Sustainability 15 12451

sustainability
Article
New Era of Artificial Intelligence in Education: Towards a
Sustainable Multifaceted Revolution
Firuz Kamalov 1, *, David Santandreu Calonge 2 and Ikhlaas Gurrib 3
1 Department of Electrical Engineering, Canadian University Dubai, Dubai 117781, United Arab Emirates
2 Academic Development, Mohamed bin Zayed University of Artificial Intelligence,
Abu Dhabi 7256, United Arab Emirates; [email protected]
3 Department of Finance and Accounting, Canadian University Dubai, Dubai 117781, United Arab Emirates;
[email protected]
* Correspondence: [email protected]
Abstract: The recent high performance of ChatGPT on several standardized academic tests has thrust
the topic of artificial intelligence (AI) into the mainstream conversation about the future of education.
As deep learning is poised to shift the teaching paradigm, it is essential to have a clear understanding
of its effects on the current education system to ensure sustainable development and deployment of
AI-driven technologies at schools and universities. This research aims to investigate the potential
impact of AI on education through review and analysis of the existing literature across three major
axes: applications, advantages, and challenges. Our review focuses on the use of artificial intelligence
in collaborative teacher–student learning, intelligent tutoring systems, automated assessment, and
personalized learning. We also report on the potential negative aspects, ethical issues, and possible
future routes for AI implementation in education. Ultimately, we find that the only way forward is to
embrace the new technology, while implementing guardrails to prevent its abuse.
Keywords: AI; deep learning; education; intelligent systems; ChatGPT
Citation: Kamalov, F.; Santandreu

Calonge, D.; Gurrib, I. New Era of
1. Introduction
Artificial Intelligence in Education:
Towards a Sustainable Multifaceted Artificial intelligence (AI) has quickly established itself as a transformative force in a
Revolution. Sustainability 2023, 15, wide range of industries, including education. The development of AI has resulted in an
12451. https://doi.org/10.3390/ array of advancements and innovations that have impacted many facets of human life. As
su151612451 a fundamental component to societal evolution and individual development, education
has had significant benefits from AI breakthroughs. The integration of AI in educational
Academic Editors: Patricio E.
Ramirez-Correa and Andreas Ch.
systems is altering the ways in which students learn, teachers educate, and institutions
Hadjichambis
function. By personalizing learning experiences, automating administrative responsibilities,
and delivering real-time feedback, AI is revolutionizing the educational landscape, bridging
Received: 5 June 2023 gaps, and encouraging a more inclusive and effective learning environment. Given the
Revised: 26 July 2023 importance of integrating AI in education, there is a need to reflect on its implications.
Accepted: 28 July 2023 Our goal is to study the potential impact of AI on education based on the review
Published: 16 August 2023
of the current literature. We focus on three major themes: applications, advantages, and
challenges. The review procedure consists of searching the Scopus database using the
relevant keywords. The results are filtered, sorted, and analyzed to extract the pertinent
information. We identify several subfields within each theme that are used to categorize the
Copyright: © 2023 by the authors.
Licensee MDPI, Basel, Switzerland.
literature and structure the problem in a more coherent fashion. The results are presented
This article is an open access article
in several sections according to the major themes and subthemes. We find that the best
distributed under the terms and way forward is to embrace the new technology, while implementing guardrails to prevent
conditions of the Creative Commons its abuse.
Attribution (CC BY) license (https:// One of the key applications of AI is natural language processing (NLP). The aim of
creativecommons.org/licenses/by/ NLP is to develop intelligent systems that can understand human text and speech. In
4.0/). particular, intelligent chatbots have been increasingly deployed in various industries to
Sustainability 2023, 15, 12451. https://doi.org/10.3390/su151612451 https://www.mdpi.com/journal/sustainability

Sustainability 2023, 15, 12451 2 of 27
provide customer service and support other tasks [1]. The development of modern chatbots
began in 2016 and has accelerated up to the current date [2]. The advent of chatbots has
also affected the field of education [3]. A recent survey found that the use of chatbots in
education has been steadily increasing [4]. Several studies found that chatbots can improve
students’ learning experiences and facilitate their education [5].
The turning point in the adoption of AI in society came in November, 2022 with
the release of ChatGPT. The advanced writing and comprehension abilities of ChatGPT
surprised many people, earning a wide-ranging audience and garnering unprecedented
attention. It was the first time that an audience outside the machine learning community
truly realized the potential and immediacy of AI. The education sector was arguably the
most affected by ChatGPT. The potential of ChatGPT to deliver intelligent tutoring systems,
on one hand, and as a tool for academic dishonesty, on the other hand, has sparked an
intense debate. Educators at secondary and tertiary education institutions have raised
the alarm over the possibility of the abuse of ChatGPT by students and called for its
restrictions. School districts in Australia’s Queensland and Tasmania schools and New York
City and Seattle have prohibited the use of ChatGPT on students’ devices and networks.
Several universities, colleges, and schools are evaluating similar restrictions [6]. However,
it appears impossible to prevent the students from using AI. As highlighted in [7], ChatGPT
has great potential to provide solutions to college students on a range of tasks from essay
writing to code creation. Ultimately, the best way forward is to incorporate AI into the
educational system and leverage its capabilities to deliver better learning outcomes for the
students. In order to advance the debate over the optimal approach to utilizing AI, in this
paper, we investigate its potential benefits as well dangers (Figure 1).
Personalized
Learning
Intelligent
Assessment
Tutoring
Automation
System
Applications
Enhanced Data Privacy

Learning AI in Education and Security
Outcomes
Benefits Challenges
Bias and
Time and Cost
Discrimination
Efficiency
Plagiarism
Global and
Access to Academic
Education Integrity
Figure 1. Multifaceted impact of AI in education.
The potential applications of AI in education include personalized learning, intelligent

tutoring systems, automation of assessment, and teacher–student collaboration [4]. Per-
sonalized learning is possible given the scalability of AI to the entire student population.
AI algorithms such as reinforcement learning can be used to dynamically learn about the
individual needs of a student and adapt the learning process accordingly. In connection
with personalized learning, intelligent tutoring systems can be developed that can actively
interact with students, giving valuable feedback. Another impactful aspect of AI is automa-
tion of the assessments. Computer vision and natural language processing systems can be
combined to automatically grade homework, quizzes, and exams. Automated grading will
provide a tremendous relief to instructors, giving them more time to spend with students.
AI can also be useful in facilitating teacher–student collaboration by providing various
feedback and analytics.
The applications of AI in education highlight the potential for huge advantages that
are made possible by intelligent systems. The impact of AI can be seen in improved learning
outcomes, time and cost efficiency, global access to quality education, and other benefits.
Personalized learning and intelligent tutoring systems can help improve learning outcomes
for students, especially in underserved populations. The global reach and scalability of
AI will allow students from both developed as well as developing nations to benefit from
better learning experiences. Automated grading will have massive cost- and time-saving
benefits in education. Currently, around 40% of teachers’ time is spent on grading and
related activities. Without the burden of grading, teachers will be able to spend more time
with students and provide more learning support.
While the applications and benefits of AI in education can paint an alluring picture,
it is important to be aware of potential hazards of introducing autonomous systems in
education. Since children are more susceptible than adults to misinformation, the use of
AI in education should be properly pretested and carefully monitored. Potential issues
include data privacy and security, bias and discrimination, and the teacher–student relation-
ship. Certain applications of AI such as personalized learning require students’ personal
information. For instance, knowing that a student has a learning disability or a mental
health issue will allow AI to select the appropriate approach and customize its content
accordingly. While students’ personal information can be used for tremendous benefit, it
can also be susceptible to privacy and security problems. Anonymizing and encrypting
the student data will alleviate some of the concerns. However, a comprehensive strategy is
required to address this issue. Another important issue is bias and discrimination. Since
AI is trained on public data it can be exposed to the biases that exist on the internet. In
addition, AI algorithms can also inadvertently learn bias on their own. Since there is a
significant amount of entropy in the AI algorithms, their behavior could be unpredictable.
Minimizing the amount of bias is one of the key challenges in applying AI in education.
New technology has historically held potential for misuse. The discovery of nuclear
fission created the devastating nuclear bomb. The advent of the internet created the dark
web, where illicit and illegal activity can be hidden from the government. However, the
society has been able to limit the potential for the abuse of technology through international
cooperation and law enforcement. In general, the benefits of new technology outweigh its
dangers. Rather than stopping or preventing the advancement of new AI technology in
education, it will be more beneficial, on balance, to integrate it into the curriculum. The
example of Khan Academy that has recently partnered with OpenAI to integrate ChatGPT
into their systems shows a potential roadmap to adopting AI in education. Ultimately,
the only way forward is to accept and embrace the new technology, while implementing
guardrails to prevent its abuse.
The main contributions of the paper can be summarized as follows:
1. Review the existing literature related to AI in education;
2. Analyze the potential impact of AI in education;
3. Identify the main avenues in applications, benefits, and challenges of AI in education.
Our paper is organized as follows. Section 2 provides background details related to
ChatGPT. In Section 3, we present key applications of AI in education. In Section 4, we
discuss the benefits of employing AI in education. Section 5 discusses potential dangers of
AI. Section 6 describes future directions and opportunities of AI in education.
2. ChatGPT
The release of ChatGPT catalyzed the discussion around the benefits as well as dangers
of AI. It marked the turning point in the conscience of many people about their perception
of AI. While hitherto AI was regarded as mostly a sci-fi fantasy that existed in a far-distant
future, the arrival of ChatGPT has suddenly made everyone keenly aware of the legitimacy
and tenability of AI. The advent of ChatGPT has increased competition and accelerated
the development of alternative AI models motivating the creation of Google’s Bard and
Meta’s LLaMA. In this section, we present a brief background of ChatGPT to provide a
more complete picture for the discussion.
ChatGPT is an AI chatbot that was developed by OpenAI. It was initially released on
30 November 2022 based on GPT-3.5 and subsequently updated on 14 March 2023 based on
GPT-4 [8]. It is considered by many as the most powerful AI tool ever crated [9]. ChatGPT
is large language model based on a generative pre-trained transformer (GPT) that is further
tuned via supervised and reinforcement learning techniques. It is able to comprehend
and respond to a large variety of prompts with a high level of expertise as well as carry
on a continuous dialogue with the user. It can perform a range of tasks from writing
poetry in a specified manner and style to generating computer code according to given
requirements. While its responses are not perfect, ChatGPT has achieved unprecedented
levels of performance. As shown in Figure 2, its latest release based on GPT-4 is able to
achieve above-average human performance on several standardized tests including AP
tests, SAT, LSAT, and GRE [10]. The success of ChatGPT has been the inflection point in the
adoption of AI in society, including education. It has highlighted the abilities of AI and
sparked the discourse about its future.
Figure 2. Performance of GPT models on various standardized tests.
2.1. Transformer
As with most of the existing large language models including Bard and LLaMA,
ChatGPT utilizes the transformer architecture. Transformer is a sequence-to-sequence
neural network model that was originally introduced for language translation but later
adopted for general purpose language modeling [11]. As shown in Figure 3, transformer
utilizes an encoder–decoder architecture to build a generative model [11].
Figure 3. The transformer model utilizes the encoder–decoder neural network architecture.
The key feature of the transformer model is the adoption of self-attention. The attention
mechanism provides context for any position in the input sequence, which allows it to
process the entire sequence simultaneously. This enables greater parallelization than the
recurrent neural networks (RNNs) and therefore reduces training times. Thus, transformers
can be trained on large amounts of data which would otherwise be impossible if using
RNNs or long short-term memory (LSTM).
The encoder portion of the transformer neural network (Figure 3) is responsible for
generating encodings that contain information about which parts of the inputs are relevant
to each other. The encoder consist of three major layers: input embedding, multi-head
attention, and feed-forward layers. The input embedding layer converts a sequence of input
tokens into vectors in the embedding space and adds positional information about each
token, which ensures the model can consider the order of elements in the sequence. The
embeddings are passed to the attention layer where the model learns various relationships
between the input tokens by weighting their importance based on their positions and
semantic content. The output of the self-attention is passed on to the feed-forward layer,
which applies non-linear transformations to further extract higher-level features. These
layers are interconnected with residual connections and layer normalization to stabilize
and accelerate training. In the end, the encoder outputs rich, context-aware representations
for downstream task.
The decoder functions in a reverse manner from the encoder. It takes the output of the
encoder and converts it into sequences. The decoder consists of three major layers: masked
multi-head self-attention, encoder–decoder attention, and position-wise feed-forward net-
works. The masked self-attention mechanism prevents the model from accessing future
tokens in the output sequence during training, which ensures that the model is trained
autoregressively. The encoder–decoder attention layer then enables the decoder to attend
to the encoder’s output, allowing it to incorporate the input sequence’s contextual infor-
mation. Finally, the feed-forward networks apply non-linear transformations to refine the
generated features using batch normalization and residual connections to improve the
training process.
2.2. GPT
The transformer architecture gives rise to several popular language models including
GPT. In particular, GPT architecture is based on the decoder part of the transformer. It is
trained to predict the next token in a sequence given the previous tokens in an autoregres-
sive manner, i.e., it is unidirectional, as it processes the input text from left to right, focusing
on learning a language model. The distinguishing feature of GPT models is their scale.
While no official information has been released regarding GPT-4, the GPT-3 model consists
of 175 billion parameters. GPT-3 was trained on 570 GB plaintext and 0.4 trillion tokens
using mostly CommonCrawl, WebText, English Wikipedia, and two books corpora (Books1
and Books2). After pretraining, the model was fine-tuned using zero-shot, one-shot, and
few-shot learning. It has been speculated that GPT-4 was trained on 1 trillion parameters
and cost USD 100 million to train.
3. Scoping Review
This study aims to review the existing literature with the goal of examining the
potential impact of AI in education. As we enter the era of technology-assisted learning,
it is important to take stock of the potential applications, benefits, and dangers of AI. We
explore the literature to gain insights into various aspects of the problem.
The review procedure begins with an initial search of the Scopus and Google Scholar
databases. We employ appropriate terms and phrases to find the relevant literature. The
results are filtered based on various criteria such as quality, applicability, recentness, and
others. Then, the screened results are categorized into major themes of applications,
benefits, and challenges in AI. The research is further divided into subfields to obtain
a more granular view of the subject. Finally, we delve into each theme and subtheme
discussing in detail the state of affairs. We provide analysis of each subtheme supported
by pertinent literature. The main tasks of the study are the following: (1) perform the
initial search of Scopus, (2) screen the pertinent works, (3) categorize the information, and
(4) provide a detailed discussion of each subtheme.
Methodology
The research conducted for this article took the form of a scoping review, following
the approach described in [12]. Unlike a systematic literature review, a scoping review has
broader research aims [13]. It serves as an effective tool for determining the overall scope
and coverage of the existing literature on a particular topic (in the context of this study, an
investigation of the effect of AI on education by examining its applications, advantages, and
challenges), providing a clear indication of the volume of available literature and studies,
as well as an overview of their focus, whether broad or detailed [14]. The search process
was carried out by two independent researchers between May and June 2023, focusing on
the Scopus and Google Scholar databases for the period of 2019–2023. The search yielded
a total of 107 results. The study then proceeded through five phases: (1) identifying the
research question(s), (2) identifying relevant studies, (3) selecting studies, (4) organizing
data, and (5) summarizing and reporting the results [12]).
In Phase 1, the research question and sub-question that were investigated are as follows:
RQ1: How can AI have an impact on Higher Education?
RQ2: What are the benefits and challenges associated with the use of AI in Higher Education?
Phase 2 involved the identification of relevant studies. To focus on the most recent
research, the database search was limited to the past five years (2019 to June 2023). The
search was conducted using Boolean terms, including “artificial AND intelligence AND
positive AND impact AND higher AND education” (yielding 55 results), “artificial AND
intelligence AND potential AND benefits AND dangers AND higher AND education”
(yielding 0 results), and “artificial AND intelligence AND applications AND advantages
OR disadvantages AND higher AND education” (yielding 52 results). Abstract and full-text
screening were performed by two authors, and the inclusion and exclusion criteria were
established by the three authors in agreement. It was argued in [15] that critical appraisal
and evaluation of article quality for inclusion in a scoping review were deemed “non-
compulsory”. Assessment of quality, reliability, and confidence using GRADE-CERQual,
for instance, was therefore not performed.
During Phase 3 of the study, a comprehensive selection process was conducted to
ensure minimal bias. To maintain consistency, a protocol based on the Preferred Reporting
Items for Systematic Reviews and Meta-Analyses guidelines for scoping reviews (PRISMA-
ScR) was developed [16]. The selection criteria for the included studies were as follows:
(1) written in English, (2) peer-reviewed (articles/book chapters), (3) reports, (4) op-eds,
and (5) published between 2019 and June 2023. Studies were excluded if they (a) were
published in a language other than English (15), (b) were published before 2019 (23), and
(c) lacked full-text availability (4), as shown in Figure 4. A total of 44 articles met the
inclusion criteria (refer to Table 1). To ensure consistency and reliability, Krippendorff’s
alpha coefficient was used to assess inter-rater reliability, resulting in scores of 0.81 for
abstracts and 1.00 for full texts. Data were shared with the third author. Any disagreements
among the three authors regarding study selection were resolved through discussion.
Table 1. Overview of included studies.
Author Year Main Contributions of the Surveyed Work

1 Hleg et al. [17] 2019 Presents ethics guidelines for trustworthy AI
2 Terzopoulos et al. [18] 2019 Explores the capabilities of voice assistants in the classroom
3 Webber et al. [19] 2019 Discusses the potential application of AI to improve teamwork
4 Marcinkowski et al. [20] 2020 Investigates algorithmic vs. human decision-making in HE admissions
5 Ahmed et al. [21] 2021 Investigates an LMS in entrepreneurship
6 Borenstein et al. [22] 2021 Examines the ethical implications of Artificial Intelligence technologies
7 González-Calatayud et al. 2021 Analyzes the use of AI for student assessment
[23]
8 Kamalov et al. [24] 2021 Explores Machine Learning for exam-cheating detection
9 Miao et al. [25] 2021 Discusses AI in Education policy
10 Timan et al. [26] 2021 Discusses data protection in the age of AI
11 UNESCO [27] 2021 Provides recommendations on the ethics of AI
12 JISC [28] 2022 Discusses and reflects on AI in Tertiary Education
13 Kulshreshtha et al. [29] 2022 Explores automatically generated questions as personalized feedback in an ITS
14 Long et al. [30] 2022 Explores Collaborative Knowledge Tracing to predict students’ correctness in
answering questions
15 Mishkin et al. [31] 2022 Presents risks and limitations for DALL-E 2
16 Nguyen et al. [32] 2022 Discusses ethical principles for AI in Education
17 Oxford Insights Govern- 2022 Compares how 160 governments are prepared to use AI in public services
ment AI Readiness Index
[33]
18 Qadir et al. [34] 2022 Discusses advantages and drawbacks on Generative AI in education
19 St-Hilaire et al. [35] 2022 Presents the results of a comparative study on learning outcomes for two
popular online learning platforms
20 Swiecki et al. [36] 2022 Discusses Generative AI and assessment practices
21 Wahle et al. [37] 2022 Explores detection of machine-paraphrased plagiarism
22 AlAfnan et al. [38] 2023 Explores advancements in Artificial Intelligence and its applications
23 Bouschery et al. [39] 2023 Focuses on product innovation management research and strategies
24 Chan et al. [40] 2023 Presents a preprint discussing specific topics related to teachers, AI and Higher
Education
25 Chen et al. [41] 2023 Investigates the use of chatbots in classrooms
26 Chetouani et al. [42] 2023 Examines human-centered AI, human-centered machine learning, ethics, law,
and the societal aspects of AI
Table 1. Cont.
Author Year Main Contributions of the Surveyed Work

27 Cotton et al. [43] 2023 Examines innovations in Education and Teaching
28 Dai et al. [44] 2023 Presents a preprint discussing specific research topics related to education and
technology
29 Dwivedi et al. [45] 2023 Discusses opportunities as well as ethical and legal challenges of Generative AI
30 Elkins et al. [46] 2023 Explores useability of educational questions generated by LLMs
31 European Schoolnet [47] 2023 Explores ethical use of digitally processed data for student learning
32 Hu et al. [48] 2023 Explores adaptive assessments with Intelligent Tutors
33 Liu et al. [49] 2023 Discusses AI and its applications in Education
34 Lodge et al. [50] 2023 Discusses the use of Generative AI in tertiary education
35 Liu et al. [51] 2023 Presents initial results of a survey on the use if Generative AI at university
36 Malmström et al. [52] 2023 Discusses the use of ChatGPT in HE
37 Perkins et al. [53] 2023 Discusses academic integrity of LLMs
38 Rasul et al. [54] 2023 Presents benefits and challenges of ChatGPT in HE
39 Rudolph et al. [9] 2023 Discusses ChatGPT and assessment
40 Sabzalieva et al. [55] 2023 Provides an overview of how ChatGPT works and explains how it can be used
in HE
41 Sullivan et al. [56] 2023 Discusses ChatGPT, academic integrity and student learning
42 UAE [57] 2023 Provides a comprehensive guide on the utilization of Generative AI applica-
tions
43 Walton Family Founda- 2023 Discusses teachers and students’ adoption of Generative AI
tion [58]
44 Wylie et al. [59] 2023 Explores the uses of Generative AI in Business Schools
Figure 4. Overview of literature search process using PRISMA-ScR.
4. Results
Phase 4: Data from the 44 eligible studies were charted (Table 1). Thematic analysis [60]
was used to identify themes.
Phase 5: Organize and Summarize the Results. Four overarching themes emerged
from the data: (1) Personalized Learning, (2) Intelligent Tutoring Systems (ITS), (3) Assess-
ment Automation, and (4) Teacher–Student Collaboration. To answer RQ 2, benefits and
challenges of each of the four themes were identified, as shown in Tables 2–5.
Table 2. Overarching themes unpacked: Personalized Learning.
Benefits Challenges
• Customized Learning Experience: Personalized • Resource-Intensive: Implementing personalized

learning allows students to progress through the cur- learning often requires significant investments in
riculum at their own pace, ensuring that they fully technology, infrastructure, and continuing profes-
understand a topic before moving on to the next sional development. Not all institutions may have
one. This customized approach can lead to better the resources to adopt personalized learning fully.
learning outcomes. • Faculty development and Support: Teachers play a
• Improved Student Engagement: When students can central role in personalized learning environments.
learn in a way that aligns with their interests, they are They need training and ongoing support to effec-
more likely to stay engaged and motivated, leading tively implement this approach, create personalized
to a deeper grasp of the subject matter. learning plans, and manage diverse student needs.
• Addressing Diverse Learning Needs: Classrooms of- • Curriculum and Content Adaptation: Personalized
ten consist of students with varying levels of knowl- learning requires a curriculum that can be easily
edge and abilities. Personalized learning helps ad- adapted to individual learners. Developing or cu-
dress the diverse learning needs of students, pro- rating such flexible content can, however, be chal-
viding additional support to those who need it and lenging.
allowing advanced learners to explore more challeng- • Data Privacy and Security: Personalized learning
ing material. relies on collecting and analyzing student data. En-
• Self-directed Learning: Personalized learning encour- suring the privacy and security of these data is cru-
ages students to take ownership of their learning and cial to protect students’ information from potential
become more self-directed learners. They learn how breaches or misuse.
to set goals, monitor their progress, and make deci- • Technical Issues: The use of technology in personal-
sions about their learning path(s). ized learning can be susceptible to technical glitches
• Data-Driven Instruction: Personalized learning re- and interruptions. Reliability and seamless integra-
lies on data and analytics to understand students’ tion of technology are essential for a smooth person-
strengths and weaknesses. Academics can use these alized learning experience.
data to make informed decisions about instructional • Balancing Structure and Freedom: Some students
strategies and interventions. may struggle with the freedom to choose their learn-
• Flexibility and Accessibility: Personalized learning ing path and may need academic advising.
can be implemented in various settings and modal- • Assessment and Accountability: Traditional forms
ities, including classrooms, online platforms, and of standardized assessment may not align well with
blended learning environments. It also allows for personalized learning approaches. Finding effective
greater accessibility for students with special needs, ways to assess (summative and formative) and mea-
refugee contexts, or those in remote areas. sure student progress in personalized settings can be
• Lifelong Learning Skills: By engaging in person- rather complex.
alized learning experiences, students can develop • Equity Concerns: Ensuring equitable access to per-
essential skills and graduate outcomes such as sonalized learning opportunities is essential. Some
critical thinking, problem solving, and indepen- students may face barriers to access, such as lim-
dent learning, which are valuable beyond their ited internet connectivity or access to devices, which
academic journey. could exacerbate existing achievement gaps.
Table 3. Overarching themes unpacked: Intelligent Tutoring Systems.
Benefits Challenges
• Enhanced Learning Outcomes: ITS can provide per- • Development and Implementation: Designing and
sonalized and adaptive instruction, tailored to indi- developing effective ITS requires significant time,
vidual student needs, which can lead to improved training, resources, and expertise in both pedagogy
learning outcomes and academic performance. and technology.
• Individualized Learning: ITS can offer personalized • Data Privacy and Security: ITS collect and analyze
feedback and guidance to students, allowing them to large amounts of student data, raising concerns about
progress at their own pace and focus on areas where data privacy, security, and the ethical use of personal
they need additional support. information.
• Continuous Assessment: ITS can provide real-time • Bias and Discrimination: If not properly designed
assessment and feedback, enabling students to mon- and trained, ITS can perpetuate biases and discrimi-
itor their progress and identify areas for improve- nation, as the underlying algorithms may replicate
ment. biases present in the training data.
• Immediate Feedback: ITS can offer immediate feed- • Technical Limitations: ITS may face limitations in
back on students’ responses, allowing them to correct accurately understanding and interpreting students’
their mistakes and reinforce their understanding in responses, especially in complex or ambiguous sit-
real time. uations, leading to potential gaps in instructional
• Access to Quality Education: ITS can provide access support.
to quality education in remote or underserved ar- • Teacher–Student Relationship: The use of ITS may
eas, reaching students who may not have access to impact the traditional teacher–student relationship,
traditional educational resources. as students may rely more on the system for in-
• Cost and Time Efficiency: ITS can reduce the cost struction and guidance, potentially reducing inter-
and time associated with one-on-one tutoring by au- personal interactions.
tomating certain aspects of instruction and support. • User Acceptance and Engagement: ITS may face re-
• Multimodal Learning: ITS can incorporate various sistance or low user acceptance from students and
forms of multimedia, such as videos, interactive sim- academics, who may prefer traditional instructional
ulations, and virtual environments, to engage stu- methods or perceive the system as impersonal or less
dents and enhance their learning experience. effective.
• Long-term Knowledge Retention: ITS can employ • Lack of Adaptability: Some ITS may struggle to
spaced repetition and other cognitive techniques adapt to individual learning styles, preferences, and
(Cognitive Load Theory) to promote long-term re- cultural differences, potentially limiting their effec-
tention of knowledge and skills. tiveness across diverse student populations.
• Integration with Existing Systems: Integrating
ITS with existing educational technologies, infras-
tructure, and curricula can present technical and
logistical challenges, requiring careful planning
and coordination.
Table 4. Overarching themes unpacked: Assessment Automation.
Benefits Challenges
• Time Efficiency: Automation can significantly reduce • Limited Applicability: Not all types of assessments
the time and effort required for grading and evaluat- can be easily automated. While diagnostic assess-
ing assessments. Computer-based grading systems ment, multiple-choice questions, and some struc-
can quickly score multiple-choice questions, fill-in- tured responses lend themselves well to automation,
the-blank responses, and even some types of open- subjective or complex assessments requiring human
ended questions, saving academics valuable time. judgment, such as essays or project evaluations, may
• Consistency and Reliability: Automated grading en- be challenging to automate fully.
sures consistent and objective evaluation, eliminating • Adaptability: Automated assessment systems may
potential bias or human error. Each student’s assess- struggle with adapting to unique or creative student
ment is evaluated against the same criteria, promot- responses that deviate from predefined answer pat-
ing fairness and accuracy in the grading process. terns/rubrics. They may not be able to recognize
• Faster Feedback: Automation allows for faster feed- innovative or unconventional thinking, limiting the
back delivery to students. Instead of waiting for scope of assessment.
manual grading, students can receive instant feed- • Learning Outcomes Assessment: Some learning out-
back on their assessments, enabling them to iden- comes, such as critical thinking, creativity, and prob-
tify areas of improvement and adjust their learning lem solving, are difficult to assess through automated
strategies promptly. systems alone. These higher-order skills often require
• Scalability: Automation enables the grading of a human judgment and qualitative evaluation, which
large number of assessments efficiently. It is par- automated assessments may not capture adequately.
ticularly beneficial for online courses, MOOCs, or • Technical Limitations: Assessment automation relies
programs with a high volume of assessments, as it on technology, and technical issues can arise, such
can handle large student populations without sacri- as system errors, software glitches, or compatibility
ficing the quality of evaluation. problems. These technical limitations can disrupt
• Analytics and Insights: Automated assessment sys- the assessment process and impact its reliability and
tems can generate data and analytics on student per- validity.
formance, allowing academics and institutions to • Lack of Contextual Understanding: Automated sys-
gain insights into student learning patterns, iden- tems may struggle to understand the context or nu-
tify common misconceptions, and make data-driven ances of student responses, leading to potential mis-
instructional decisions. interpretation or incomplete evaluation. They may
• Personalization: Automated assessment tools can not grasp the underlying reasoning or provide tar-
be designed to provide personalized feedback and geted feedback to address individual student needs
recommendations based on individual student per- effectively.
formance. This customization helps students under- • Teacher–Student and Student–Student Interaction(s):
stand their strengths and weaknesses, guiding them Automated assessments, particularly those lack-
toward targeted learning activities. ing human involvement, can limit opportunities
• Standardization: Automated assessments can be for meaningful interaction(s) and dialogue between
aligned with predetermined program/course learn- teachers and students and students with other stu-
ing outcomes, ensuring that students are evaluated dents (peer assessment). Personalized feedback and
consistently against specific educational standards. guidance may be absent, impacting the overall learn-
This standardization helps maintain the quality and ing experience.
integrity of assessments and helps ensure construc- • Ethical Considerations: Assessments that involve
tive alignment. sensitive or personal information require careful han-
dling and protection of student data. The automation
of assessments raises ethical concerns related to data
privacy, security, and potential biases in algorithms
or automated decision-making processes.
Table 5. Overarching themes unpacked: Teacher–Student Collaboration.
Benefits Challenges
• Personalized Learning: AI tools can analyze student • Lack of Human Interaction: AI tools, while beneficial,
data and provide personalized recommendations cannot fully replace human interaction and the value
and resources based on individual learning needs. of face-to-face communication between teachers and
This personalization helps academics tailor their in- students. Over-reliance on AI tools may result in
struction to each student, fostering a more effective reduced opportunities for meaningful interactions,
and targeted learning experience. personal connections, sense of belonging, and emo-
• Real-time Feedback: AI tools can provide instant tional support.
feedback to students on their work, allowing them to • Technical Issues and Reliability: AI tools rely on tech-
identify mistakes, misconceptions, or areas needing nology, and technical glitches or system failures can
improvement promptly. This immediate construc- disrupt the collaboration process. Reliability and sta-
tive feedback helps students make adjustments in bility issues may affect the trust and confidence in
their learning strategies and promotes continuous AI tools, potentially impacting their adoption and
learning and growth. effectiveness.
• Enhanced Communication: AI tools facilitate com- • Data Privacy and Security: AI tools often require
munication and collaboration between teachers and the collection and analysis of student data, raising
students through various channels. These tools can concerns about privacy and security. Proper pro-
support online discussions, virtual classrooms, and tocols and safeguards must be in place to protect
interactive platforms, enabling seamless interaction sensitive student information and ensure compliance
and engagement regardless of physical location, for with data protection regulations.
both the academic and the student. • Bias and Fairness: AI tools are only as unbiased as
• Resource Accessibility: AI-powered platforms can of- the algorithms and data they are trained on. If the un-
fer a wide range of educational resources, including derlying data or algorithms have biases, it can lead to
e-books, multimedia materials, and interactive sim- unfair outcomes or perpetuate existing inequalities.
ulations. These tools provide students with access It is crucial to regularly evaluate and mitigate biases
to diverse learning materials and enable teachers to to ensure fairness and equity in the use of AI tools.
share resources easily, expanding the learning oppor- • Skills and Training: Effective utilization of AI tools
tunities beyond traditional classroom boundaries. requires teachers and students to have the necessary
• Data Analysis and Insights: AI tools can analyze skills and training. Academics need support and pro-
large volumes of student data, such as performance, fessional development opportunities to understand
behavior, and navigational/engagement patterns. and leverage AI tools effectively. Similarly, students
This data analysis provides valuable insights to aca- need guidance on how to use AI tools appropriately
demics, enabling them to track student progress, and critically evaluate the information provided.
identify trends, and make data-driven instructional • Ethical Considerations: The use of AI tools raises eth-
decisions for individual students and/or the entire ical considerations, such as the responsible use of stu-
class. dent data, transparency in algorithms and decision-
• Time Efficiency: AI tools can automate routine ad- making processes, and ensuring that AI tools do not
ministrative tasks, such as grading assignments or or- compromise student well-being or autonomy.
ganizing schedules, allowing academics to dedicate
more time to instruction and personalized support.
This time efficiency frees up academics’ workload
and enables them to focus on higher-value teaching
and learning activities.
The papers suggest that AI has the potential to positively impact Higher Education.
As was argued in [61], for instance, that AI, big data, and learning analytics can become a
powerful tool for advancing higher education institutions further, at the same time, AI can
have a detrimental effect without a vigilant eye. However, the papers also acknowledge
the challenges and potential risks of introducing AI in educational settings, such as the
need to pay attention to the ethical side of AI.
Our review reveals that the recent popularity of AI in education (Figure 5) is driven
by four main applications: (1) Personalized Learning, (2) Intelligent Tutoring Systems,
(3) Assessment Automation, and (4) Teacher–Student Collaboration.
100
80
Google Trends
60
40
20
-15
-24
-02
-11
-19
-30
-05
-07
-10
-12
-02
-04
22
22
22
22
23
23
20
20
20
20
20
20
Figure 5. Popularity worldwide of the search query “AI in education” [62].
4.1. Theme 1: Personalized Learning

Personalized learning powered by AI involves tailoring educational content and
experiences to cater to each student’s unique needs, learning styles, and preferences. This
approach enables students to learn at their own pace, thereby enhancing their engagement
and overall learning outcomes [63]. The individualized approach benefits both students
and teachers, leading to better learning results and overall educational experiences. AI has
the ability to help tailor education in a variety of ways. Adaptive learning systems driven
by AI may analyze students’ performance, strengths, and shortcomings in order to offer
tailored learning courses. AI algorithms and adaptive learning systems can analyze student
data, recognize patterns, and suggest personalized content and resources to optimize their
learning experience. Based on each student’s achievement, these platforms may change
the speed, material, and complexity of the curriculum, to deliver an optimal learning
experience. AI can power virtual tutors who provide students with one-on-one education
customized to their unique learning and emotional needs [64,65]. Automated tutoring
systems can provide instant feedback, answer queries, and walk pupils through complex
ideas, supplementing or even replacing traditional tutoring services.
AI has the potential to analyze massive volumes of data created by students during the
learning process, revealing patterns and trends that might assist instructors in identifying
areas where students require additional support or resources. It enables instructors to
develop individualized learning experiences for the students based on data. The power
of AI can help improve learning experiences by including gamification features such as
incentives, challenges, and competition. Engaging and individualized learning environ-
ments can be developed that increase motivation and active engagement by adapting these
components to individual students. AI has potential to help students with special needs or
impairments receive tailored education by designing assistive technologies that meet their
specific demands. Speech recognition and text-to-speech technologies can be employed to
develop custom content that fits the individual needs of affected students.
4.2. Theme 2: Intelligent Tutoring Systems

AI technology has the potential to accelerate intelligent tutoring systems (ITS), which
are aimed at providing students with tailored, one-on-one teaching, emulating the experi-
ence of learning from a human tutor (Figure 6). To understand students’ learning needs and
customize their teaching methods, ITS employs powerful algorithms and machine learning
approaches. Natural language processing (NLP) enables AI to perceive and interpret writ-
ten or spoken input from students, allowing ITS to engage in meaningful dialogues, answer
questions, and provide instruction on a variety of subjects. ITS employs AI to provide
customized instruction and feedback to students, thus bridging the gap between traditional
classroom learning and individualized tutoring [66].
Figure 6. Intelligent humanoid leading a tutoring session (DALL·E 2 image).
Student modeling is another critical component of ITS. It is a process in which AI

generates dynamic models of students’ knowledge, skills, and learning preferences based
on their interactions with the system. These models assist ITS in understanding students’
present comprehension of the material, misconceptions, and areas where they may re-
quire additional help. These systems analyze students’ learning behaviors, track their
progress, and offer personalized guidance, contributing to improved learning outcomes
and increased student motivation [67]. AI may develop ITS that customizes education
to individual learners and successfully addresses their specific needs by combining NLP,
student modeling, and adaptive content delivery.
4.3. Theme 3: Assessment Automation

Automated student assessment is one of the major areas of application. AI has the
potential to significantly improve student assessment by automating it, expediting the
evaluation process, and providing fast, individualized feedback [68]. It can revolutionize
the way assessments are conducted by providing real-time feedback to students [23]. AI can
examine students’ work, from simple multiple-choice questions to sophisticated written
assignments and even spoken responses, by employing machine learning techniques and
NLP. AI-driven assessment tools can evaluate complex tasks, such as essays and problem-
solving exercises, reducing the burden on educators while improving the accuracy and
consistency of grading [69]. This feature enables more objective evaluation, eliminating
human bias and errors while allowing educators to focus on other areas of teaching, such as
curriculum creation and student engagement. The technology has fewer time and resource
constraints compared to a human educator, which allows AI to provide more in depth
feedback on student work. Since learning from mistakes is one of the most effective ways
of learning, high-quality feedback is important in mastering the material.
Automated AI assessment tools can detect patterns in students’ performance, identify-
ing specific areas where they may require additional assistance or resources. It helps instruc-
tors to make data-driven decisions and build personalized interventions that suit individual
learning needs by offering real-time feedback and deep insights into students’ strengths and
weaknesses. This level of personalization and adaptation in evaluation can help students
achieve better learning outcomes and have a more engaging educational experience.
4.4. Theme 4: Teacher–Student Collaboration

The collaboration between teachers and students can be facilitated by AI, enhancing
the overall learning experience. By providing real-time analytics and insights, AI can
help educators identify students’ strengths, weaknesses, and learning patterns, allowing
them to adjust their teaching strategies accordingly [70]. In situ assessments and instant
feedback allow teachers to make real-time adjustments during the class. AI can be used to
notify teachers when students, and which students, are struggling while providing possible
remedies. As a brainstorming partner, AI can help identify effective solutions to support
student learning.
AI technology can help teachers to answer a variety of student questions in class. Since
human teachers have a limited amount of knowledge, they may be challenged by unex-
pected and out-of-box questions from the students. AI can help fill the gap in knowledge
and provide high-quality responses to student questions. AI-powered chatbots can address
students’ queries and facilitate peer-to-peer interactions, promoting a collaborative learning
environment [71]. The AI-assisted approach has been particularly popular in English as a
second language (ESL) courses [72,73].
Voice-activated technology that uses speech recognition technology is another area of
advancement that will help improve teacher–student collaboration.
5. Advantages of AI in Education
AI technology holds tremendous potential for enhancing the quality of education at
all levels of study. The key advantages of AI can be summarized as follows: (1) enhanced
learning outcomes, (2) time and cost efficiency, and (3) global access to quality education.
5.1. Enhanced Learning Outcomes

AI in education has the capacity to significantly improve student learning outcomes
by providing tailored learning experiences with the help of knowledge tracing (KT) and
collaborative filtering. AI technology can be used to customize learning experiences to
match and predict specific student needs (and future performance) by (a) identifying in-
dividual strengths and limitations, (b) tracing knowledge states [30], and (c) analyzing
past interactions and navigational patterns. As a result, more engaged and motivated
students, increased information retention, and, ultimately, enhanced academic success may
be achieved. Patterns and trends in student data (actionable intelligence) can be identified
with the assistance of AI to help educators identify and address any learning gaps or
challenges in real time. A more refined and granular adaptive exercise recommendation
based on (students’) cognitive level and collaborative filtering with an 8 percent improve-
ment in recommendation effectiveness was proposed in [49]. The potential of intelligent
tutoring systems to enhance learning outcomes by delivering tailored instruction and
just-in-time feedback to students [29] was recognized as early as 1990 [74]. More recently,
several studies found that by catering to individual learning styles and preferences, AI can
increase student engagement, motivation, and retention [75,76] with the use, for instance,
of personalized feedback systems based on deep learning-based Transformer models [29]
or Intelligent Tutoring Systems/AI-powered Tutoring (e.g., Chrome’s Galileo or Khan-
migo). A recent study [35] indicated, for instance, that learning outcomes were “better for
participants on Korbit”, a Canadian AI-powered learning platform, which uses machine
learning models to adapt the learning process to students, “than participants on either of
the platforms that” did “not provide personalized feedback”. Learning outcomes can also
be improved by automating often time-consuming administrative tasks like assessment,
grading, and teaching and learning activity planning, allowing educators to focus on direct
student interaction and effective teaching strategies. Existing technology-driven instruction
tools [77,78] can also be further enhanced with the power of AI.
AI tools and resources, such as virtual tutors, voice assistants [18], text-to-image gen-
eration (Stable Diffusion, Dall-e-2, Midjourney), text summarization (summarizer.org), AI
video generation platforms (Synthesia or Elai.io), AI-enabled adaptive assessment [48], and
smart content, may also supplement traditional learning techniques by providing students
with additional material and interactive experiences that improve their understanding
of complex concepts [34,40]. The Joint Information Systems Committee (JISC) interactive
map shows how UK tertiary institutions use AI technologies to improve processes. The
ChatGPT quick start guide by Sabzalieva and Valentini [55] and the United Arab Emirates
100 practical applications and uses of generative AI [57] provide useful and valuable exam-
ples of how AI can be used to improve productivity and facilitate student collaboration by
promoting teamwork, team effectiveness [19], and problem-solving skills, all of which are
required to succeed in today’s workforce. Teamwork during the class is a commonly used
teaching strategy that helps students to work together and learn from one another. Since
teachers can attend to only one team at a time, AI can be employed to assist and lead the
discussion in each group, in a “hybrid team”, according to intent and topics (intelligent
triage). AI can help steer teamwork in the right direction. The researchers in [39] explored
how GPT-3 could act as an innovator in a hybrid (augmented) team in the new product
development process. In general, incorporating artificial intelligence into education has
the potential to create a more engaging, efficient, and effective learning environment that
benefits both students and educators.
5.2. Time and Cost Efficiency

One of the key advantages of AI is the automation of manual tasks [36]. The intro-
duction of AI in education has resulted in a paradigm shift, with potential to dramatically
improve time and cost efficiency for both students and educators. The automation of
teachers’ duties will provide time savings that would allow greater focus of personalized
learning (Figure 7) [79].
AI can streamline various educational processes such as generating questions using
prompts [46], grading, and content creation, reducing the workload on educators and
enabling them to focus on higher-order tasks including thinking and problem solving.
Specifically, AI can be trained to grade a range of subjects and assessments including
essays, problem-solving questions, and even graphing assessments [23]. Since grading
consumes up to half of educators’ time, the potential time savings are enormous. In
addition, the instant grading provided on assessments ensures that students receive timely
and constructive feedback on their performance [44,50]. AI-Generated Lesson Plan creation
is another avenue for improved efficiency [21]. AI can be used to create lecture slides,
custom images and videos, homework and exam questions, and other course content,
allowing academics to focus on human-centric tasks. AI tools can provide cost-effective
solutions for institutions, particularly in resource-constrained settings, by offering scalable
and accessible learning opportunities [80].
Figure 7. Top barriers to providing personalized learning, % of teachers identifying area as a

primary barrier.
AI has the potential to alter educational institution budget allocations by reducing

reliance on traditional, resource-intensive teaching methods in terms of cost efficiency.
With the incorporation of AI into the educational scene, schools can save money on hiring
many teachers and other staff while still providing a high-quality education to their pupils.
AI-based educational resources such as Massive Open Online Courses (MOOCs) and digital
textbooks can be delivered at a fraction of the cost of their traditional counterparts. As
a result, these low-cost frugal solutions and innovations improve access to education,
especially for students in low-resource settings or in remote regions [81], democratizing
information and contributing to the wider educational setting. In the new paradigm, the
role of the educator will be to oversee the work of AI. Quality control and fine-tuning the
AI will become key responsibilities of a teacher. While AI can be utilized to perform various
tasks, educators must approve the results.
5.3. Global Access to Quality Education

The lack of access to quality education is one of the key issues in many developing nations [82].
It is a sentiment that is expressed by the leading experts in international education [47].
Since education is an important factor in driving economic growth, providing universal
education has huge financial benefits to both individuals and countries. The emergence of
AI has paved the way for significant breakthroughs in worldwide access to high-quality
education. As a result, AI-based education has recently attracted a significant amount of
interest from around the globe (Figure 8) [62]. AI-powered educational tools and materials
can cross geographical, socioeconomic, and linguistic boundaries, allowing for a fairer
distribution of information. In this regard, it is important to consider the technological
challenges of delivering AI-based education to remote locations. Thus, the next-generation
communication networks, such as 6G, play a crucial role in enabling the above vision [83].
Figure 8. Popularity of the search query “AI in education” by region.
One of the key advantages of AI is its relatively low cost. While the initial fixed cost
of building an AI system can be significant, as in the case of GPT-4, which was estimated
to cost USD 100 million to train, the variable cost is almost zero. In other words, AI
systems can scale at low cost. AI has the potential to democratize education by providing
access to high-quality learning resources and personalized instruction across geographic
and socioeconomic barriers [84]. Online AI-driven platforms and chatbots can bridge the
gap between students and quality education, helping to reduce inequalities and create
a more inclusive learning environment [85]. Since English presentation skills are still a
major barrier for some participants, ChatGPT can help tell a more convincing story when
pitching their ideas in front of the jury, as well as written reports. It means an equal
playing field for participants from all backgrounds [59]. By adapting educational content
and pedagogical approaches to meet the needs and abilities of different communities,
AI-powered adaptive learning systems can improve students’ engagement, motivation, and
academic achievement, making education more effective and accessible around the world.
6. Challenges and Ethical Considerations

While AI has potential to revolutionize the education system, there are several concerns
over its dangers [86]. As shown in Figure 9, the number of ethics incidents has continued
to grow rapidly in line with the rise of AI [87]. Several AI ethics initiatives have emerged
recently [88,89]. However, very few schemes currently exist for the specific issues raised
by AI in education [86], except maybe the European Commission Ethics guidelines for
trustworthy AI [17] and the forthcoming UNESCO IESALC Manual on AI in Higher
Education. It is important to move carefully in the adoption of AI in schools and universities.
There is a real danger of AI becoming pervasive in every sense where those involved may
be exposed to risks without being aware of them [32,90]. Although governments have
recognized the importance of AI for future development, there is lack of comprehensive
policy or guidance around AI in education [25,42,91].
Figure 9. Number of ethics incidents related to AI.
Proper testing and evaluations are needed to ensure the safety of the technology. The
use of AI with young children should be particularly investigated. The potential issues
related to AI include the following: (1) data privacy and security, (2) bias and discrimination,
(3) plagiarism and academic integrity, and (4) the teacher–student relationship.
6.1. Data Privacy and Security

Data privacy and security concerns affect almost every technology. User personal data,
communications, and location details are vulnerable to hacking or monitoring by govern-
ments. These concerns also exist in the applications of AI within the educational domain.
As AI tools penetrate educational institutions, they collect and process vast quantities of
sensitive information, including students’ personal data, academic records, and behav-
ioral patterns. Consequently, the potential for misuse or unauthorized access to these data
presents significant risks to stakeholders [26]. The violation of individual privacy rights and
the potential for discriminatory practices are among key dangers. In addition, inadequate
data security measures may lead to breaches or leaks, further exacerbating these concerns
and undermining trust in AI-based educational systems. The implementation of AI in edu-
cation raises concerns regarding data privacy and security, as these systems often rely on
large volumes of student data for analysis and personalization [25,92]. The challenge lies in
striking a delicate balance between harnessing AI’s potential to revolutionize education and
ensuring the protection of data privacy and security. Various solutions have been proposed,
including the implementation of robust data protection policies, encryption techniques, and
stringent access controls. Ensuring the protection of sensitive information and complying
with data privacy regulations, such as the General Data Protection Regulation (GDPR), is
crucial to maintain trust and prevent potential misuse [93]. Moreover, fostering a culture of
privacy awareness within educational institutions is critical to ensure compliance with data
protection laws and regulations. It requires the development of comprehensive guidelines
for educators and AI developers that outline best practices and promote transparency,
accountability, and ethical use of AI in education. Ultimately, addressing data privacy and
security issues in the context of AI-enabled education is fundamental to ensuring that the
benefits of these technologies are fully realized while minimizing the associated risks.
6.2. Bias and Discrimination

The use of AI in education has shown great promise in enhancing teaching and learn-
ing experiences. However, the issues of bias [45] and discrimination present a significant
challenge to the effective and equitable utilization of AI in education. Bias in AI systems is
mainly derived from the data they are trained on, which is frequently influenced by histori-
cal and societal biases. Large language models such as GPT and Bard are trained on huge
amounts of publicly available data from the internet, which contain different points of view.
These biases, when embedded in educational AI applications, can lead to a perpetuation
of existing disparities, and even exacerbate them, ultimately impacting students’ learning
outcomes, opportunities, and access to resources. AI algorithms can inadvertently perpetu-
ate biases present in the data they are trained on, leading to discriminatory outcomes [94].
Additionally, the use of algorithmic decision making (ADM) by tertiary institutions to
admit students to programs raises issues of fairness and objectivity [20]. It is therefore
essential to address these biases and develop fair and equitable AI systems that promote
diversity and inclusion in the learning environment [95].
Discrimination in AI-based educational tools arises when these systems inadvertently
disadvantage specific groups of students based on factors such as race, gender, or socioeco-
nomic background. In content creation, AI-generated material may contain cultural biases,
acknowledged by Open AI, in its initial risk analysis and mitigations document for the
DALL·E 2 system, where it expresses the following: “DALL·E 2 tends to serve completions
that suggest stereotypes, including race and gender stereotypes. For example, the prompt
“lawyer” results disproportionately in images of people who are White-passing and male-
passing in Western dress, while the prompt “nurse” tends to result in images of people who
are female-passing” [31]. Since a vast majority of the information that exists on the internet
has been produced in the Global North, the AI models trained on these data will be molded
in the same fashion. Thus, to mitigate the AI divide highlighted in the Oxford Insights
Government AI Readiness Index [33], the use of AI in non-Western educational institutions
requires further tuning and customization according to the regional and local cultural
norms. In assessment evaluation, AI tools might exhibit bias towards certain linguistic
or cultural backgrounds, consequently affecting the performance evaluation of students
from diverse backgrounds. Additionally, adaptive learning systems, which tailor learning
experiences to individual students’ needs, may inadvertently favor students who exhibit
learning patterns similar to those of the majority in the training data, thereby (1) widening
the achievement gap and (2) potentially exacerbating inequalities. As pointed out in [22],
the root of the problem is with people, even though there is a need to fix the bias in the
data, within the algorithms, and in the outputs. Addressing these issues therefore requires
greater collaboration between software engineers, educators, and policymakers to develop
transparent, accountable, and inclusive systems to ensure responsible deployment of AI
in education.
6.3. Plagiarism and Academic Integrity

AI plagiarism has been one of the hottest topics of debate since the introduction of
AI chatbots and virtual assistants, such as ChatGPT, Bard, YouChat, Hubspot ChatSpot,
Bing Chat, or Vicuna. Whilst chatbots can be responsive pedagogical tools [41], educators
have raised the alarm about students using chatbots to compose essays, write computer
code, and complete other homework assignments [56]. The convenience and effectiveness
of AI chatbots are tantalizing for students. A survey of 5894 students from across Swedish
universities about their use of and attitudes towards AI for learning purposes, both about
chatbots (such ChatGPT) and other AI language tools (such as Grammarly), indicated
that 95% of students were familiar with ChatGPT; 56% were positive about using chatbots
in their studies; 35% use ChatGTP regularly; 60% were opposed to a ban on chatbots,
and 77% were against a ban on other AI tools (such as Grammarly) in education [52].
As online education is growing increasingly popular, an environment where the use of
chatbots is particularly easy, the issues of (a) AI plagiarism and (b) detecting academic
misconduct with the use of artificial intelligence [24,43] have become more pressing [96,97].
The new technology is “a clear threat to academic integrity for HEIs, requiring a range of
adjustments to be made in both practice and policy” [53]. Thus, it is imperative to identify
and implement solutions that address the issue of AI plagiarism.
Several studies have already considered the effectiveness of Generative AI (ChatGPT,
davinci-003) in completing student assignments [38,45]. Researchers have compared the
quality of authentic student work to that of AI. In [98], the researchers considered the
performance of ChatGPT (version GPT-3.5) on the final exam of a typical MBA core course,
Operations Management. They found that it does an excellent job at basic operations
management and process analysis questions, including those that are based on case studies.
On the other hand, it makes surprising mistakes in relatively simple math calculations.
Another research direction has been the study of anti-plagiarism software in detecting AI
work [37]. Most of the currently available anti-plagiarism software is aimed at detecting
plagiarism from existing literature [99]. Since AI chatbots often produce original content,
the off-the-shelves anti-plagiarism software is not well-suited to detect AI plagiarized
content [100]. Given the inefficiencies of the traditional software to detect plagiarized
content, several attempts have been made to create software that is aimed specifically at
identifying AI-generated text. A popular AI plagiarism tool, GPTZero, utilizes perplexity as
a measure of the complexity of text, and burstiness—variations of the sentences—to detect
AI-generated text. The software creators claim a 95% detection rate of AI-generated text.
Other options include Turnitin Feedback Studio (TFS with Originality, Turnitin Originality,
Turnitin Similarity, Simcheck, Originality Check, and Originality Check+), with a 98%
detection confidence rate.
6.4. Teacher–Student Relationship

The arrival of AI in education has potential to produce significant transformations
in the teacher–student relationship [54]. While AI technology offers promising ability for
personalized learning experiences, efficient assessment, and adaptive feedback, it also
presents novel challenges to the dynamics of the traditional educational bond [9]. As the
technology continues to permeate the educational sphere, concerns arise regarding the
potential dilution of human connection, as well as the erosion of teacher–student rapport,
with the consequent impact on the development of socio-emotional skills and a shared
sense of classroom community.
The increased reliance on AI will invariably affect the interpersonal dynamics between
teachers and students. The introduction of AI will undoubtedly alter the traditional role of
teachers and affect the relationship between the educators and students [101]. A survey
of 1002 K—12 teachers and 1000 students between 12 and 17 in the United States in
2023, commissioned by the Walton Family Foundation, indicated that 51% of teachers
had used ChatGPT (10% almost every day). Three in ten teachers had used it for lesson
planning (30%), coming up with creative ideas for classes (30%), and building background
knowledge for lessons and classes (27%); 73% indicated that they thought the tool can help
students learn more. Interestingly, and in terms of continuing professional development,
three-quarters of teachers said using ChatGPT could help them grow as teachers (77%) [58].
The consequences of the new teacher–student paradigm can be unexpected. As AI

becomes more embedded in content creation and assessment evaluation, one potential
outcome could be the decline in teachers’ authority, as students no longer consider them as
the ultimate authority. Ensuring that the integration of AI does not undermine the human
connection and emotional support provided by educators is essential for the holistic devel-
opment of students [102]. It has been argued that AI has potential to engender dramatic
and profound changes in teaching and assessment of students, forming an indispensable
component of a new paradigm within which we design and deliver education [103]. The
potential issues highlighted in this section require a critical examination of the potential
consequences of AI integration in the educational domain, as well as a commitment to
the establishment of guidelines and protocols that safeguard the rights and well-being
of students. The delicate balancing act between leveraging AI’s potential benefits and
addressing the associated challenges underscores the need for a multidisciplinary approach
in the ongoing discourse surrounding AI implementation in education.
7. Future Directions and Opportunities

As AI technology continues to advance, it will generate new and unimaginable appli-
cations in education. One of the most exciting future opportunities involves the fusion of
AI and virtual reality to provide learners with visually rich educational content. Another
direction for future application is lifelong learning, where AI is poised to transform the
landscape of continuous education and upskilling, laying the foundation for a more adapt-
able and resilient workforce in the future. On the other hand, as AI permeates multiple
facets of daily life, it is important to educate people about AI literacy. Given the power of
AI, it is essential to be aware of the ethical consideration when using the technology. A
recent example is the launch of the Frontiers of Computing initiative at the University of
Southern California, which aim is to embed digital/AI literacies, ethics, and responsibilities
across all disciplines. JISC’s AI in tertiary Education: A summary of the current state of
play report [28] also provides key legal and regulatory frameworks and guidance on the
ethical uses of AI.
7.1. Augmented and Virtual Reality

Augmented reality (AR), Mixed Reality (MR), and virtual reality (VR, with HoloLens
or Meta Quest 2 headsets) are three of the most exciting new technologies which hold a
tremendous capacity to enrich human life. In the rapidly evolving landscape of education,
the integration of AI with AR, MR, and VR promises to revolutionize learning experiences
and pedagogical approaches. AI-driven AR, MR, and VR technologies have the potential to
create immersive and interactive educational environments (Metaverse), fostering engage-
ment and enhancing students’ cognitive abilities [104]. These advances enable personalized
and adaptive learning pathways, accommodating various learning styles and addressing
individual needs. The development of AI-powered AR/VR applications can also provide
students with access to simulations and virtual environments, enabling them to acquire
practical skills and knowledge in various disciplines [105]. However, the widespread
adoption of AI, AR, MR, and VR technologies in education raises numerous ethical and
practical concerns that warrant further investigation.
The potential for exacerbating socio-economic disparities and widening the digital
divide must be considered, as access to these advanced educational tools may be limited
to those who can afford them. In addition, while AI-driven AR and VR applications
have demonstrated promising outcomes in various educational contexts, the long-term
efficacy and impact on learning outcomes remain to be thoroughly assessed. As a result,
future research should focus on addressing these challenges, ensuring the responsible and
equitable integration of AI-driven AR and VR technologies into educational ecosystems.
7.2. Lifelong Learning and Skill Development

In today’s fast-changing job landscape, lifelong learning and skill development have
emerged as critical aspects of modern education. The popularity of online education
platforms such as Coursera, Edx, and Udemy, which allow individuals to upskill [81] or
even transition to completely new fields of expertise, shows the importance of continued
education beyond college. We anticipate the role of AI in this domain to be transformative.
AI holds the potential to have a big impact on the way individuals engage with learning
processes throughout their lives, enabling highly personalized educational experiences
that adapt to the evolving needs and abilities of learners. Specifically, AI-driven learning
platforms could seamlessly integrate formal and informal learning opportunities, thereby
encouraging self-directed skill acquisition and continuous intellectual growth.
We expect AI to address the challenges inherent to lifelong learning and skill devel-
opment by delivering a dynamic and customizable educational framework. It can play a
pivotal role particularly in the context of the rapidly evolving job market and the need for
continuous upskilling [106]. AI-driven learning platforms can help individuals identify
skill gaps, access relevant learning resources, and track their progress, facilitating career
growth and adaptability [107]. One of the key aspects of AI is the potential to harness
massive datasets to identify emerging trends, skill gaps, and workforce requirements. The
extracted information can be used to design and deliver education and training programs
that align with the evolving demands of the global economy. By intelligently identify-
ing and addressing individual skill gaps, AI systems can facilitate efficient upskilling
and reskilling, enabling individuals to adapt more effectively to shifting job markets and
technological advancements.
7.3. AI Literacy and Ethics Education

As the proliferation of AI-based systems and applications continues to accelerate, the
need for individuals to possess a foundational understanding of AI principles, techniques,
and potential applications becomes increasingly crucial. AI literacy will be essential not
only for engineers but for all professions. Educators must prioritize the integration of
AI literacy into curricula, ensuring that students are equipped with the knowledge and
skills necessary to navigate the rapidly evolving AI landscape. This focus will enable
students to better comprehend and utilize AI technologies as well as prepare them for
careers in the AI-dominated future job market [22]. The increased integration of AI into
various aspects of society requires that we equip students with AI literacy and ethical
awareness [30]. Incorporating AI and ethics education in curricula can help students
understand the implications and responsibilities associated with AI technologies, fostering
responsible innovation and informed decision making [108].
As the impact of AI becomes more pervasive, ethical considerations surrounding its
deployment and use will become paramount [22]. Initial findings of an ongoing survey of
more than 450 students in Hong Kong and pilot focus group panels with 13 Australian stu-
dents indicated that educating students about risks, biases, and limitations was crucial [51].
Consequently, ethics education must be incorporated into AI-related curricula. It is crucial
to instill in students an awareness of the ethical implications and potential risks associated
with AI technologies. By fostering a critical mindset that balances the benefits of AI with its
potential harms, educators can promote responsible AI usage to mitigate adverse societal
consequences. The proposed approach will support the development of a generation of
AI-literate and ethically conscious individuals who are equipped to confront and navigate
the complex challenges presented by the rise of AI.
8. Conclusions
In this paper, we reviewed and analyzed the current literature to better understand
the potential effects of AI in education. We aimed to provide both a general overview as
well as a more specific discussion of various aspects of the subject. Our review focused
on three major themes: applications, benefits, and challenges. We found that the advent
of AI brings tantalizing possibilities and applications in the education sector. Its impact
is multifaceted and holds the potential to revolutionize the way learning is delivered and
experienced. As we enter the new era in education, the present study allows for a moment
of reflection based on the aggregate survey of the existing knowledge.
The applications of AI in education include personalized learning, intelligent tutor-
ing systems, assessment automation, and teacher–student collaboration, which can help
improve learning outcomes, efficiency, and global access to quality education. The scala-
bility of AI means that its benefits can be shared by large swaths of the society, providing
high quality education around the world. While AI has the capacity to make a significant
positive impact on education, it is important to keep in mind the dangers of misusing AI.
There are several concerns related to the deployment of AI; these include data privacy,
security, bias, and teacher–student relationships, and they must be addressed to ensure
the responsible and ethical implementation of AI in education. To meet the challenges
presented by the rise of the technology, AI literacy and ethics education must become a
part of the curricula. By leveraging these advancements, educators and policymakers can
work towards creating inclusive, equitable, and effective learning environments that cater
to the diverse needs of learners in the 21st century.
While this study presented a theoretical overview of the potential effects of AI in
education and can serve as a springboard for the development of the subject, an empirical
study is required to provide more concrete results. In the future, studies based on student
cohorts measuring the difference in the learning outcomes between AI-driven and tradi-
tional teaching methods or teacher surveys measuring the actual number of saved hours
when using automated grading systems are needed.
Author Contributions: Conceptualization, F.K.; methodology, F.K. and D.S.C.; validation, F.K., D.S.C.
and I.G.; formal analysis, F.K.; investigation, F.K. and D.S.C.; resources, F.K. and D.S.C.; data curation,
F.K. and D.S.C.; writing—original draft preparation, F.K., D.S.C. and I.G.; writing—review and
editing, F.K., D.S.C. and I.G.; visualization, F.K.; supervision, F.K.; funding acquisition, F.K., D.S.C.
and I.G. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Data Availability Statement: No new data were created or analyzed in this study.
Conflicts of Interest: The authors declare no conflict of interest.
References
1. Nicolescu, L.; Tudorache, M.T. Human-Computer Interaction in Customer Service: The Experience with AI Chatbots—A
Systematic Literature Review. Electronics 2022, 11, 1579. [CrossRef]
2. Adamopoulou, E.; Moussiades, L. Chatbots: History, technology, and applications. Mach. Learn. Appl. 2020, 2, 100006. [CrossRef]
3. Hwang, G.J.; Chang, C.Y. A review of opportunities and challenges of chatbots in education. Interact. Learn. Environ.
2021 , 1952615. [CrossRef]
4. Wollny, S.; Schneider, J.; Di Mitri, D.; Weidlich, J.; Rittberger, M.; Drachsler, H. Are we there yet?-A systematic literature review.
Front. Artif. Intell. 2021, 4, 654924 [CrossRef] [PubMed]
5. Okonkwo, C.W.; Ade-Ibijola, A. Chatbots applications in education: A systematic review. Comput. Educ. Artif. Intell. 2021,
2, 100033. [CrossRef]
6. Wilcox, H. Cheating Aussie Student Fails Uni Exam after Being Caught Using Artificial Intelligence Chatbot to Write Essay—
Now Australia’s Top Universities Are Considering a Bizarre Solution to Stop It Happening Again. DailyMail. 2023. Available
online: https://www.dailymail.co.uk/news/article-11688905/UNSW-student-fails-exam-using-OpenAIs-ChatGPT-write-essay.
html (accessed on 2 February 2023).
7. Stokel-Walker, C. AI bot ChatGPT writes smart essays-should academics worry? Nature 2022 . [CrossRef]
8. OpenA. GPT-4 Technical Report. arXiv 2023, arXiv:2303.08774.
9. Rudolph, J.; Tan, S.; Tan, S. ChatGPT: Bullshit spewer or the end of traditional assessments in higher education? J. Appl. Learn.
Teach. 2023, 6, 342–363.
10. GPT-4. Available online: https://openai.com/research/gpt-4 (accessed on 21 April 2023 ).
11. Vaswani, A.; Shazeer, N.; Parmar, N.; Uszkoreit, J.; Jones, L.; Gomez, A.N.; Kaiser, Ł; Polosukhin, I. Attention is all you need. Adv.
Neural Inf. Process. Syst. 2017, 30.
12. Arksey, H.; O’Malley, L. Scoping studies: Towards a methodological framework. Int. J. Soc. Res. Methodol. 2005, 8, 19–32.
[CrossRef]
13. Schwendimann, R.; Blatter, C.; Dhaini, S.; Simon, M.; Ausserhofer, D. The occurrence, types, consequences and preventability of
in-hospital adverse events—A scoping review. BMC Health Serv. Res. 2018, 18, 521. [CrossRef]
14. Munn, Z.; Peters, M.D.; Stern, C.; Tufanaru, C.; McArthur, A.; Aromataris, E. Systematic review or scoping review? Guidance for
authors when choosing between a systematic or scoping review approach. BMC Med. Res. Methodol. 2018, 18, 143. [CrossRef]
[PubMed]
15. Pollock, D.; Tricco, A.C.; Peters, M.D.; Mclnerney, P.A.; Khalil, H.; Godfrey, C.M.; Alex, L.; Munn, Z. Methodological quality,
guidance, and tools in scoping reviews: A scoping review protocol. JBI Evid. Synth. 2022, 20, 1098–1105. [CrossRef]
16. Tricco, A.C.; Lillie, E.; Zarin, W.; O’Brien, K.K.; Colquhoun, H.; Levac, D.; Moher, D.; Peters, M.D.J.; Horsley, T.; Weeks, L. PRISMA
extension for scoping reviews (PRISMA-ScR): Checklist and explanation. Ann. Intern. Med. 2018, 169, 467–473. [CrossRef]
[PubMed]
17. Hleg, A.I. Ethics Guidelines for Trustworthy AI. B-1049 Brussels. Available online: https://digital-strategy.ec.europa.eu/en/
library/ethics-guidelines-trustworthy-ai (accessed on 10 May 2023).
18. Terzopoulos, G.; Satratzemi, M. Voice assistants and artificial intelligence in education. In Proceedings of the 9th Balkan
Conference on Informatics, Sofia, Bulgaria, 26–28 September 2019; pp. 1–6.
19. Webber, S.S.; Detjen, J.; MacLean, T.L.; Thomas, D. Team challenges: Is artificial intelligence the solution? Bus. Horizons 2019, 62,
741–750. [CrossRef]
20. Marcinkowski, F.; Kieslich, K.; Starke, C.; Lünich, M. Implications of AI (un-) fairness in higher education admissions: The effects
of perceived AI (un-) fairness on exit, voice and organizational reputation. In Proceedings of the 2020 Conference on Fairness,
Accountability, and Transparency, Barcelona, Spain, 27–30 January 2020; pp. 122–130.
21. Ahmed, A.A.A.; Ganapathy, A. Creation of automated content with embedded artificial intelligence: A study on learning
management system for educational entrepreneurship. Acad. Entrep. J. 2021, 27, 1–10.
22. Borenstein, J.; Howard, A. Emerging challenges in AI and the need for AI ethics education. AI Ethics 2021, 1, 61–65. [CrossRef]
23. González-Calatayud, V.; Prendes-Espinosa, P.; Roig-Vila, R. Artificial intelligence for student assessment: A systematic review.
Appl. Sci. 2021, 11, 5467. [CrossRef]
24. Kamalov, F.; Sulieman, H.; Sant, R., Calonge, D. Machine learning based approach to exam cheating detection. PLoS ONE 2021,
16, e0254340. [CrossRef]
25. Miao, F.; Holmes, W.; Huang, R.; Zhang, H. AI and Education: A Guidance for Policymakers; UNESCO Publishing: Paris, France, 2021.
26. Timan, T.; Mann, Z. Data protection in the era of artificial intelligence: Trends, existing solutions and recommendations for
privacy-preserving technologies. In The Elements of Big Data Value: Foundations of the Research and Innovation Ecosystem; Springer
International Publishing: Cham, Switzerland, 2021; pp. 153–175.
27. United Nations Educational, Scientific and Cultural Organization. Recommendations on the Ethics of Artificial Intelligence. 2021.
Available online: https://unesdoc.unesco.org/ark:/48223/pf0000380455 (accessed on 10 April 2023).
28. JISC. AI in Tertiary Education: A Summary of the Current State of Play. Second Edition, June 2022. Available online: https:
//repository.jisc.ac.uk/8783/1/ai-in-tertiary-education-report-june-2022.pdf (accessed on 20 April 2023).
29. Kulshreshtha, D.; Shayan, M.; Belfer, R.; Reddy, S.; Serban, I.V.; Kochmar, E. Few-shot Question Generation for Personalized
Feedback in Intelligent Tutoring Systems. arXiv 2022, arXiv:2206.04187.
30. Long, T.; Qin, J.; Shen, J.; Zhang, W.; Xia, W.; Tang, R.; He, X.; Yu, Y. Improving knowledge tracing with collaborative information.
In Proceedings of the Fifteenth ACM International Conference on Web Search and Data Mining, Tempe, AZ, USA, 21–25 February
2022; pp. 599–607.
31. Mishkin, P.; Ahmad, L.; Brundage, M.; Krueger, G.; Sastry, G. DALL• E 2 Preview: Risks and Limitations. Noudettu 2022, 28, 2022.
Available online: https://github.com/openai/dalle-2-preview/blob/main/system-card.md (accessed on 22 April 2023).
32. Nguyen, A.; Ngo, H.N.; Hong, Y.; Dang, B.; Nguyen, B.P.T. Ethical principles for artificial intelligence in education. Educ. Inf.
Technol. 2022 , 28, 4221–4241. [CrossRef] [PubMed]
33. Government AI Readiness Index. Oxford Insights. 2022. Available online: https://www.oxfordinsights.com/government-ai-
readiness-index-2022 (accessed on 10 May 2023).
34. Qadir, J. Engineering education in the era of ChatGPT: Promise and pitfalls of generative AI for education. TechRxiv 2022
TechRxiv:techrxiv.21789434.v1.
35. St-Hilaire, F.; Vu, D.D.; Frau, A.; Burns, N.; Faraji, F.; Potochny, J.; Robert, S.; Roussel, A.; Zheng, S.; Glazier, T.; et al. A New era:
Intelligent tutoring systems will transform online learning for millions. arXiv 2022, arXiv:2203.03724.
36. Swiecki, Z.; Khosravi, H.; Chen, G.; Martinez-Maldonado, R.; Lodge, J.M.; Milligan, S.; Selwyn, N.; Gašević, D. Assessment in the
age of artificial intelligence. Comput. Educ. Artif. Intell. 2022, 3, 100075. [CrossRef]
37. Wahle, J.P.; Ruas, T.; Foltýnek, T.; Meuschke, N.; Gipp, B. Identifying machine-paraphrased plagiarism. In Proceedings of the
Information for a Better World: Shaping the Global Future: 17th International Conference, iConference 2022, Virtual Event, 28 February–4
March 2022; Proceedings, Part I; Springer International Publishing: Cham, Switzerland, 2022; pp. 393–413.
38. AlAfnan, M.A.; Dishari, S.; Jovic, M.; Lomidze, K. Chatgpt as an educational tool: Opportunities, challenges, and recommenda-
tions for communication, business writing, and composition courses. J. Artif. Intell. Technol. 2023, 3, 60–68. [CrossRef]
39. Bouschery, S.G.; Blazevic, V.; Piller, F.T. Augmenting human innovation teams with artificial intelligence: Exploring transformer-
based language models. J. Prod. Innov. Manag. 2023, 40, 139–153. [CrossRef]
40. Chan, C.K.Y.; Tsi, L.H. The AI Revolution in Education: Will AI Replace or Assist Teachers in Higher Education? arXiv 2023,
arXiv:2305.01185.
41. Chen, Y.; Jensen, S.; Albert, L.J.; Gupta, S.; Lee, T. Artificial intelligence (AI) student assistants in the classroom: Designing
chatbots to support student success. Inf. Syst. Front. 2023, 25, 161–182. [CrossRef]
42. Chetouani, M.; Dignum, V.; Lukowicz, P.; Sierra, C. (Eds.) Human-Centered Artificial Intelligence: Advanced Lectures; Springer
Nature: Berlin, Germany, 2023; Volume 13500.
43. Cotton, D.R.; Cotton, P.A.; Shipway, J.R. Chatting and cheating: Ensuring academic integrity in the era of ChatGPT. Innov. Educ.
Teach. Int. 2023, 1–12. [CrossRef]
44. Dai, W.; Lin, J.; Jin, F.; Li, T.; Tsai, Y.; Gasevic, D.; Chen, G. Can large language models provide feedback to students? A case study
on ChatGPT. EdArVix 2023.
45. 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. “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]
46. Elkins, S.; Kochmar, E.; Cheung, J.C.; Serban, I. How Useful are Educational Questions Generated by Large Language Models?
arXiv 2023, arXiv:2304.06638.
47. European Schoolnet. ChatGPT and the Role of AI in Assessment, Data4Learning Webinar Series; European Schoolnet: Brussels,
Belgium, 2023.
48. Hu, X.; Shubeck, K.; Sabatini, J. Artificial Intelligence-enabled adaptive assessments with Intelligent Tutors. In Innovating
Assessments to Measure and Support Complex Skills; Foster, N., Piacentini, M., Eds.; OECD Publishing: Paris, France, 2023.
49. Liu, Z.; Li, Y.; Wei, L.; Wang, W. Adaptive Exercise Recommendation Based on Cognitive Level and Collaborative Filtering. In
Proceedings of the Computer Science and Education: 17th International Conference, ICCSE 2022, Ningbo, China, 18–21 August 2022;
Revised Selected Papers, Part II; Springer Nature: Singapore, 2023; pp. 503–518.
50. Lodge, J.M.; Thompson, K.; Corrin, L. Mapping out a research agenda for generative artificial intelligence in tertiary education.
Australas. J. Educ. Technol. 2023, 39, 1–8. [CrossRef]
51. Liu, D.; Bridgeman, A.; Chan, Y.K.C. ‘Please Do Not Assume the Worst of Us’: Students Know AI Is Here to Stay and Want Unis
to Teach Them How to Use It. The Conversation. Available online: https://theconversation.com/please-do-not-assume-the-
worst-of-us-students-know-ai-is-here-to-stay-and-want-unis-to-teach-them-how-to-use-it-203426 (accessed on 15 May 2023).
52. Malmström, H.; Stöhr, C.; Ou, A.W. Chatbots and other AI for learning: A survey of use and views among university students in
Sweden. Chalmers Stud. Commun. Learn. High. Educ. 2023, 1.
53. Perkins, M. Academic Integrity considerations of AI Large Language Models in the post-pandemic era: ChatGPT and beyond. J.
Univ. Teach. Learn. Pract. 2023, 20, 7. [CrossRef]
54. Rasul, T.; Nair, S.; Kalendra, D.; Robin, M.; de Oliveira, Santini, F.; Ladeira, W.J.; Sun, M.; Day, I.; Rather, R.; Heathcote, L. The role
of ChatGPT in higher education: Benefits, challenges, and future research directions. J. Appl. Learn. Teach. 2023, 6.
55. Sabzalieva, E.; Valentini, A. ChatGPT and Artificial Intelligence in Higher Education: Quick Start Guide; UNESCO: Paris, France, 2023.
56. Sullivan, M.; Kelly, A.; McLaughlan, P. ChatGPT in higher education: Considerations for academic integrity and student learning.
J. Appl. Learn. Teach. 2023, 6.
57. UAE. United Arab Emirates 100 Practical Applications and Uses of Generative AI; Minister of State for Artificial Intelligence, Digital
Economy & Remote Work Applications Office: Dubai, United Arab Emirates, 2023.
58. Walton Family Foundation. Teachers and Students Embrace ChatGPT for Education. 2023. Available online: https://www.
waltonfamilyfoundation.org/learning/teachers-and-students-embrace-chatgpt-for-education (accessed on 12 May 2023).
59. Wylie, I. Can Artificial Intelligence Deliver Real Learning at Business School? Financial Times 2023. Available online: https:
//www.ft.com/content/34de1ae6-fdde-43e3-a39a-50e30b8b97a1 (accessed on 4 May 2023).
60. Clarke, V.; Braun, V.; Hayfield, N. Thematic analysis. Qualitative psychology: A practical guide to research methods. Qual.
Psychol. 2015, 3, 222–248.
61. Popkhadze, N. The Good, The Bad and The Ugly: AI in the higher education. Hallinnon Tutk. 2021, 40, 254–263 [CrossRef]
62. Google Trends. Google Trends. Available online: https://www.google.com/trends (accessed on 25 April 2023).
63. Luan, H.; Tsai, C.C. A review of using machine learning approaches for precision education. Educ. Technol. Soc. 2021, 24, 250–266.
64. Jonnalagadda, A.; Rajvir, M.; Singh, S.; Chandramouliswaran, S.; George, J.; Kamalov, F. An Ensemble-Based Machine Learning
Model for Emotion and Mental Health Detection. J. Inf. Knowl. Manag. 2023, 22, 2250075. [CrossRef]
65. Mahfood, B.; Elnagar, A.; Kamalov, F. Emotion Recognition from Speech Using Convolutional Neural Networks. In Proceedings of
the Data Analytics and Management: ICDAM 2022; Springer Nature: Singapore, 2023; pp. 719–731.
66. Mousavinasab, E.; Zarifsanaiey, N.; Niakan, R.; Kalhori, S.; Rakhshan, M.; Keikha, L.; Ghazi, S.M. Intelligent tutoring systems:
A systematic review of characteristics, applications, and evaluation methods. Interact. Learn. Environ. 2021, 29, 142–163. [CrossRef]
67. Kochmar, E.; Vu, D.D.; Belfer, R.; Gupta, V.; Serban, I.V.; Pineau, J. Automated data-driven generation of personalized pedagogical
interventions in intelligent tutoring systems. Int. J. Artif. Intell. Educ. 2022, 32, 323–349. [CrossRef]
68. Minn, S. AI-assisted knowledge assessment techniques for adaptive learning environments. Comput. Educ. Artif. In-
tell. 2022, 3, 100050. [CrossRef]
69. Vittorini, P.; Menini, S.; Tonelli, S. An AI-based system for formative and summative assessment in data science courses. Int. J.
Artif. Intell. Educ. 2021, 31, 159–185. [CrossRef]
70. Villegas-Ch, W.; Arias-Navarrete, A.; Palacios-Pacheco, X. Proposal of an Architecture for the Integration of a Chatbot with
Artificial Intelligence in a Smart Campus for the Improvement of Learning. Sustainability 2020, 12, 1500. [CrossRef]
71. Lee, D.; Yeo, S. Developing an AI-based chatbot for practicing responsive teaching in mathematics. Comput. Educ. 2022,
191, 104646. [CrossRef]
72. Klimova, B.; Seraj, P.M.I. The use of chatbots in university EFL settings: Research trends and pedagogical implications. Front.
Psychol. 2023, 14, 1146. [CrossRef]
73. Lin, C.J.; Mubarok, H. Learning analytics for investigating the mind map-guided AI chatbot approach in an EFL flipped speaking
classroom. Educ. Technol. Soc. 2021, 24, 16–35.
74. Nwana, H.S. Intelligent tutoring systems: An overview. Artif. Intell. Rev. 1990, 4, 251–277. [CrossRef]
75. Bhutoria, A. Personalized education and artificial intelligence in United States, China, and India: A systematic Review using a
Human-In-The-Loop model. Comput. Educ. Artif. Intell. 2022, 3, 100068. [CrossRef]
76. Hao, K. China has started a grand experiment in AI education. It could reshape how the world learns. MIT Technol. Rev. 2019, 123.
77. Kamalov, F.; Leung, H.H. Numerical computing in engineering mathematics. In Proceedings of the 2022 Advances in Science and
Engineering Technology International Conferences (ASET), Dubai, United Arab Emirates, 21–24 February 2022; IEEE: Manhattan,
NY, USA, 2022; pp. 1-5.
78. Kamalov, F.; Santandreu, D.; Leung, H.H.; Johnson, J.; El Khatib, Z. Leveraging computer algebra systems in calculus: A case
study with SymPy. In Proceedings of the 2023 IEEE Global Engineering Education Conference (EDUCON), Kuwait City, Kuwait,
1–4 May 2023; IEEE: Manhattan, NY, USA, 2023; pp. 1-6.
79. Bryant, J.; Heitz, C.; Sanghvi, S.; Wagle, D. How Artificial Intelligence Will Impact K-12 Teachers; McKinsey and Company: New
York, NY, USA, 2020
80. Luan, H. Democratizing education through AI-driven learning platforms. In Handbook of Artificial Intelligence in Education;
Springer: Berlin/Heidelberg, Germany, 2020; pp. 59–74.
81. Calonge, D.S.; Aman, Shah, M.; Riggs, K.; Connor, M. MOOCs and upskilling in Australia: A qualitative literature study. Cogent
Educ. 2019, 6, 1687392. [CrossRef]
82. Shah, M.A.; Calonge, D.S. Frugal MOOCs: An adaptable contextualized approach to MOOC designs for refugees. Int. Rev. Res.
Open Distrib. Learn. 2019, 20, 1–19.
83. Khanh, Q.V.; Chehri, A.; Quy, N.M.; Han, N.D.; Ban, N.T. Innovative Trends in the 6G Era: A Comprehensive Survey of
Architecture, Applications, Technologies, and Challenges. IEEE Access 2023.
84. Tondeur, J.; van Braak, J.; Zhu, C. Unpacking the potential of AI-driven learning in education: A multilevel framework. Comput.
Educ. 2020, 151, 103858.
85. Huang, R.H.; Soman, R. A new approach to promoting equity in education through AI-driven learning. J. Comput. Educ. 2020, 7,
279–299.
86. Holmes, W.; Porayska-Pomsta, K.; Holstein, K.; Sutherl, E.; Baker, T.; Shum, S.B.; Koedinger, K.R. Ethics of AI in education:
Towards a community-wide framework. Int. J. Artif. Intell. Educ. 2021, 1–23. [CrossRef]
87. Maslej, N.; Fattorini, L.; Brynjolfsson, E.; Etchemendy, J.; Ligett, K.; Lyons, T.; Manyika, J.; Ngo, H.; Niebles, J.C.; Parli, V. The AI
Index 2023 Annual Report; AI Index Steering Committee, Institute for Human-Centered AI, Stanford University: Stanford, CA,
USA, 2023.
88. AI Ethics Initiative. 2017. Available online: https://aiethicsinitiative.org (accessed on 11 May 2023).
89. Universite de Montréal. Declaration of Montréal for a Responsible Development of AI. Université de Montréal. 2018. Available
online: https://www.montrealdeclaration-responsibleai.com (accessed on 11 May 2023).
90. Nazir, A.; Mitra, R.; Sulieman, H.; Kamalov, F. Suspicious Behavior Detection with Temporal Feature Extraction and Time-Series
Classification for Shoplifting Crime Prevention. Sensors 2023, 23, 5811. [CrossRef]
91. Schiff, D. Education for AI, not AI for Education: The role of education and ethics in national AI policy strategies. Int. J. Artif.
Intell. Educ. 2022, 32, 527–563. [CrossRef]
92. Zeide, E. The structural consequences of big data-driven education. Big Data 2017, 5, 164–172. [CrossRef]
93. Bessen, J.E.; Impink, S.M.; Reichensperger, L.; Seamans, R. GDPR and the Importance of Data to AI Startups; NYU Stern School of
Business: New York, NY, USA, 2020.
94. Ntoutsi, E.; Fafalios, P.; Gadiraju, U.; Iosifidis, V.; Nejdl, W.; Vidal, M.E.; Ruggieri, S.; Turini, F.; Papadopoulos, S.;
Krasanakis E.; et al. Bias in data-driven artificial intelligence systems—An introductory survey. Wiley Interdiscip. Rev. Data Min.
Knowl. Discov. 2020, 10, e1356. [CrossRef]
95. Holstein, K.; Wortman, Vaughan, J.; Daumé, H., III; Dudík, M.; Wallach, H. Improving fairness in machine learning systems:
What do industry practitioners need? In Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems,
Glasgow, UK, 4–9 May 2019; pp. 1–16.
96. Khalil, M.; Prinsloo, P.; Slade, S. In the nexus of integrity and surveillance: Proctoring (re) considered. J. Comput. Assist. Learn.
2022, 38, 1589–1602. [CrossRef]
97. Selwyn, N.; O’Neill, C.; Smith, G.; Andrejevic, M.; Gu, X. A necessary evil? The rise of online exam proctoring in Australian
universities. Media Int. Aust. 2021, 186, 149–164. [CrossRef]
98. Terwiesch, C. Would Chat GPT Get a Wharton MBA? [White Paper]. Wharton School of Business. 2023. Available online:
https://mackinstitute.wharton.upenn.edu/wp-content/uploads/2023/01/Christian-Terwiesch-Chat-GTP-1.24.pdf (accessed on
8 May 2023).
99. Meo, S.A.; Talha, M. Turnitin: Is it a text matching or plagiarism detection tool? Saudi J. Anaesth. 2019, 13 (Suppl. S1), S48.
[CrossRef]
100. Khalil, M.; Er, E. Will ChatGPT get you caught? Rethinking of plagiarism detection. arXiv 2023, arXiv:2302.04335.
101. Buckingham, D. The Media Education Manifesto; Polity Press: Cambridge, UK, 2019.
102. Lindsey, R.; Dalenberg, A.; Kramer, L. The role of human touch in leveraging AI for education. J. Educ. Technol. Syst. 2020, 49,
167–183.
103. Yeadon, W.; Inyang, O.O.; Mizouri, A.; Peach, A.; Testrow, C.P. The death of the short-form physics essay in the coming AI
revolution. Phys. Educ. 2023, 58, 035027. [CrossRef]
104. Hwang, G.J.; Chien, S.Y. Definition, roles, and potential research issues of the metaverse in education: An artificial intelligence
perspective. Comput. Educ. Artif. Intell. 2022, 3, 100082. [CrossRef]
105. Radianti, J.; Majchrzak, T.A.; Fromm, J.; Wohlgenannt, I. A systematic review of immersive virtual reality applications for higher
education: Design elements, lessons learned, and research agenda. Comput. Educ. 2020, 147, 103778. [CrossRef]
106. Bessen, J. Artificial intelligence and jobs: The role of demand. In The Economics of Artificial Intelligence: An Agenda; University of
Chicago Press: Chicago, IL, USA, 2018; pp. 291–307.
107. Kabudi, T.; Pappas, I.; Olsen, D.H. AI-enabled adaptive learning systems: A systematic mapping of the literature. Comput. Educ.
Artif. Intell. 2021, 2, 100017. [CrossRef]
108. Boddington, P.; Rump, B. Ethics in computing, AI and data use: A case for a key stage 1-4 computer ethics curriculum. ACM
Trans. Comput. Educ. 2020, 20, 1–20.
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.

Sustainability 15 12451

Uploaded by

Copyright:

Available Formats

Sustainability 15 12451

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sustainability 15 12451

Uploaded by

Copyright:

Available Formats

sustainability

Keywords: AI; deep learning; education; intelligent systems; ChatGPT

Citation: Kamalov, F.; Santandreu

Sustainability 2023, 15, 12451. https://doi.org/10.3390/su151612451 https://www.mdpi.com/journal/sustainability

Enhanced Data Privacy

Figure 1. Multifaceted impact of AI in education.

The potential applications of AI in education include personalized learning, intelligent

Figure 2. Performance of GPT models on various standardized tests.

Table 1. Overview of included studies.

Author Year Main Contributions of the Surveyed Work

Author Year Main Contributions of the Surveyed Work

Figure 4. Overview of literature search process using PRISMA-ScR.

Table 2. Overarching themes unpacked: Personalized Learning.

• Customized Learning Experience: Personalized • Resource-Intensive: Implementing personalized

Table 3. Overarching themes unpacked: Intelligent Tutoring Systems.

Table 4. Overarching themes unpacked: Assessment Automation.

Table 5. Overarching themes unpacked: Teacher–Student Collaboration.

4.1. Theme 1: Personalized Learning

4.2. Theme 2: Intelligent Tutoring Systems

Figure 6. Intelligent humanoid leading a tutoring session (DALL·E 2 image).

Student modeling is another critical component of ITS. It is a process in which AI

4.3. Theme 3: Assessment Automation

4.4. Theme 4: Teacher–Student Collaboration

5.1. Enhanced Learning Outcomes

5.2. Time and Cost Efficiency

Figure 7. Top barriers to providing personalized learning, % of teachers identifying area as a

AI has the potential to alter educational institution budget allocations by reducing

5.3. Global Access to Quality Education

Figure 8. Popularity of the search query “AI in education” by region.

6. Challenges and Ethical Considerations

Figure 9. Number of ethics incidents related to AI.

6.1. Data Privacy and Security

6.2. Bias and Discrimination

6.3. Plagiarism and Academic Integrity

6.4. Teacher–Student Relationship

The consequences of the new teacher–student paradigm can be unexpected. As AI

7. Future Directions and Opportunities

7.1. Augmented and Virtual Reality

7.2. Lifelong Learning and Skill Development

7.3. AI Literacy and Ethics Education

You might also like