage and strong indexing of peer-reviewed literature; ACM
Digital Library, a key resource for research related to computer science and technology;
Education Research Complete, covering a wide range of educational research articles;
Computers and Applied Sciences, applied computer science and technology; Web of
Science, multidisciplinary in scope with high-impact journals; and IEEE Xplore, specialized
in engineering, computer science, and technology research. These databases were selected
to ensure both depth and breadth in the literature search, aligning with the objectives of
this review.
Published articles on ChatGPT integration in higher education, academic writing, per-
sonalized learning, programming, and coding were included. The publication period was
from January 2023 to February 2024. The exclusion criteria included non-English articles
and articles from non-academic sources. Table 1 summarizes the inclusion and exclusion
criteria in selecting the articles for review. In the final search, 551 papers were identified to
enable a comprehensive literature review on ChatGPT integration in higher education.

Table 1. Inclusion and exclusion criteria for article selection.

Criterion Inclusion Exclusion

Addresses the integration of Does not address the integration of
ChatGPT in higher education ChatGPT in higher education
Article topic
contexts, focusing on its role as contexts, focusing on its role as an
an educational resource/tool educational resource/tool
Non-academic articles, such as
Article type Academic articles
articles from mass and social media
1 January 2023 to
Time Articles outside the specified time
1 February 2024
Language English Non-English

2.2. Screening
After the collection of relevant papers, Zotero 6.0.30 was utilized for effective man-
agement. The first step involved eliminating duplicate records, totaling 148 in number.
Next, papers published in non-English languages, those not aligned with the scope of the
review, and those published in non-academic sources or conference papers were excluded,
totaling 120 papers. Only the journal articles were retained during this process. The title
and abstract screening process was subsequently performed on the remaining records
(n = 283). In this phase, the title and abstract of each paper were reviewed independently
by two authors to establish their relevance to the scope of the current review. Exclusion
criteria involved papers that did not meet the objectives of this review or failed to fulfill
set eligibility criteria. This intensive process selected only those studies directly relevant
to the research question and excluded records that focused on technical aspects rather
than educational implications, explored broader AI ethics without specific educational
applications, or had limited relevance to personalized learning approaches (n = 113). The
Computers 2025, 14, x FOR PEER REVIEW 6 of 34

Computers 2025, 14, 53 6 of 36

rather than educational implications, explored broader AI ethics without specific educa-
tional applications, or had limited relevance to personalized learning approaches (n = 113).
The next step was a full-text eligibility screening. During this phase, non-full-text articles
next step was a full-text eligibility screening. During this phase, non-full-text articles
(n = 27) were excluded. A full-text reading of the remaining papers led to the exclusion of
(n = 27) were excluded. A full-text reading of the remaining papers led to the exclusion of
literature reviews and additional records outside the scope of the current review (n = 117).
literature reviews and additional records outside the scope of the current review (n = 117).
This further refinement resulted in a final set of papers (n = 26) (See Figure 1).
This further refinement resulted in a final set of papers (n = 26) (See Figure 1).

Figure 1. The
Figure 1. The PRISMA
PRISMA flow
flow diagram
diagram for
for the
the systematic
systematic review,
review, detailing
detailing the
the database
database searches,
searches, the
the
number of abstracts screened, and the full texts retrieved.
number of abstracts screened, and the full texts retrieved.
2.3. Data Extraction and Analysis
2.3. Data Extraction and Analysis
An extraction sheet was developed to include the key components of each study.
An extraction sheet was developed to include the key components of each study. See
See Supplementary Materials for the extracted data. These components included the
Supplementary Materials for the extracted data. These components included the study’s
study’s objective, concept of ChatGPT, discipline, context, methods, findings, conclusions,
objective, concept of ChatGPT, discipline, context, methods, findings, conclusions, limita-
limitations, and recommendations. Each record was scrutinized for its journal source,
tions, and recommendations. Each record was scrutinized for its journal source, research
research aims, and support for personalized learning, academic writing, coding, and ethical
aims, and support for personalized learning, academic writing, coding, and ethical con-
considerations. Each record was then reviewed by two independent authors on journal
siderations. Each record was then reviewed by two independent authors on journal qual-
quality and source relevance. The reputation of the journal, based on its past reputation and
ity and source relevance. The reputation of the journal, based on its past reputation and
linkage to known academic or professional bodies, was considered. For this reason, only
linkage to known academic or professional bodies, was considered. For this reason, only
credible sources directly relevant to the research found their way into the review. Thematic
credible sources directly relevant to the research found their way into the review. The-
analysis, as outlined by Braun and Clarke [41], was considered to interpret the gathered
matic analysis, as outlined by Braun and Clarke [41], was considered to interpret the gath-
data. This six-step process involves familiarization, coding, generating themes, reviewing
ered data. This six-step process involves familiarization, coding, generating themes, re-
themes, defining and naming themes, and documentation. Initially, we examined the
viewing themes, defining and naming themes, and documentation. Initially, we examined
extracted data to become familiar with its content before assigning codes for categorization.
From each study, we extracted information (e.g., personalized learning, academic writing,
coding, ethical issues, and others) based on our domain of interest, according to Table 2.
 the extracted data to become familiar with its content before assigning codes for categori-
zation. From each study, we extracted information (e.g., personalized learning, academic
Computers 2025, 14, 53 7 of 36
writing, coding, ethical issues, and others) based on our domain of interest, according to
Table 2. These categories were developed based on our aim and research questions of the
study. We then analyzed studies reporting similar issues and grouped them into a cohe-
These categories were developed based on our aim and research questions of the study. We
sive theme. We re-read the themes to be certain they reflected an accurate representation
then analyzed studies reporting similar issues and grouped them into a cohesive theme.
of the data. With the final list, we delineated each theme and gave it a name that was
We re-read the themes to be certain they reflected an accurate representation of the data.
related to our research questions. See Appendix A (Table A1) for the summary of included
With the final list, we delineated each theme and gave it a name that was related to our
studies.
research questions. See Appendix A (Table A1) for the summary of included studies.
Table 2. Our domains of interest with their categories.
Table 2. Our domains of interest with their categories.
Domain Categories/Themes
Domain Categories/Themes
Real-time feedback, adaptive learning, collaborative learning,
Personalized Learning Real-time feedback, and
adaptive learning, collaborative
Personalized Learning student engagement, self-regulatory learning.
learning, student engagement, and self-regulatory learning.
Academic writing process, content generation, writing tools,
Academic Writing Academic writing process, content generation, writing
Academic Writing over-reliance, and critical thinking.
tools, over-reliance, and critical thinking.
Coding
Coding Code
Codegeneration,
generation,code
codedebugging,
debugging, and code
and codeexplanation.
explanation.
Ethical issues
Ethical issues Plagiarism, biases, and academic misconduct.
Plagiarism, biases, and academic misconduct.
Others
Others Privacy, accuracy,
Privacy, accuracy,and
andrecommendations
recommendations

3. Results
3. Results
3.1. General
3.1. General Description
Description ofof the
the Included
Included Studies
Studies
The bar
The bar chart
chart (Figure
(Figure 2)2) illustrates
illustrates the
the included
included studies’
studies’ coverage
coverage of of different
different higher
higher
education disciplines
education disciplines inin which
which thethe ChatGPT
ChatGPT waswas used.
used. Among
Among these
these disciplines,
disciplines, general
general
higher education is the most prominent, and Computer Science and
higher education is the most prominent, and Computer Science and Education are Education are the
the
most frequent,
most frequent,considering
consideringa aspecific
specificstudy
study area.
area. SixSix papers
papers closely
closely trailtrail
the the domains
domains of
of the
the Humanities and Social Sciences, and the fields of business and social sciences,
Humanities and Social Sciences, and the fields of business and social sciences, medicine, medi-
cine,interdisciplinary
and and interdisciplinary
sciencescience are represented
are represented by onebypaper
one paper
each. each.

Figure 2. Disciplinary coverage of the included studies around the use of ChatGPT in higher education.
Figure 2. Disciplinary coverage of the included studies around the use of ChatGPT in higher edu-
cation.
The pie chart (Figure 3) illustrates the distribution of the studied papers across various
countries and continents. In terms of continents, Asia appears to have the highest frequency
The pie chart (Figure 3) illustrates the distribution of the studied papers across vari-
of papers (45.2%), followed by Europe (31.0%) and Oceania (11.9%).
ous countries and continents. In terms of continents, Asia appears to have the highest fre-
Research methods such as quantitative (e.g., surveys) (n = 11), qualitative (e.g., ob-
quency of papers (45.2%), followed by Europe (31.0%) and Oceania (11.9%).
servations and interviews) (n = 12) studies, and mixed methods (n = 3) were used in the
included studies. The key aim of most of the studies (n = 11) was to understand ChatGPT’s
applications across various fields. On the other hand, a significant chunk of studies (n = 5)
also focused on the effectiveness of ChatGPT, majorly in essay writing and programming.
The interest in researching the issues and limitations in using ChatGPT, above all in view of
academic integrity and biases, was high (n = 4). Comparison with other tools or methods,
the effect of ChatGPT on teaching and learning and student perception was discussed, but
lower in frequency (n = 3).
Computers
Computers 2025,
2025, 14,
14, x53FOR PEER REVIEW 8 of 34
8 of 36

Figure 3. Countries/locations of the included records (N = 26).

Figure 3. Countries/locations of the included records (N = 26).

Research
3.2. RQ1: Howmethods
Can ChatGPTsuch Be
as Utilized
quantitative (e.g., surveys)
for Personalized (n = 11), qualitative (e.g., ob-
Learning?
servations and interviews) (n = 12) studies, and mixed methods
The included studies pointed out that ChatGPT can provide feedback (n = 3) were usedofinneed,
at times the
included
facilitate studies.
learningThe key aim
changes overofthemost of the
course ofstudies (n = 11)
education, andwas to understand
create ChatGPT’s
conditions for students
applications across various fields. On the other hand, a significant chunk
to work together. In addition, ChatGPT has been shown to increase learner engagement of studies (n = 5)
also focused
through on the effectiveness
interactive of ChatGPT,
platforms, which majorlyself-regulation.
also supports in essay writingThe
andtool
programming.
encourages
The interest in researching the issues and limitations
students to monitor their progress and make adjustments. in using ChatGPT, above all in view
of academic
Real-timeintegrity andrefers
feedback biases,towas high (n either
responses, = 4). Comparison with otherortools
for action, question, or meth-
performance,
ods, the effect of ChatGPT on teaching and learning and student
provided for immediate and continued action. That would imply time for adjustment, perception was dis-
cussed, but lower
improvement, in frequency
constant learning,(nand
= 3).development of skills [42]. Of the 26 studies analyzed,
ten focused on exploring ChatGPT as a personal tutor and, more specifically, on the function
3.2. RQ1: How Can ChatGPT Be Utilized for Personalized Learning?
of its real-time feedback. Several studies (n = 7) have emphasized ChatGPT’s role in enhanc-
The includedlearning.
ing personalized studies pointed
Wang etout that ChatGPT
al. (2021) noted not can provide
only feedback
real-time feedback at times
but also of
need, facilitate
goal setting andlearning
progresschanges
monitoring. overSeveral
the courseotherofresearchers
education,have and noted
createsimilar
conditions for
benefits,
students to work together. In addition, ChatGPT has been shown
such as quick and effective responses [43], effective personalized support across learning to increase learner en-
gagement
tasks [44],through
on-demand interactive
support platforms,
[45], and which
real-time alsowriting
supports self-regulation.
guidance [46]. Rasul Theettool en-
al. [47]
courages students to monitor their progress and make adjustments.
state that ChatGPT’s just-in-time feedback provision fosters understanding and access to
Real-time
resources. feedback
Sullivan et al.refers to responses,
[48] highlighted its either
value in forproviding
action, question, or performance,
plain language pro-
explanations,
vided for immediate and continued action. That would imply
organizational suggestions, grammatical feedback, and quiz-question development. time for adjustment, improve-
ment,However,
constant learning, and development
the limitations of real-timeoffeedback
skills [42].systems
Of the 26 studies
should be analyzed,
understood tenandfo-
cused on exploring ChatGPT as a personal tutor and, more specifically,
acknowledged. Seven of the ten articles studying real-time learning acknowledged that on the function of its
real-time
ChatGPTfeedback.
sometimes Several
gavestudies
incorrect (n =answers.
7) have emphasized
Ellis and SladeChatGPT’s
[49] in role
theirinstudy
enhancing
observedper-
sonalized learning. Wang et al. (2021) noted not only real-time
its inability to adjust instructions that were not in text format, and occasional delaysfeedback but also goal setting
and progress
in its responses.monitoring. Severalthe
Furthermore, other researchers
efficacy have noted
of ChatGPT similar benefits,
is contingent on the such as quick
complexity
and effective responses [43], effective personalized support
of the problem at hand [49,50]. It points to a more profound problem in educational across learning tasks [44], on-de-
mand supporthow
technology: [45],ChatGPT
and real-timeresponds writing guidance
poorly [46]. Rasul especially
to complexity, et al. [47] state
within thatdiverse
ChatGPT’s and
just-in-time feedback provision fosters understanding and access
changing learning contexts. Such an educational tool, like ChatGPT, would fail to respond to resources. Sullivan et al.
[48] highlighted
appropriately toitsthevalue in providing
complexities of theplain
needslanguage
of learners,explanations,
particularly organizational
those requiring sugges-
more
tions, grammatical feedback, and quiz-question development.
nuanced approaches. While AI tools provide useful features, they also have limitations
that However,
may affectthe limitations
learning outcomes.of real-time feedback
Pedagogical systems should
seriousness designatesbe understood
the importance and
acknowledged. Seven of the ten articles studying real-time
of accurately understanding and conveying knowledge in an educational setting. The learning acknowledged that
ChatGPT
mistakes of sometimes
AI or the gave incorrect
inability to handleanswers.
complex Ellisorand Slade information
nuanced [49] in their undermine
study observed such
its
learning, mainly when students rely on AI’s output without fully mastering concepts. its
inability to adjust instructions that were not in text format, and occasional delays in In
responses.
other words, Furthermore,
overlooking the efficacy
these of ChatGPT
weaknesses in AIisapplications
contingent on the compromising
risks complexity of the the
problem at hand [49,50]. It points to a more profound problem in educational technology:
Computers 2025, 14, 53 9 of 36

pedagogical value of these tools, which may go so far as to accidentally mislead students
or fail to engage them in critical thinking about the content. Therefore, although GPT tools
assist greatly, limitations in their functionalities should not be avoided, particularly in
learning situations that are quite complex or subtle.
Adaptive learning is an educational approach in which instructional materials, re-
sources, and activities are tailored to the needs of each learner [51]. However, as seen with
ChatGPT, the inability to adjust to complex learning needs—such as non-text formats or
varied cognitive styles—can present significant limitations. Out of the 26 reviewed papers,
15 emphasized adaptive learning. However, again, in that context, integration with Chat-
GPT depends on either the complexity of the task or the learner’s familiarity with that tool.
In various educational systems, such biases and limitations may be further exacerbated,
for example, by difficulties faced by students who are hands-on or experiential learners,
leading to inequalities whereby some learners may not be able to engage effectively with
ChatGPT [52]. Therefore, in an educational context where complexity or learner diversity is
not addressed, AI tools like ChatGPT may inadvertently reinforce existing biases, making
it harder for all students to benefit equally.
The NLP in ChatGPT encourages student independence and makes their learning
experience active and engaging [53]. When used as a co-teacher virtual assistant, Chat-
GPT provides individualized support to make learning more engaging [54]. Kiryakova
and Angelova [55] and Qureshi [56] found it to be an intelligent tutoring system with
potential benefits for creating customized instruction. Natural language processing models,
including ChatGPT, can be used to construct highly individualized learning plans based
on performance and feedback [45]. The integration of ChatGPT appears to enhance edu-
cational support and sustainably create positive learning experiences [50]. ChatGPT can
be beneficial to students in such a way that proficient users become better at using this
tool effectively through critical engagement with materials [57]. In addition, once learners
grasp how to properly use ChatGPT, it becomes more useful and effective in augmenting
their learning experiences [57].
Applying the learning phases based on the 5Es model (Engagement, Exploration,
Explanation, Elaboration, and Evaluation), researchers showed that ChatGPT provides
personalized assistance in each phase, which increases learner commitment. Through
instant access to data, scholars’ enthusiasm and self-belief are boosted in ways that make
them more committed to their work [46]. Moreover, the use of the tool is believed to create
more enjoyment and enthusiasm, as shown by the positive perceptions of ChatGPT and
personalized coaching [58]. According to Fuchs [45], the support provided by ChatGPT
upon request contributes a great deal to the learning among students, especially those
who study online. This enables teachers to concentrate on upper-level abilities as well
as mentorship aided by AI systems, such as ChatGPT. The shift from educators’ roles
can result in increased satisfaction among learners [59]. In a way that enhances learning
experiences for students who study together in interactive learning environments, ChatGPT
fosters teamwork and the verification of solutions during problem-solving processes [50].
However, there are concerns that learners may rely solely on interactions with the ChatGPT.
In this scenario, collaborative learning and critical thinking can be hindered [47]. There-
fore, ChatGPT’s promotion of robust student engagement may vary depending on the
instructional context and implementation strategies [43].
Self-regulatory learning is a process whereby students set goals, monitor progress,
and adjust strategies to manage their learning [60]. Of the 26 papers analyzed, 11 discussed
self-regulatory learning. ChatGPT plays an important role in self-regulated learning by
responding to different learning styles and preferences [53]. Personalized learning mate-
rials enhance ownership and agency for learners in the learning process [54]. ChatGPT’s
Computers 2025, 14, 53 10 of 36

judgment-free environment serves as a private space for learners to seek clarification, which
alleviates classroom anxiety [46]. Moreover, personalized tutoring by ChatGPT brought
more satisfaction, engagement, and self-confidence. This underlines the importance of
subjective factors for effective learning interventions [58]. ChatGPT provides respectful
and motivating comments and supports engaging with the content better [57]. This type of
interaction is particularly beneficial for students with prior knowledge, who are developing
critical thinking skills and can enhance their learning efficiency [57].
Despite Esmaeil et al.’s [46] finding that ChatGPT can reduce classroom anxiety by
providing a judgment-free zone to clarify doubts, Stojanow [57] found that ChatGPT may
not fully support students in achieving their learning goals independently. Furthermore,
although Yilmaz and Yilmaz [43] indicated that incorporating ChatGPT in programming
tasks may enhance self-confidence and motivation, Zou and Huang [44] found that it may
not adequately promote self-regulated learning strategies. See Table 3 for the summary of
ChatGPT’s support of real-time feedback, adaptive learning, and self-regulatory learning.

Table 3. Summary of ChatGPT’s support of real-time feedback, adaptive learning, and self-regulatory
learning based on 15 studies.

Aspect of Learning Summary of Support from Literature Limitations and Considerations

Provides immediate guidance, corrections,
Occasionally generates misinformation and
and explanations, enhancing continuous
incorrect answers (Various sources).
learning [42].
Real-time Feedback Offers personalized feedback, goal-setting Unable to accommodate non-text prompts;
assistance, and progress monitoring [2]. occasionally has slow response times [49].
Facilitates on-demand support and
Efficacy is dependent on task complexity [50].
guidance in various tasks [45,46].
Tailors learning paths, providing
Efficacy in offering customized instructions is
personalized guidance and support
uncertain [56].
(Various sources).
Adaptive Learning Acts as a virtual co-teacher, offering
It is not suitable for students who prefer
customized instructions and enhancing
hands-on, experiential learning [52].
interactivity [55].
Constructs individualized curricula and
generate customized learning plans [58].
Empowers learners, fosters ownership, and Potential over-reliance on ChatGPT could
increases motivation through personalized hinder collaborative learning and critical
learning [53]. thinking [47].
Student Engagement Varies in effectiveness depending on
Sequential progression guided by the 5Es
instructional context and implementation
model enhances engagement [54].
strategies [43].
Immediate access to information boosts
motivation and confidence [46].
May not fully support students in
Facilitates diverse learning styles and
independently setting and achieving learning
preferences [53].
goals [57].
Self-regulatory Incorporation in programming tasks may not
Provides a judgment-free zone for clarifying
Learning adequately promote self-regulated learning
doubts and reducing anxieties [46].
strategies [44].
Promotes active learning and critical
engagement through respectful and
encouraging feedback [57].

These studies have a number of recommendations that can ensure the effective usage
of ChatGPT for personalized learning: training educators and providing them with clear
Computers 2025, 14, 53 11 of 36

guidelines on how to integrate ChatGPT into their practice is necessary [45,53]. Such
training may encourage responsible and transparent use of the tool. Besides, digital compe-
tencies at both the student and faculty levels need to be developed for the full exploitation
of AI tools in educational settings [55]. Knowledge empowerment of stakeholders in the use
of such a tool as ChatGPT is the key to building an enabling environment for personalized
learning. Future research should explore collaborative learning environments that leverage
ChatGPT’s capabilities to collectively solve complex problems collectively [53]. Another
important avenue of research is the investigation of the long-term impact of ChatGPT on
learning outcomes beyond the classroom [53].

3.3. RQ2: How Can ChatGPT Effectively Enhance Academic Writing Skills?
Nine studies discussed ChatGPT’s usage across a full academic writing process and
described effective content produced within instances of this AI performing a standard
tool’s function. In contrast, however, they express several concerns since overuse will
decrease motivation toward developing skills on their own.
All nine studies that explored the role of ChatGPT in writing addressed its ability to
generate content. ChatGPT provides invaluable assistance in idea generation and sum-
marization. It offers students a platform for brainstorming and quickly generating texts
that would otherwise require substantial time and effort [46]. Zou and Huang [44] further
suggested that, in addition to brainstorming, ChatGPT enhances content creation through
the provision of individual feedback to students in developing their written drafts. The
effectiveness of ChatGPT in content generation also varies across different contexts. For
instance, it has been observed that it generates adequate scripts for modules at the under-
graduate level, while differences in quality vary with formats and topics of questions [61].
Additionally, as McMurtrie [62] suggests, ChatGPT’s integration into everyday writing
seems inevitable, akin to incorporating calculators and computers into mathematics and
science. While this use of ChatGPT for writing is undoubtedly an increasingly common and
continuing phenomenon, it must be placed within the perspective that ChatGPT is not the
only AI model available. Various companies are integrating other GPT models and many
other language models into their products. For instance, Microsoft has already integrated
GPT models into its products, such as Microsoft Word and Excel, to show that this broader
trend of integrating AI into everyday tasks goes beyond just ChatGPT. ChatGPT is only a
small part of a bigger trend in which many AI technologies, including scores of variants of
GPT, are finding their way into all sorts of products.
Contrary to expectations, there is no evidence that this would speed up writing. The
essays written with ChatGPT are not faster than conventionally written essays [63]. Apart
from text stimuli, responses by ChatGPT cannot be initiated with pictures, for instance [49].
It has still been an open question whether such generated text material is credible at all.
This is a point of great importance that the student will learn to check and give sources in
general so that the content is valid [55].
Of the 26 papers analyzed, 5 discussed the capability of ChatGPT as a writing tool. The
advantages pointed out include how ChatGPT saves time in aggregating and summarizing
information and in paraphrasing [64]. Furthermore, ChatGPT is a good reference for
grammatical feedback; it can serve as a quasi-translator for non-native English-speaking
people, especially in very complex terms that may present problems [48]. Meanwhile,
several aspects of ChatGPT were found useful across the writing process: brainstorming,
personalized feedback, translation, and draft enhancements [44]. However, ChatGPT
struggles to consistently generate high-quality content, especially for non-native English
speakers [48].
Computers 2025, 14, 53 12 of 36

Based on the recommendations, this study suggests various ways in which ChatGPT
can be used to enhance writing skills. First, users need to understand how their prompts
can affect the quality of the responses they get from ChatGPT [63]. The nature and depth of
students’ prompts present a good starting point for understanding how prompts can be
optimized [63]. For example, instructors can look at students’ prompts to get an idea of the
type of questions they are asking and at what level of specificity. Instructors can then adjust
their teaching to better position students to avoid the knowledge gaps and get higher value
responses with ChatGPT. ChatGPT is recommended for collaborative learning because
it provides a context for developing writing skills in teamwork and problem-solving
work [53]. For example, students can work together to generate and refine written content
while leveraging the capabilities of ChatGPT, thus enhancing their collaboration skills. For
the proper integration of this tool into academic writing, instructors also need training in
how to do so effectively with ChatGPT, and educators should receive training to integrate
this tool into their teaching [45,53]. Addressing the ethical considerations associated with
ChatGPT use is also essential [45,53]. This involves educating students about the limitations
and potential biases [49] and emphasizing critical thinking alongside ChatGPT usage [50].
See Table 4 for the summary of ChatGPT’s support for writing-related activities.

Table 4. A summary of ChatGPT’s support of content generation process, grammar and clarity, idea
generation, and writing tool based on the literature.

Academic Writing Skills

Summary of Support from Literature Limitations and Considerations
Although ChatGPT proves useful in content
generation, it does not necessarily enhance
ChatGPT supports students in their writing
or shorten writing time. Its impact on
process by providing personalized feedback
writing efficiency varies depending on
Content generation and improving drafts [44]
individual writing styles and
process preferences [63].
ChatGPT provides students with a platform for
The credibility of ChatGPT-generated text
brainstorming and quickly generating written
remains an open question [55].
content [46].
ChatGPT lacks the expertise to provide
detailed guidance on subject-specific
ChatGPT effectively identifies and rectifies
terminology, non-text prompts, non-English
Grammar and grammatical errors, improving clarity,
Clarity content, citation styles, or content relevance,
coherence, and overall quality of written
particularly in specialized fields requiring
content [48].
precise terminology and logical
reasoning [48,55].
ChatGPT may propagate biases present in
ChatGPT proves an invaluable platform for
Idea Generation exploring different perspectives and enriching its training data and provide not up to date
their writing [46,64]. information due to its reliance on
pre-existing text data [53].
There is a risk of students becoming overly
dependent on ChatGPT, potentially
ChatGPT has proven to be useful in assisting hindering the development of critical
various aspects of the writing process, from thinking, creativity, and independent
brainstorming ideas, and personalized writing skills [63].Additionally, the risk of
Writing tool feedback to translating language items, plagiarism and academic dishonesty is
paraphrasing and language improvement [48]. increased if students use
ChatGPT-generated content without proper
citation [53].
ChatGPT streamlines the process of ChatGPT struggles to consistently generate
aggregating, summarizing, and paraphrasing high-quality content, especially for
information, saving valuable time [64]. non-native English speakers [64].
Computers 2025, 14, 53 13 of 36

3.4. RQ3: How Can ChatGPT Be Used for Coding Tasks?

Seven of the included studies highlighted ChatGPT’s ability to aid students with
code generation and explanation, particularly at the introductory level. However, the
accuracy of ChatGPT in this regard remains contingent upon further advancements. Its
efficacy for code debugging is limited. Seven of the analyzed papers discussed ChatGPT’s
capabilities in generating code. The utility of ChatGPT extends across a spectrum of
programming tasks, from code completion and correction to document generation and
chatbot development [65]. ChatGPT shows potential as a comprehensive tool for assisting
programmers in software development. Coello et al. (2024) delved deeper into the dynamics
of code generation by comparing GPT-based models with non-GPT-based approaches [66].
Their findings highlight the tendency of GPT-based models to produce shorter, more concise
codes [66]. Despite this advantage, challenges such as compatibility issues and the need for
human oversight persist. AI models should play a complementary role along with human
developers [66].
The effectiveness of ChatGPT in generating executable code varies among program-
ming languages. While it excels in languages like Julia, in other languages such as C++,
it performs considerably worse [67]. This leads to inconsistent results and raises ethical
issues based on the choice of language [67]. Moreover, if students abuse the tool and use it
without acquiring the concepts themselves, academic integrity becomes a problem [56,59].
Sridhara et al. [68] highlighted ChatGPT’s proficiency in natural language processing tasks
related to data analysis but noted limitations in feature engineering and accuracy.
Of the 26 papers analyzed, 5 discussed the explaining code capability of ChatGPT.
These studies showed the adaptability of ChatGPT in generating documentation for pro-
gramming tasks [65]. It can automatically generate explanations available for functions,
classes, variables, and API references, thus helping in their comprehension [65]. A clear
explanation of code and programming concepts with the help of ChatGPT reinforces its
role as a learning aid for novice students [43]. ChatGPT works well for method name
suggestion, log summarization, and commit message generation [68]. It can generate code
snippets adeptly and provide clear explanations, which will be particularly helpful for
novice learners working to grasp the basics of programming [66,67].
However, there were instances where the model did not follow some instructions.
For example, it would give comments in the code generated even when instructed not
to comment [67]. Other problems arise when trying to convert the model-generated
explanations into codes that can be executed [56]. Coello et al. [66] stressed that human
feedback is necessary for fine-tuning the performance of the model. While ChatGPT can
indeed make a much-needed boost for better coding practices, it works best with, not
instead, human developers themselves [66].
Out of the 26 papers, 3 were on the capability of ChatGPT to debug code. The ChatGPT
chatbot indeed helps in the real-time detection of syntax errors and provides ways of cor-
recting some common programming errors. Thus, it can improve the accuracy and quality
of code [65]. However, the suggested solutions sometimes do not compile successfully
in integrated development environments. This frustrates users [56]. Moreover, when the
capabilities of ChatGPT were tested in complex debugging tasks, its effectiveness was
limited. ChatGPT may provide generic debugging suggestions for complex issues. Addi-
tional difficulties arose in tasks such as duplicate bug report detection due to issues such as
normalized identifier names [68]. See Table 5 for a summary of ChatGPT’s capabilities in
code generation, explanation, and debugging.
Computers 2025, 14, 53 14 of 36

Table 5. Summary of ChatGPT’s capabilities in code generation, explanation, and debugging based
on the literature.

Aspect of Code Summary of Support from Literature Limitations and Considerations

Support
Effectiveness varies across programming
Aids in code completion, correction, documentation,
languages, with inconsistent outcomes and
and chatbot development [65].
ethical considerations [67].
It demonstrates proficiency in generating code
Code Generation solutions for common coding problems, providing
Risk of academic dishonesty if used without
valuable examples for learners, and assisting in code
understanding underlying concepts [56,59].
completion, correction, documentation, and chatbot
development [65,69].
Tends to produce shorter, more concise code than Limited in feature engineering and
non-GPT-based approaches [66]. accuracy [68].
Sometimes, it deviates from provided
Generates comprehensive documentation for
instructions and has difficulty translating
programming tasks, aids understanding of code
generated explanations into executable
Code Explanation elements, and simplifies complex codebases [65].
code [56,67].
Proficient in specific tasks such as method name Human feedback and collaboration enhance
suggestion and log summarization [68]. effectiveness [66].
Provides real-time detection of syntax errors and Suggestions may fail to compile in integrated
suggests solutions for common mistakes [65]. development environments [56].
Code Debugging It may offer generic debugging suggestions for Faces challenges in tasks like duplicate bug
complex issues, with effective constraints for more report detection due to issues such as
intricate tasks [70]. normalized identifier names [68].

3.5. RQ4: Challenges in the Integration of ChatGPT in Higher Education

We identified several challenges and barriers to the integration of the ChatGPT in
education. These include concerns regarding accuracy, privacy, and plagiarism. There
are concerns about the tool’s impact on critical thinking skills and that its responses can
be influenced by biases or by the quality of training data. There is also the potential
for privacy risks associated with misuse of personal data and the potential to facilitate
academic misconduct, such as plagiarism.
Of the 26 papers analyzed, 15 discussed the impact of the ChatGPT on students’ critical
thinking. This topic is of significant concern in educational research [46]. Some perspectives
suggest that ChatGPT can stimulate critical thinking and creativity [47] because crafting
effective prompts for ChatGPT requires a deep understanding of the tool and underly-
ing content. Therefore, Crafting prompts demand students’ critical thinking skills [56].
Rasul et al. [47] suggested that educators can promote discovery learning (i.e., through
play and investigation, students come up with new concepts) to enhance problem-solving
skills so that students’ use of ChatGPT involves deeper reflection. While ChatGPT offers
unparalleled ease and convenience in generating responses, there is growing apprehension
that its use may promote intellectual laziness among students [45]. Becoming reliant on
suggestions could hinder students’ development of critical thinking and problem-solving
abilities [55]. Students could fall into passiveness, whereby they might receive what-
ever ChatGPT provides without ever assessing either the veracity or relevance of such
responses [45]. However, its impact on critical thinking still requires investigation [47].
Of the 26 papers analyzed, 20 discussed the accuracy of the ChatGPT responses. Sev-
eral studies (n = 9) have pointed out instances where ChatGPT responses are inaccurate
or inconsistent [49,53,54,57,63,67,71]. Many inaccuracies seem to arise from model biases
or variations in the quality of the training data [45,49,53]. Privacy concerns and copyright
issues also affect the accuracy of ChatGPT responses [47,53]. Accuracy may also vary
depending on the complexity of the task and the programming language involved [66,67].
Although ChatGPT can generate code for a broad range of tasks, its performance is incon-
Computers 2025, 14, 53 15 of 36

sistent, resulting in syntactically correct code in some instances and bugged or no code in
others [67].
Of the 26 analyzed papers, 4 discussed privacy issues related to the ChatGPT. The use
of ChatGPT in educational settings has raised significant privacy concerns [53]. A primary
concern is that personal information can be used for purposes beyond personalized learning,
such as targeted advertising or other commercial activities [53]. Moreover, data breaches or
disclosure of sensitive information cannot be completely eliminated [53]. Recent incidents
have increased the AI technology’s privacy vulnerabilities, including in ChatGPT, where
leaks took place to raise the alarm about the protection of user information [71]. In this
regard, ChatGPT often operates without explicit user consent [71]. ChatGPT may also raise
copyright concerns, as the model may have been trained on content protected by copyright
and may have provided responses similar to that content [47].
A total of 5 of the 26 analyzed papers discussed bias in the ChatGPT responses.
Biased training data may lead to ChatGPT perpetuating inaccuracies, thus hindering
genuine learning [45,47]. Without proper verification, students may internalize misleading
content [55]. There is much concern that biased information could reinforce negative
stereotypes and harm learners [57]. Efforts have been made to address biases, such as
implementing plugins for updated data, but this issue currently persists [47].
Educators are at the forefront of developing habits of using ChatGPT among students.
Various studies recommend the design of training programs that could assist instructors in
effectively integrating ChatGPT into teaching. Such training should focus on the supple-
mentary nature of ChatGPT and the need to encourage critical thinking [45,50]. To avoid
overdependence on ChatGPT, educational institutions should focus on active learning
experiences that foster critical thinking, problem-solving, and independent inquiry [49,59].
Institutions should establish clear policies and ethical guidelines for using AI models like
ChatGPT to protect student privacy and academic integrity [45,71]. To incentivize student
engagement and discourage passive reliance on AI-generated content, assessments should
measure the application of concepts and not rote memorization [56]. Assessment strategies
must evaluate learning processes and foster collaborative learning while safeguarding
against plagiarism [47]. Educational initiatives should prioritize students’ development
of AI literacy, enabling them to navigate ethical considerations and potential biases in
AI-generated outputs [43].

4. Discussion
This study explored the potential of integrating ChatGPT into education to enhance
learning experiences, academic writing, and coding tasks. The key findings are discussed
in the following sections, and recommendations are provided for the successful, ethical,
and responsible integration of ChatGPT in higher education.
Educators increasingly use ChatGPT to personalize learning by tailoring content to
individual needs and fostering engagement and self-regulation of learning processes [72].
As such, ChatGPT provides targeted explanations and activities according to students’
proficiency and learning style. On the other hand, challenges such as response accuracy and
reliability need consideration; uncritical acceptance of AI suggestions can harm cognitive
development in learners [49,53]. Ethical use guidelines and accuracy measures should be
integrated into the curriculum to ensure responsible use of AI while enhancing personalized
learning and supporting student success [73].
ChatGPT has the potential to help improve academic writing by providing brainstorm-
ing and suggesting alternatives; however, it also brings a number of challenges, such as
over-dependence on the tool, which reduces critical thinking and independent writing [36].
This should be used not as a solution but as a supporting tool for writing improvement.
Computers 2025, 14, 53 16 of 36

Educators should guide students in responsibly using ChatGPT, emphasizing information

verification, critical thinking, and ethics in AI use. This includes integrating AI content
with the students’ styles and voices in their writing, ensuring a culture of integrity to avoid
overreliance, and enhancing the educational value of the tool [63].
The integration of ChatGPT into coding tasks is limited; it may help explain code and
solve frequent problems mainly, as stated in [43,69]. While ChatGPT can develop code
snippets and explain concepts in programming, critical thinking, and problem-solving
need to remain central. This encourages students to review and verify the code gener-
ated by ChatGPT for better understanding and peer learning. The educator can guide
students in workshops or labs on using AI technology responsibly and interpreting the
responses correctly.

4.1. Recommendations for the Integration of ChatGPT in Higher Education

Our findings on the integration of ChatGPT into higher education are replete with
opportunities and challenges. First, there is the implication of how the instructors contem-
plate the continually developing role that AI technologies, such as ChatGPT, take within
a classroom environment. While ChatGPT is able to support teaching with its ability to
provide fast feedback and individual attention, limitations related to its level of accuracy
and depth create great pedagogical concerns. Educators have to find a delicate balance
between using the capabilities of ChatGPT and making students develop critical thinking
skills and domain-specific knowledge. Another teaching implication in higher education
could be that traditional methods of instruction will need to be revisited due to the growing
demand to integrate with AI. As AI technologies in education continue to increase, so too
must strategies to have students interact with and critically evaluate content generated
by AI. This includes the integration of overt instruction in information literacy, digital
ethics, and critical thinking within the curriculum. Moreover, the results highlight the
need for a collaborative culture and an interdisciplinary dialogue between educators, AI
developers, and domain experts. Working together allows stakeholders to co-create inno-
vative approaches that harness the power of AI technology while guarding against bias,
misinformation, and over-reliance.
While ChatGPT holds great potential for personalized learning, educators must ensure
that such a utility does not undermine student autonomy. Students need to be encouraged
to critically use AI-generated content. Teachers should encourage active learning techniques
for self-regulation, critical analysis, and problem-solving rather than passively relying on
ChatGPT for an answer. According to Hattie [20], students have to develop higher-order
thinking skills, which can only be achieved if students are engaged in active learning.
Therefore, educators should integrate ChatGPT in a manner that would spur curiosity
and independent thinking, yet also give guidance on how to verify information, question
responses, and integrate AI-generated content with their ideas.
The integration of ChatGPT into the classroom requires teachers to shift from tradi-
tional providers of knowledge to facilitators of learning. As AI increasingly supports the
process of information retrieval and content creation, educators’ roles will shift toward fos-
tering critical thinking, ethical decision-making, and collaborative learning. Teachers must
guide students in evaluating and interpreting the vast amounts of information provided
by AI [74]. This shift requires a professional development process for teachers to equip
them with the skillsets needed to responsibly integrate AI into teaching practices. Teachers
should, in turn, work on training them to use AI to support learning processes while not
forging academic creations. We summarized recommendations from the studies that are
listed in Table 6 below, along with the implications of our findings.
Computers 2025, 14, 53 17 of 36

Table 6. A summary of the recommendations for institutions, educators, and students.

For Institutions For Educators For Students

Formulate wide-ranging policies and Recognize the limitations of ChatGPT
Establish clear guidelines on when and
directives on ethical utilization of and not rely solely on its
ChatGPT [45,48,52,53,59]. how ChatGPT should be used [45,52].
responses [43,44,49,52].
Offer faculty training programs on
Design assessments that focus on critical Cross-check information obtained from
ChatGPT’s capabilities and ChatGPT with reliable sources [43,44,52].
thinking skills [52,59].
limitations [50,55].
Foster transparency in ChatGPT usage Educate students on ChatGPT’s Evaluate ChatGPT suggestions critically,
and encourage reporting of concerns [52]. capabilities and limitations [44,50,52]. considering context and relevance [43,44].
Conduct regular evaluations of Incorporate ChatGPT insights into the
Prompt students to reflect on ChatGPT
ChatGPT’s impact on teaching and learning process alongside other
interactions [52].
learning [50,59]. resources [52].
Leverage ChatGPT’s adaptive learning Seek clarification from instructors or
Address ethical and legal considerations
features for personalized consult additional sources if unsure
related to ChatGPT usage [45,52]. about ChatGPT responses [50].
feedback [48,59].
Encourage collaboration among Educate students on privacy and data Avoid plagiarism by adequately citing
educators and departments for effective security when interacting with ChatGPT-generated content in academic
ChatGPT integration [52,59]. ChatGPT [45,50,52,63]. writing [71].
Stay informed about advancements in AI Continuously review and update Seek guidance from instructors or peers
technology and adapt institutional guidelines based on feedback and when uncertain about ChatGPT
policies accordingly [52]. advancements [45,63]. usage [50].

4.2. Implications
The findings of this study form a basis for integrating ChatGPT into educational
practice, as it may enhance educational outcomes under certain conditions. Personalization
in learning and immediate feedback raise the level of participation of students for better
learning outcomes [62]. Therefore, this section presents recommendations for institutions,
educators, and students in integrating ChatGPT into higher education and discusses ethics
and future research directions.

4.2.1. Implications for Institutions

To this end, an institution is expected to provide an enabling environment through
which AI tools, including ChatGPT, can be utilized. Therefore, they need to make broad
policies and guidelines on effective ways of using AI technologies for teaching and
learning while ensuring institutional ethical practices and fostering responsible AI prac-
tices [45,48,52,53,59]. Institutions should also focus on infrastructure development and
resource allocation to train educators regarding these technologies. In this regard, they
would be able to create synergies between academia and the technology industry to sup-
port the integration of AI tools, like ChatGPT, into curricula in a way that is both effective
and responsible [50,55]. Moreover, institutions must keep themselves updated on the
developments of AI technologies and adjust their policies accordingly. As ChatGPT and
other AI tools continue to evolve, institutions must also be nimble and revise guidelines
in light of emerging best practices to maintain an educational environment conducive to
innovation [50,59].
As universities increasingly use AI technologies like ChatGPT, it becomes a balancing
act between embracing innovation and ensuring such uses of AI tools do not undermine
academic integrity or traditional teaching methods. The challenge lies in regulating the use
of ChatGPT in ways that foster educational enhancement while maintaining standards of
quality, fairness, and ethical conduct. With clear policies on academic integrity, personalized
learning, and ethical use, universities would, in effect, be able to control ChatGPT’s use
and plant the seeds for an innovative educational environment. The secret to success in
higher education will be finding a balance between integrating AI and traditional teaching
Computers 2025, 14, 53 18 of 36

practices so that AI enhances learning experiences without undermining critical thinking

and independent learning. Following best practices, universities can ensure that ChatGPT
and similar AI tools become valuable, ethical partners in the classroom. For example, a
university can state that students may use ChatGPT in draft form or to see how to say
something in the text. Students always need to locate where the AI-generated content is
used and edit and refine such output themselves, demonstrating critical thinking in idea
development. This ensures that AI is a tool to enhance, not replace, students’ creative
and academic abilities. Universities can use ChatGPT as AI tools that grade or provide
comments on assignments but ensure human checks to eliminate objectivity and equity. A
particular institutional policy may define how the faculty members will review assessments
or feedback from AIs so that the educational objectives would align and have academic
rigor intact.

4.2.2. Implications for Educators

For educators, this provides another avenue in which ChatGPT can aid in student-
centered learning. The teacher can utilize ChatGPT to provide feedback and encourage
students toward active learning. It is also critical, however, that educators be aware of
the technology’s limitations: it cannot recognize context fully, nor does it allow for many
prompts in critical thinking ability [44,50,52]. Thus, educators should design assessments
emphasizing critical thinking skills and encourage students to cross-check information
obtained from ChatGPT with other reliable sources [52,59]. To fully exploit the potency
of ChatGPT in teaching, there is a need to provide training for educators on both its
capabilities and limitations [45,52], which would enable it to be better integrated into an
educator’s practice in such a way as to promote learners’ autonomy rather than dependency
on the AI tool itself.

4.2.3. Implications for Students

The role of ChatGPT is to inform students about their strengths and weaknesses.
Although ChatGPT can support students richly, a student has to realize that it cannot
replace critical thinking and personal effort, as stated by [36,37,47]. Students should be
encouraged to evaluate ChatGPT’s suggestions critically about the context and relevance
of the information [43,44]. In addition, ChatGPT should be used in combination with other
resources, not as a primary source of information. Furthermore, students should be taught
about privacy and data security when using ChatGPT or other AI platforms [43,44,49,52].
They should also be warned against plagiarism by correctly referencing any AI output in
academic work [71]. As a last resort, they should also check with instructors for anything
they are unsure of by using ChatGPT or seek additional sources for further clarification to
fully understand what is being presented to them [50].

4.2.4. Ethical Considerations

The ethical implications of using ChatGPT in education are significant and need to be
considered by educators and institutions alike. Overcoming biases and privacy concerns
is crucial in gaining trust in AI-powered educational tools. Ethical considerations should
be embedded in policies and training programs so that AI will be used responsibly in
educational settings [75]. It is essential to conduct more longitudinal research showing how
ChatGPT influences student learning over a longer period and leads to the workforce. Re-
search can also be used to investigate how ChatGPT can be further customized for different
kinds of learners, providing personalized support for a wider population of students. By
focusing on this, further research can lead the way to the appropriate development of AI
tools to achieve students’ learning outcomes while infusing academic integrity.
 how ChatGPT influences student learning over a longer period and leads to the work-
force. Research can also be used to investigate how ChatGPT can be further customized
Computers 2025, 14, 53 19 of 36
for different kinds of learners, providing personalized support for a wider population of stu-
dents. By focusing on this, further research can lead the way to the appropriate development
of AI tools to achieve students’ learning outcomes while infusing academic integrity.
4.2.5. Extending ChatGPT’s Tutor or Assistant Role and Future Research Agenda
4.2.5. Extending ChatGPT’s Tutor or Assistant Role and Future Research Agenda
We provide a matrix (See Figure 4) that gives a theoretical framework for positioning
We providebeyond
AI (ChatGPT) a matrix the
(Seetypical
Figure 4) that role.
tutor gives Incorporating
a theoretical framework for positioning
key critical AI perspectives, like
AI (ChatGPT) beyond the typical tutor role. Incorporating key critical AI perspectives, like
Human-in-the-Loop learning models and AI as an Epistemic Agent, ChatGPT, and other
Human-in-the-Loop learning models and AI as an Epistemic Agent, ChatGPT, and other
tools, can be further developed toward collaborative and reflective partners in educational
tools, can be further developed toward collaborative and reflective partners in educational
processes. This
processes. This notnot only
only extends
extends theofrole
the role of AI
AI but but
also also ensures
ensures students students develop critical
develop critical
thinking,autonomy,
thinking, autonomy, andand deeper
deeper engagement
engagement with content,
with content, with the with theguiding
teacher teachertheguiding the
ethicaland
ethical andreflective
reflective aspects
aspects oflearning
of the the learning process.
process.

Engagement
Figure4.4.Engagement
Figure matrix
matrix showing
showing interactions
interactions betweenbetween
Students,Students, AI (ChatGPT),
AI (ChatGPT), and Teachersand Teachers
acrosskey
across keylearning
learning activities.
activities.

Passive Engagement: Information Retrieval and Contextualization

Passive Engagement: Information Retrieval and Contextualization
Atthe
At the level
level of passive
of passive engagement,
engagement, ChatGPTChatGPT acts classically
acts classically as an ChatGPT
as an assistant. assistant. ChatGPT
engages the student upon demand by retrieval or explanations; it
engages the student upon demand by retrieval or explanations; it is about retrieving is about retrieving
in- infor-
mation with
formation withAIAIwhile
whilethe the teacher
teacher controls againstcompliance
controls against compliance andand norms
norms for for
con-consistency
sistency on AI
on AI use use or purposefulness
or purposefulness in pursuing
in pursuing learning
learning objectives. This
objectives. This supports
supportsthe the findings
findings by Luckin et al. [76], who stated that the earliest phases of adopting AI in
by Luckin et al. [76], who stated that the earliest phases of adopting AI in education often educa-
tion often position the role of AI as supportive, providing information to students or ad-
position the role of AI as supportive, providing information to students or advice. Here,
vice. Here, the teacher is a facilitator who ensures that AI is appropriate and benefits the
the teacher is a facilitator who ensures that AI is appropriate and benefits the student’s
student’s learning process. According to Baker et al. [77], the teacher’s role in contextual-
learning
izing AI forprocess. According
the student to Baker
is important et al.the[77],
in making the teacher’s
interaction betweenrole in contextualizing
the student and AI
for the student is important in making the interaction between the student and the AI
tool responsible and productive. The teacher positions ChatGPT as a passive engagement
partner to ensure its capabilities align with pedagogical objectives.

Active Collaboration: Guidance and Epistemic Agent

As the student becomes more familiar with AI, the model shifts into Active Collabo-
ration, where ChatGPT takes on a more involved role. Beyond information retrieval, AI
facilitates Collaborative Learning: it may suggest novel ideas, stimulate questions, and
assist the learner in exploring several perspectives on one issue. Based on the replies,
the learner actively interacts with the AI as he refines his understanding of the subject of
interest. The role of the teacher in this phase is to guide and monitor the students to ensure
Computers 2025, 14, 53 20 of 36

they are not passively following but rather thinking deeply about what AI suggests. The
Human-in-the-Loop (HITL) model comes into play during this phase, where the AI and
student work in tandem, each influencing the other’s input. Holstein et al. [78] emphasize
how such HITL models encourage more active and adaptive learning processes where
AI does not just present the student with inert content but modify that content to the
developing needs of the student. The collaborative interaction underlines that AI can do
much more than answer questions; it can create an environment where students engage
with AI as collaborators in dynamic inquiry, not just as tutors.

Critical Thinking—Reflection and Ethical Oversight

In the Critical Thinking stage, students interact with the AI tool to analyze and reflect
on the information provided. In this respect, ChatGPT acts as an Epistemic Agent at this
stage, leading the student through processes of inquiry, critical evaluation, and synthesis.
ChatGPT does not deliver answers but rather creates a dialogical space where the student
can explore ideas, question assumptions, and construct deeper understanding. In this
model, the teacher’s role is further emphasized. Teachers are responsible for facilitating
critical thinking and ensuring that AI is used to support autonomous development. The
teacher encourages students to question and assess AI responses, fostering a reflective
process that helps the student construct knowledge in a deeper, more meaningful way.
Popenici & Kerr [79] emphasize the teacher’s role in encouraging students to question
the AI output and reflect on their own thinking skills, which are considered higher order
in nature.

Knowledge Construction—Collaboration and Reflection

The last stage is Knowledge Construction, where co-construction involves the student,
AI, and teacher. With ChatGPT now fully placed as an epistemic agent, it works with the
student to guide new knowledge creation. Students synthesize information and develop
insights into applying learning to real-life situations through active collaboration. At this
stage, the teacher assumes the role of overseer and facilitator of reflection to keep the
process in line with academic and ethical standards. The teacher gives students space to
reflect on how they construct their knowledge, manages the application of critical thinking,
and ensures that the learning process is rigorous and ethically sound. Stages also align with
views [80] that AI most powerfully contributes to education in knowledge creation, where
all participants—AI, students, and teachers—collaborate to shape new understandings.
Considering the above discussions, the following research questions need to be ex-
plored in future work.
(a) How does the use of ChatGPT in educational settings influence students’ academic
performance compared to the traditional teaching methodology? (b) How does ChatGPT
enhance the development of students’ critical thinking, problem-solving, and analytical
skills? (c) How does ChatGPT influence student participation and interaction in individual
and group learning environments? (d) How do educators balance AI-generated content
with human feedback to promote meaningful student learning?
These research questions represent ways ChatGPT, and similar AI tools will continue
shaping educational experiences, creating possibilities for improved learning outcomes,
student engagement, and deeper development of students’ paths. In working out how to
answer such areas, researchers will better understand how AI within the classroom should
be leveraged to support students and educators alike.

4.3. Limitations
This review had several limitations. The data collection for this study was conducted
over a period of only one year, and more papers could have been published at the time
Computers 2025, 14, 53 21 of 36

of writing. Additionally, the number of articles in the three categories was not evenly
distributed: personalized learning, academic writing, and coding tasks. Similarly, the distri-
bution of articles across categories-personalized learning, academic writing, and coding-is
not uniform, which may affect the recommendations suggested in each domain. Given
these limitations, further research should be done to extend the search period and include
more studies to analyze the long-term effect of ChatGPT on student learning outcomes
at different levels of academic disciplines. Additionally, assessing the attitudes and per-
ceptions of students toward ChatGPT would shed light on the usage of this AI tool in
educational contexts. There are some methodological limitations of the included studies.
Limitations identified in the reviewed studies are small sample sizes and lack of representa-
tiveness (e.g., [54,63]). Many focus on narrow data types or short-term effects (e.g., [63,66],
while methodological clarity (e.g., [53,65]) is often insufficient. Quite oftentimes, quality
output generated by AI, particularly in coding, is not assessed properly, as noted in two
studies (e.g., [66,67]). Also, some of these studies have sampling bias [58] and lack of
diverse data sources [64], limiting the generalizability of the findings. These recurrent
issues raise the need for more holistic, transparent, and ethical research in AI applications
in education.

5. Conclusions
This review points out ChatGPT’s potential to enhance educational practices, especially
writing, coding, and personalized learning. While it is a tool that provides valuable support,
challenges include the risk of overreliance, the propagation of biases, and the inability to
fully encourage critical thinking, self-directed learning, or the development of problem-
solving skills necessary to pursue autonomous learning. Suitable integration into the
curriculum with special emphasis on ethical guidelines by educators would contribute
to mitigating these risks. This review highlights areas where ChatGPT might benefit
education. However, most of the studies reviewed included small samples, and the studies
lack longitudinal data, limiting the generalizability of findings. Although ChatGPT has
shown promising results in some contexts, there is a need for more comparative studies
regarding its effectiveness compared to other learning tools. Long-term research on learning
outcomes of ChatGPT use, for example, is warranted, as is capturing student perceptions
and experiences of this tool to define best practices in its integration into higher education.
Future studies should involve more extensive and diverse samples and include longitudinal
designs to assess the long-term impact of ChatGPT on learning outcomes.

Supplementary Materials: The following supporting information can be downloaded at: https:
//www.mdpi.com/article/10.3390/computers14020053/s1.

Author Contributions: Conceptualization, K.N. and M.T.N.; methodology, K.N. and A.A.M.;
validation, K.N., formal analysis, K.N., A.A.M., C.C. and M.T.N.; investigation, K.N., A.A.M.,
C.C. and M.T.N.; resources, K.N., A.A.M. and M.T.N.; data curation, K.N., A.A.M. and M.T.N.;
writing—original draft preparation, K.N. and M.T.N.; writing—review and editing, K.N., A.A.M.
and C.C.; supervision, K.N. and A.A.M.; project administration, K.N. and C.C.; funding acquisition,
K.N. and C.C. All authors have read and agreed to the published version of the manuscript.

Funding: This research received no external funding.

Data Availability Statement: The reference list and Appendix A include all the information regarding
the articles that were reviewed.

Conflicts of Interest: The authors declare no conflicts of interest.

Computers 2025, 14, 53 22 of 36

Appendix A

Table A1. Summary of the included studies.

Author & Year Aim Country Samples Methods Data Collection Findings Limitations
This paper lacks clarity
Educators, administrators, regarding the
and policymakers should comprehensive literature
know the possible benefits review methodologies
To discuss the and risks of using employed, as no explicit
opportunities and details are provided.
ChatGPT in personalized
challenges of using A comprehensive learning. An ethical While it recommends the
Abbas et al., 2023 [53] Malaysia Not reported Not reported
ChatGPT for literature review framework and guidelines development of ethical
personalized learning guidelines for integrating
in higher education. regarding its
implementation and use in ChatGPT in higher
higher education should education, it falls short of
offering a thorough
be developed. discussion to support these
recommendations.
Its integration into the
instruction of the
experimental group,
embedded in a 5E
instructional model, The sample size for the
To examine proved reasonably study was relatively small,
ChatGPT’s potential A case study design successful concerning the comprising only 10
as a research tool for with a mixed students’ level of students in the control
data science methods approach: A control group and involvement with it and group and 10 students in
Albdrani and educators to quantitativedata with well-ordered learning.
AI-Shargabi, 2023 [54] Saudi Arabia 20 students an experimental the experimental group.
investigate the from quiz results and group Moving through its Additionally, the study
effectiveness of AI in qualitative successive phases clearly primarily focused on the
personalized learning observations promoted comprehensive short-term effects of using
experiences. learning while ChatGPT ChatGPT on student
played the important role learning outcomes.
of the virtual co-teacher in
encouraging agency and
greater student
engagement.
Computers 2025, 14, 53 23 of 36

Table A1. Cont.

Author & Year Aim Country Samples Methods Data Collection Findings Limitations
The primary limitation
of this study was its
small sample size, with
only nine students per
group. Moreover, the
Results showed that on analysis was restricted
To examine none of the two to quantitative data,
students’ indicators did the focusing solely on final
essay-writing A control group ChatGPT group do any essay scores evaluated
18 s-year Quantitative data
performances with and an better; the students based on mechanics,
Basic et al., 2023 [63] Croatia master’s using Excel and R
or without experimental produced content of no style, content, and
students studio
ChatGPT as an group higher quality, did not format. Incorporating
essay-writing write faster, and did not qualitative data, such as
assistance tool. have a higher degree of student feedback and
authentic text. responses to targeted
questions, could
provide a more
comprehensive
understanding of the
outcomes.
ChatGPT could explain
complex notions and
This paper lacks a
technologies, provide
transparent
To outline the role examples, guide
methodology for
and capabilities of resources, and identify
examining the role and
ChatGPT, a and solve different
capabilities of ChatGPT
language model technical issues. Its use
Biswas, 2023 [65] USA Not reported Not reported Not reported in computer
developed by can further enhance
programming, leaving
OpenAI for overall satisfaction with
significant gaps in the
computer the support services of
understanding of its
programming an organization so that
application and
it’s considered a
effectiveness.
reputable and
dependable company.
Computers 2025, 14, 53 24 of 36

Table A1. Cont.

Author & Year Aim Country Samples Methods Data Collection Findings Limitations
ChatGPT 3.5 already
demonstrates the ability
to generate code that
solves many tasks. The
model is
To investigate the
non-deterministic,
coding proficiency
meaning it may develop This paper assessed
of ChatGPT 3.5 to
different code solutions various factors,
identify potential
to the same problem. including time
areas for Testing, 10
This usually results in performance, code
development and programming
inconsistent length, ethical
examine the languages were A total 4 of
Buscemi, 2023 [67] Luxembourg Not reported. performance: given a considerations, and
ramifications of chosen, and 000 tests
specific task, the model others. However, it
automated code 40 tasks were
generates syntactically overlooked a critical
generation on the employed to test.
correct code in some aspect: evaluating the
evolution of
instances, but in other quality of code
programming
cases, it produces either produced by ChatGPT.
languages and the
bugged code or no code
tech industry.
at all. Also, the
programming language
choice influences
comprehension of the
task’s requirements.
Computers 2025, 14, 53 25 of 36

Table A1. Cont.

Author & Year Aim Country Samples Methods Data Collection Findings Limitations
It was observed that
LLMs developed some
difficulties in code
generation that led to
To explore the
bugs and errors, and
effectiveness and
multiple solutions when
efficiency of the
Four hundred sixty a slight change or
popular OpenAI This study did not
programming wrong query was
model ChatGPT, discuss ethical issues
prompts were used provided. This may lead
powered by related to using
from the Basic to inefficient time use by
GPT-3.5 and GPT-4, Testing, A ChatGPT in coding.
Coello et al., Python programmers. While
in programming Germany Not reported quantitative Also, they did not test
2024 [66] Programming ChatGPT and other
tasks to approach the prompts with
dataset LLMs can generate code
understand its students and did not
(1000 crowd-sourced effectively and thus can
impact on focus on evaluating the
programming be used as
programming and, generating codes.
problems. programming assistant
potentially,
tools, they are not
software
meant to replace human
development.
software developers, as
they always need
human feedback and
monitoring.
They briefly discussed
Educators can guide the
the challenges of using
To examine the use of generative AI
AI tools in the
potential of tools in statistics and
Three prompts, classroom. More
Ellis and Slade, ChatGPT as an A literature data science classrooms
USA Not recorded statistical data (p discussion would be
2023 [49] educational tool for review so that students and
values) useful in terms of
statistics and data educators can leverage
limitations and legal
science. the benefits of this
and ethical concerns of
technology.
using AI.
Computers 2025, 14, 53 26 of 36

Table A1. Cont.

Author & Year Aim Country Samples Methods Data Collection Findings Limitations
The discussion section
of this paper is brief,
presenting the
findings with some
Document Although the students
support from existing
analysis is used acknowledge the extensive
To understand literature. Expanding
to collect data capabilities of ChatGPT, including
perception regarding A qualitative this section to provide
Esmaeil et al., and analyze its ability to provide information
the use of ChatGPT in Malaysia 17 students research a more in-depth
2023 [46] insights into and guidance and decrease both
their argumentative approach analysis and including
complex topics research expenses and time
writing. practical
related to student consumption, they also voice
recommendations
writing. apprehensions.
based on their
findings would
significantly enhance
its value.
To ascertain what
The participants were
factors influence the The ChatGPT can increase student
not selected randomly
Fernando et al., length of time Quantitative questionnaires engagement by utilizing
Poland 287 students for conducting this
2023 [58] undergraduates questionnaires were used to multimedia and interactive
study; hence, the
receive individual collect data. teaching aids.
results may be biased.
tutoring
To discuss a range of Some of the possible benefits of
challenges and using NLP models for
This study could
opportunities for personalized learning and
benefit from a clearer
higher education, as on-demand support include the
explanation of the
well as conclude with creation of customized learning
implications of its
implications that plans, generating feedback and
Not A comprehensive findings. Additionally,
Fuchs, 2023 [45] (hopefully) expose Thailand Not reported support, and providing resources
reported review it should directly and
gaps in the literature, to students at any time and from
precisely address the
stimulate research any location. However, there are
question posed in its
ideas, and, finally, also some challenges that NLP
title: “Is ChatGPT a
advance the models may bring, including the
blessing or a curse?”
discussion about NLP loss of human interaction, bias,
in higher education. and ethical implications.
Computers 2025, 14, 53 27 of 36

Table A1. Cont.

Author & Year Aim Country Samples Methods Data Collection Findings Limitations
To highlight the use of The information
AI and AI-assisted provided by ChatGPT
technologies such as was not appropriate; it
The research conducted
the ChatGPT and Experiments were can also bring about
experiments and shared
other chatbots in conducted to test implications for medical
Guleria et al., insights on the use of AI;
scientific writing and India Not reported the authenticity Not reported science and engineering.
2023 [71] however, it lacks clarity
research, which results and accuracy of Here, critical thinking
regarding the students’
in bias, the spread of ChatGPT. should be encouraged
perspectives.
inaccurate to show awareness of
information, and related privacy and
plagiarism. ethical risks.
To present an
Teachers should permit
argument in favor of
the use of ChatGPT and
incorporating This paper did not
even be the ones to start
ChatGPT into consider many
using it since students
education, educators peer-reviewed research
will use it anyway.
should be provided papers on the use of
Halaweh, United Arab Permitting them to use
with a set of strategies Not reported Literature review Not reported ChatGPT in education.
2023 [64] Emirates the tool puts them on an
and techniques to The only consideration
equal footing in
ensure the responsible is Google Scholar as
developing ideas and
and successful their database for
improving their writing,
implementation of collecting papers.
as the faculty
ChatGPT in teaching
encourages.
or research.
To explore the opinion ChatGPT is a means to A study focusing on
of university support learners’ attitudes
professors at a time-consuming toward the use of
A questionnaire
Kiryakova and Bulgarian university teaching activities, ChatGPT in education
87 University was distributed by
Angelova, regarding the Bulgaria Survey provoke interest, would provide valuable
professors email to the
2023 [55] possibilities and activate and engage insights into the benefits
participants.
challenges of ChatGPT students, and stimulate and challenges of
in carrying out their critical thinking integrating ChatGPT
teaching activities. and creativity. into classrooms.
Computers 2025, 14, 53 28 of 36

Table A1. Cont.

Author & Year Aim Country Samples Methods Data Collection Findings Limitations
Integrating AI in
The study employed an
To identify the education offers many
open-ended question to
implications of opportunities to
identify the main
ChatGPT, an enhance learning
themes and their
AI-powered language 14 PhD experiences, personalize
Mehmet Firat, A thematic content frequencies. It would be
model, for students Turkey students, and Open question instruction, and
2023 [59] analysis approach beneficial to incorporate
and universities by seven scholars transform educators.
additional open-ended
examining the However, this shift
questions to gather
perceptions of scholars challenges assessment,
more comprehensive
and students. digital literacy, and
data.
ethical considerations.
To explore the
prospects and
obstacles associated
with using ChatGPT Students using
Providing detailed
as a tool for learning ChatGPT had an
recommendations for
and assessment in Quasi- The control and advantage in earned
Qureshi, Saudi both educators and
undergraduate 24 students experimental experimental scores, but
2023 [56] Arabia students regarding the
Computer Science research groups inconsistencies in the
use of ChatGPT would
curriculum, submitted code affected
be highly valuable.
particularly in the overall performance.
teaching and learning
fundamental
programming courses.
To examine the The discussion could be
Tertiary educators and
potential benefits and more thorough by
students must exercise
challenges of using exploring the benefits of
caution when using
Rasul et al., ChatGPT in higher Perspective type incorporating
Australia Not reported Not reported ChatGPT for academic
2023 [47] education, in the study constructivist learning
purposes to ensure its
backdrop of the principles alongside the
ethical, reliable, and
constructivist theory use of ChatGPT in the
effective use.
of learning classroom.
Computers 2025, 14, 53 29 of 36

Table A1. Cont.

Author & Year Aim Country Samples Methods Data Collection Findings Limitations
In most cases, across a
To provide a baseline range of question
The study design lacks
understanding of how A mix of formats, topics, and
sufficient detail, as it
the public release of descriptive study levels, ChatGPT
Richards et al., A dual-anonymous does not clearly outline
generative AI is likely UK Not reported statistics and is at least capable of
2024 [61] study protocol the sample size or the
to impact quality graphing of the producing adequate
procedure followed in
assurance processes data answers for
conducting the study.
significantly. undergraduate
assessment.
To present the
technology’s ChatGPT can be used to
implications for higher check sentences for This paper only
education and discuss plagiarism input by the considered two
Rudolph et al., the future of learning, A desktop analysis user and then modify peer-reviewed and eight
Singapore Not reported Not reported
2023 [81] teaching, and approach them so that academic papers on
assessment in higher anti-plagiarism software ChatGPT and higher
education in the reports a low originality education.
context of AI chatbots index score.
such as ChatGPT.
There are currently no
To systematically
A-students and no
compare selected
A Systematic B-students in this bot The study could discuss
Rudolph et al., chatbots across a
Singapore Not reported comparison within 15 Test questions cohort, despite all the implications of their
2023 [52] multi-disciplinary test
the Chatbots. publicized and research.
relevant to higher
sensationalist claims to
education.
the contrary.
Computers 2025, 14, 53 30 of 36

Table A1. Cont.

Author & Year Aim Country Samples Methods Data Collection Findings Limitations
Of particular note, the
majority of students
involved in this research
showed interest in the
usage of the tool as a
Half of the students
Three distinct supportive tool for
struggled to effectively
stages: Initial tests teaching in the
utilize ChatGPT’s
To gauge the viability on ChatGPT on classroom to develop
resources, highlighting
of ChatGPT in chatbot, learning sustainable and
Silva et al., the need for a deeper
programming Brazil 40 students and teaching code A Questionnaire improved learning. The
2024 [70] understanding of these
education and on ChatGPT, and integration of ChatGPT
challenges in order to
sustainability. student experience into coding and
enhance the AI tool’s
using ChatGPT in programming courses
support for their
coding changes students’
learning.
perceptions regarding
educational support,
sustainability, and
individual learning
experiences.
ChatGPT can be helpful
To compare selected in learning/teaching This study did not
chatbots across a activities, but better compare student
Singh et al.,
multi-disciplinary test UK 430 students A Survey A Questionnaire guidelines should be performance based on
2023 [50]
relevant to higher provided for the their level of proficiency
education. students in using the in using ChatGPT.
tool.
Computers 2025, 14, 53 31 of 36

Table A1. Cont.

Author & Year Aim Country Samples Methods Data Collection Findings Limitations
While ChatGPT indeed
does a credible job for This study could offer
many tasks, its response more insight into the
is detailed, often academic integrity
beyond what even a concerns and risks
To explore how
human expert would associated with using
ChatGPT can be used 15 common
Sridhara et al, Testing and have come up with or ChatGPT for coding
to help with common India Not reported software
2023 [68] observation even state of the art. tasks. Additionally,
software engineering engineering tasks
However, it also turns conducting a test with
tasks.
out that ChatGPT students to gather their
provides incorrect perspectives on using
answers for a few other this tool would provide
tasks and, hence, is not valuable information.
suited for such tasks.
Focusing on a single
ChatGPT gave users
instance, particularly
sufficient content to
the feature of
form an overall
autoethnographies, is a
impression of its
limitation of this study.
technical features, and
To examine ChatGPT’s Learning approaches,
An users felt the response it
Stojanow, use as a tool aiding the New experiences, and how
Not reported autoethnographic Not reported provided to be engaging
2023 [82] learning process has Zealand students interact with
study examining and relevant. Answers
not been examined. technology vary among
were, however,
individuals, which
somewhat superficial;
means the insights
generated text was not
gained by the author
always logical, even
may not be applicable to
contradictory at times.
others.
Computers 2025, 14, 53 32 of 36

Table A1. Cont.

Author & Year Aim Country Samples Methods Data Collection Findings Limitations
There was mixed public
discussion and
university responses,
To examine news
with a focus mainly on
articles (N = 100)
academic integrity
about how ChatGPT is
concerns and
disrupting higher They analyzed coverage
opportunities for
education, in mainstream news
Sullivan et al., Content analysis/a innovative assessment
concentrating Australia Not reported Not reported databases but did not
2023 [48] systematic search design.There has also
specifically on explore alternative news
been a lack of public
Australia, New sources.
discussion about the
Zealand, the United
potentialfor ChatGPT to
States, and the United
enhance participation
Kingdom.
and success for students
from disadvantaged
backgrounds.
This research was
It would be useful to limited to an 8-week
To analyze the
A questionnaire integrate generative AI period. Future studies
students’ perspectives
Yilmaz and and a form tools into programming could benefit from a
on using ChatGPT in The case study
Karaoglan Turkey 41 students consisting of courses, considering the longitudinal approach
the field of method
Yilmaz, 2023 [43] open-ended advantages they to examine the
programming and
questions provide in perspectives of students
programming learning
programming teaching. who use ChatGPT over
a longer duration.
Computers 2025, 14, 53 33 of 36

Table A1. Cont.

Author & Year Aim Country Samples Methods Data Collection Findings Limitations
This study was
exploratory and relied
solely on survey
questions to gauge
students’ acceptance of
ChatGPT. It did not
There was powerful
doctoral students’ employ a case study or
evidence for the
acceptance toward descriptive mixed-methods
Zou and Huang, 242 doctoral applicability of the
ChatGPT in writing China Online survey analysis and research design, nor did
2023 [44] students Technology Acceptance
and the factors that correlation analysis it collect multiple
Model in the acceptance
influence it. sources of data to gain a
of ChatGPT in writing
more nuanced and
in-depth understanding
of students’ actual
processes and outcomes
when using ChatGPT
for writing.
Computers 2025, 14, 53 34 of 36

References
1. Brown, T.; Mann, B.; Ryder, N.; Subbiah, M.; Kaplan, J.D.; Dhariwal, P.; Neelakantan, A.; Shyam, P.; Sastry, G.; Askell, A. Language
models are few-shot learners. Adv. Neural Inf. Process. Syst. 2020, 33, 1877–1901.
2. Wu, H.; Wang, W.; Wan, Y.; Jiao, W.; Lyu, M. ChatGPT or Grammarly? Evaluating ChatGPT on Grammatical Error Correction
Benchmark. arXiv 2023, arXiv:2303.13648.
3. Das, A.; Malaviya, S.; Singh, M. The Impact of AI-Driven Personalization on Learners’ Performance. Int. J. Comput. Sci. Eng. 2023,
11, 15–22. [CrossRef]
4. Imran, M.; Almusharraf, N. Analyzing the role of ChatGPT as a writing assistant at higher education level: A systematic review
of the literature. Contemp. Educ. Technol. 2023, 15, ep464. [CrossRef]
5. Roumeliotis, K.I.; Tselikas, N.D. ChatGPT and Open-AI Models: A Preliminary Review. Future Internet 2023, 15, 192. [CrossRef]
6. Ray, P.P. ChatGPT: A comprehensive review on background, applications, key challenges, bias, ethics, limitations and future
scope. Internet Things Cyber-Phys. Syst. 2023, 3, 121–154. [CrossRef]
7. Dwivedi, Y.K.; Kshetri, N.; Hughes, L.; Slade, E.L.; Jeyaraj, A.; Kar, A.K.; Baabdullah, A.M.; Koohang, A.; Raghavan, V.; Ahuja,
M.; et al. Opinion Paper: “So what if ChatGPT wrote it?” Multidisciplinary perspectives on opportunities, challenges and
implications of generative conversational AI for research, practice and policy. Int. J. Inf. Manag. 2023, 71, 102642. [CrossRef]
8. Javaid, M.; Haleem, A.; Singh, R.P.; Khan, S.; Khan, I.H. Unlocking the opportunities through ChatGPT Tool towards ameliorating
the education system. BenchCouncil Trans. Benchmarks Stand. Eval. 2023, 3, 100115. [CrossRef]
9. Tian, H.; Lu, W.; Li, T.O.; Tang, X.; Cheung, S.-C.; Klein, J.; Bissyandé, T.F. Is ChatGPT the Ultimate Programming Assistant—How
far is it? arXiv 2023. [CrossRef]
10. Akbar, M.A.; Khan, A.A.; Liang, P. Ethical Aspects of ChatGPT in Software Engineering Research. arXiv 2023. [CrossRef]
11. Flower, L.; Hayes, J.R. A cognitive process theory of writing. Coll. Compos. Commun. 1981, 32, 365–387. [CrossRef]
12. Kellogg, R.T. Training writing skills: A cognitive developmental perspective. J. Writ. Res. 2008, 1, 1–26. [CrossRef]
13. Grover, S.; Pea, R. Computational thinking in K–12: A review of the state of the field. Educ. Res. 2013, 42, 38–43. [CrossRef]
14. Lye, S.Y.; Koh, J.H.L. Review on teaching and learning of computational thinking through programming: What is next for K-12?
Comput. Hum. Behav. 2014, 41, 51–61. [CrossRef]
15. Hattie, J.; Yates, G.C. Visible Learning and the Science of How We Learn; Routledge: London, UK, 2013.
16. Walkington, C.A. Using adaptive learning technologies to personalize instruction to student interests: The impact of relevant
contexts on performance and learning outcomes. J. Educ. Psychol. 2013, 105, 932. [CrossRef]
17. Pashler, H.; McDaniel, M.; Rohrer, D.; Bjork, R. Learning styles: Concepts and evidence. Psychol. Sci. Public Interest 2008, 9,
105–119. [CrossRef] [PubMed]
18. Coffield, F.; Ecclestone, K.; Hall, E.; Moseley, D. Learning Styles and Pedagogy in Post-16 Learning: A Systematic and Critical Review;
Learning & Skills Research Centre: London, UK, 2004.
19. Riener, C.; Willingham, D. The myth of learning styles. Change Mag. High. Learn. 2010, 42, 32–35. [CrossRef]
20. Hattie, J. Visible Learning: A Synthesis of over 800 Meta-Analyses Relating to Achievement; Routledge: London, UK, 2008.
21. Taylor, D.L.; Yeung, M.; Bashet, A. Personalized and adaptive learning. In Innovative Learning Environments in STEM Higher
Education: Opportunities, Challenges, and Looking Forward; Springer: Cham, Switzerland, 2021; pp. 17–34.
22. Al Shloul, T.; Mazhar, T.; Iqbal, M.; Ghadi, Y.; Malik, F.; Hamam, H. Role of activity-based learning and ChatGPT on students’
performance in education. Comput. Educ. Artif. Intell. 2024, 6, 100219. [CrossRef]
23. Yılmaz, Ö. Personalised learning and artificial intelligence in science education: Current state and future perspectives. Educ.
Technol. Q. 2024, 2024, 255–274. [CrossRef]
24. Labadze, L.; Grigolia, M.; Machaidze, L. Role of AI chatbots in education: Systematic literature review. Int. J. Educ. Technol. High.
Educ. 2023, 20, 56. [CrossRef]
25. Vygotsky, L.S. Mind in Society: The Development of Higher Psychological Processes; Harvard University Press: Cambridge, MA, USA,
1978; Volume 86.
26. Hatmanto, E.D.; Sari, M.I. Aligning Theory and Practice: Leveraging Chat GPT for Effective English Language Teaching and
Learning. E3S Web Conf. 2023, 440, 05001. [CrossRef]
27. Jonassen, D.H. Objectivism versus constructivism: Do we need a new philosophical paradigm? Educ. Technol. Res. Dev. 1991, 39,
5–14. [CrossRef]
28. Kostka, I.; Toncelli, R. Exploring applications of ChatGPT to English language teaching: Opportunities, challenges, and recom-
mendations. Tesl-Ej 2023, 27, n3. [CrossRef]
29. Norman Donald, A. The Design of Everyday Things; MIT Press: Cambridge, MA, USA, 2013.
30. Jo, H.; Park, D.-H. Effects of ChatGPT’s AI capabilities and human-like traits on spreading information in work environments.
Sci. Rep. 2024, 14, 7806. [CrossRef] [PubMed]
31. Menon, D.; Shilpa, K. Chatting with ChatGPT”: Analyzing the factors influencing users’ intention to Use the Open AI’s ChatGPT
using the UTAUT model. Heliyon 2023, 9, e20962. [CrossRef]
Computers 2025, 14, 53 35 of 36

32. Liu, J. ChatGPT: Perspectives from human–computer interaction and psychology. Front. Artif. Intell. 2024, 7, 1418869. [CrossRef]
33. Lo, C.K. What Is the Impact of ChatGPT on Education? A Rapid Review of the Literature. Educ. Sci. 2023, 13, 410. [CrossRef]
34. Mai, D.T.T.; Da, C.V.; Hanh, N.V. The use of ChatGPT in teaching and learning: A systematic review through SWOT analysis
approach. Front. Educ. 2024, 9, 1328769. [CrossRef]
35. Zhang, P.; Tur, G. A systematic review of ChatGPT use in K-12 education. Eur. J. Educ. 2024, 59, e12599. [CrossRef]
36. Chan, C.K.Y.; Hu, W. Students’ voices on generative AI: Perceptions, benefits, and challenges in higher education. Int. J. Educ.
Technol. High. Educ. 2023, 20, 43. [CrossRef]
37. Montenegro-Rueda, M.; Fernández-Cerero, J.; Fernández-Batanero, J.M.; López-Meneses, E. Impact of the Implementation of
ChatGPT in Education: A Systematic Review. Computers 2023, 12, 153. [CrossRef]
38. Bettayeb, A.M.; Abu Talib, M.; Sobhe Altayasinah, A.Z.; Dakalbab, F. Exploring the impact of ChatGPT: Conversational AI in
education. Front. Educ. 2024, 9, 1379796. [CrossRef]
39. Baig, M.I.; Yadegaridehkordi, E. ChatGPT in the higher education: A systematic literature review and research challenges. Int. J.
Educ. Res. 2024, 127, 102411. [CrossRef]
40. Page, M.J.; McKenzie, J.E.; Bossuyt, P.M.; Boutron, I.; Hoffmann, T.C.; Mulrow, C.D.; Shamseer, L.; Tetzlaff, J.M.; Akl, E.A.;
Brennan, S.E.; et al. The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. Int. J. Surg. 2021, 88,
105906. [CrossRef]
41. Clarke, V.; Braun, V. Thematic analysis. J. Posit. Psychol. 2017, 12, 297–298. [CrossRef]
42. Rietsche, R.; Aier, S.; Rivera, M. Does Real-Time Feedback Matter? A Simulation Study to Link Individual and Organizational
Performance. 2021. Available online: https://www.alexandria.unisg.ch/entities/publication/2faa947b-0832-4db2-a205-38362
119eb5a (accessed on 1 January 2025).
43. Yilmaz, R.; Karaoglan Yilmaz, F.G. Augmented intelligence in programming learning: Examining student views on the use of
ChatGPT for programming learning. Comput. Hum. Behav. Artif. Hum. 2023, 1, 100005. [CrossRef]
44. Zou, M.; Huang, L. To use or not to use? Understanding doctoral students’ acceptance of ChatGPT in writing through technology
acceptance model. Front. Psychol. 2023, 14, 1259531. [CrossRef] [PubMed]
45. Fuchs, K. Exploring the opportunities and challenges of NLP models in higher education: Is Chat GPT a blessing or a curse?
Front. Educ. 2023, 8, 1166682. [CrossRef]
46. Esmaeil, A.; Maakip, I.; Ag Kiflee, D.N.; Marshall, S.; Matanluk, O. Understanding Student Perception Regarding The Use of
ChatGPT in Their Argumentative Writing: A Qualitative Inquiry. J. Komun. Malays. J. Commun. 2023, 39, 150–165. [CrossRef]
47. Rasul, T.; Nair, S.; Kalendra, D.; Robin, M.; De Oliveira Santini, F.; Ladeira, W.J.; Sun, M.; Day, I.; Rather, R.A.; Heathcote, L. The
role of ChatGPT in higher education: Benefits, challenges, and future research directions. J. Appl. Learn. Teach. 2023, 6, 41–56.
[CrossRef]
48. Sullivan, M.; Kelly, A.; McLaughlan, P. ChatGPT in higher education: Considerations for academic integrity and student learning.
J. Appl. Learn. Teach. 2023, 6, 1–10. [CrossRef]
49. Ellis, A.R.; Slade, E. A New Era of Learning: Considerations for ChatGPT as a Tool to Enhance Statistics and Data Science
Education. J. Stat. Data Sci. Educ. 2023, 31, 128–133. [CrossRef]
50. Singh, H.; Tayarani-Najaran, M.H.; Yaqoob, M. Exploring Computer Science Students’ Perception of ChatGPT in Higher Education:
A Descriptive and Correlation Study. Educ. Sci. 2023, 13, 924. [CrossRef]
51. Peng, H.; Ma, S.; Spector, J.M. Personalized adaptive learning: An emerging pedagogical approach enabled by a smart learning
environment. Smart Learn. Environ. 2019, 6, 9. [CrossRef]
52. Rudolph, J.T.S.; Tan, S. ChatGPT: Bullshit spewer or the end of traditional assessments in higher education? J. Appl. Learn. Teach.
2023, 6, 342–363. [CrossRef]
53. Abbas, M.; Jam, F.A.; Khan, T.I. Is it harmful or helpful? Examining the causes and consequences of generative AI usage among
university students. Int. J. Educ. Technol. High. Educ. 2024, 21, 10. [CrossRef]
54. Albdrani, R.N.; Al-Shargabi, A.A. Investigating the Effectiveness of ChatGPT for Providing Personalized Learning Experience: A
Case Study. Int. J. Adv. Comput. Sci. Appl. 2023, 14, 1208–1213. [CrossRef]
55. Kiryakova, G.; Angelova, N. ChatGPT—A Challenging Tool for the University Professors in Their Teaching Practice. Educ. Sci.
2023, 13, 1056. [CrossRef]
56. Qureshi, B. Exploring the Use of ChatGPT as a Tool for Learning and Assessment in Undergraduate Computer Science Curriculum:
Opportunities and Challenges. arXiv 2023. [CrossRef]
57. Stojanov, A. Learning with ChatGPT 3.5 as a more knowledgeable other: An autoethnographic study. Int. J. Educ. Technol. High.
Educ. 2023, 20, 35. [CrossRef]
58. Fernando, C.; Banarse, D.; Michalewski, H.; Osindero, S.; Rocktäschel, T. Promptbreeder: Self-Referential Self-Improvement Via
Prompt Evolution. arXiv 2023, arXiv:2309.16797.
59. Firat, M. What ChatGPT means for universities: Perceptions of scholars and students. J. Appl. Learn. Teach. 2023, 6. [CrossRef]
Computers 2025, 14, 53 36 of 36

60. Pintrich, P.R. Chapter 14—The Role of Goal Orientation in Self-Regulated Learning; Elsevier Inc.: Amsterdam, The Netherlands, 2000;
pp. 451–502. [CrossRef]
61. Richards, M.; Waugh, K.; Slaymaker, M.; Petre, M.; Woodthorpe, J.; Gooch, D. Bob or Bot: Exploring ChatGPT’s Answers to
University Computer Science Assessment. ACM Trans. Comput. Educ. 2024, 24, 1–32. [CrossRef]
62. McMurtrie, B. AI and the Future of Undergraduate Writing. Available online: https://www.chronicle.com/article/ai-and-the-
future-of-undergraduate-writing (accessed on 1 January 2025).
63. Bašić, Ž.; Banovac, A.; Kružić, I.; Jerković, I. ChatGPT-3.5 as writing assistance in students’ essays. Humanit. Soc. Sci. Commun.
2023, 10, 750–755. [CrossRef]
64. Halaweh, M. ChatGPT in education: Strategies for responsible implementation. Contemp. Educ. Technol. 2023, 15, ep421.
[CrossRef]
65. Biswas, S. Role of ChatGPT in Computer Programming. Mesopotamian J. Comput. Sci. 2023, 2023, 9–15. [CrossRef]
66. Coello, C.E.A.; Alimam, M.N.; Kouatly, R. Effectiveness of ChatGPT in Coding: A Comparative Analysis of Popular Large
Language Models. Digital 2024, 4, 114–125. [CrossRef]
67. Buscemi, A. A Comparative Study of Code Generation using ChatGPT 3.5 across 10 Programming Languages. arXiv 2023.
[CrossRef]
68. Sridhara, G.; G, R.H.; Mazumdar, S. ChatGPT: A Study on its Utility for Ubiquitous Software Engineering Tasks. arXiv 2023.
[CrossRef]
69. Arefin, S.E.; Heya, T.A.; Al-Qudah, H.; Ineza, Y.; Serwadda, A. Unmasking the giant: A comprehensive evaluation of ChatGPT’s
proficiency in coding algorithms and data structures. arXiv 2023. [CrossRef]
70. Silva, C.A.G.d.; Ramos, F.N.; de Moraes, R.V.; Santos, E.L.d. ChatGPT: Challenges and Benefits in Software Programming for
Higher Education. Sustainability 2024, 16, 1245. [CrossRef]
71. Guleria, A.; Krishan, K.; Sharma, V.; Kanchan, T. ChatGPT: Ethical concerns and challenges in academics and research. J. Infect.
Dev. Ctries. 2023, 17, 1292–1299. [CrossRef]
72. Hasanein, A.M.; Sobaih, A.E.E. Drivers and Consequences of ChatGPT Use in Higher Education: Key Stakeholder Perspectives.
Eur. J. Investig. Health Psychol. Educ. 2023, 13, 2599–2614. [CrossRef] [PubMed]
73. Prince, M. Does Active Learning Work? A Review of the Research. J. Eng. Educ. 2004, 93, 223–231. [CrossRef]
74. Holmes, W. Artificial Intelligence in Education: Promises and Implications for Teaching and Learning; Center for Curriculum Redesign:
Boston, MA, USA, 2019.
75. Yu, H. The application and challenges of ChatGPT in educational transformation: New demands for teachers’ roles. Heliyon 2024,
10, e24289. [CrossRef]
76. Luckin, R.; Holmes, W. Intelligence Unleashed: An Argument for AI in Education; UCL Knowledge Lab: London, UK, 2016.
77. Baker, T.; Smith, L.; Anissa, N. Exploring the Future of Artificial Intelligence in Schools and Colleges. 2019. Available online:
https://www.nesta.org.uk/report/education-rebooted/ (accessed on 1 January 2025).
78. Holstein, K.; Aleven, V. Designing for human–AI complementarity in K-12 education. AI Mag. 2022, 43, 239–248. [CrossRef]
79. Popenici, S.A.; Kerr, S. Exploring the impact of artificial intelligence on teaching and learning in higher education. Res. Pract.
Technol. Enhanc. Learn. 2017, 12, 22. [CrossRef]
80. Kukulska-Hulme, A.; Lee, H.; Norris, L. Mobile learning revolution: Implications for language pedagogy. In The Handbook of
Technology and Second Language Teaching and Learning; Wiley: Hoboken, NJ, USA, 2017; pp. 217–233.
81. Rudolph, J.; Tan, S.; Tan, S. War of the chatbots: Bard, Bing Chat, ChatGPT, Ernie and beyond. The new AI gold rush and its
impact on higher education. J. Appl. Learn. Teach. 2023, 6, 364–389.
82. Stojanov, A. Achievement Goal Motivation and Rliance on ChatGPT for learning. International Psychological Applications Conference and
Trends (InPACT 2024); Science Press: Porto, Portugal, 2024.

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.