Harnessing The Power of Artificial Intel
Harnessing The Power of Artificial Intel
Irfan Jahic
Educational Technology
Graz University of Technology
Austria
irfan.jahic@student.tugraz.at
Martin Ebner
Educational Technology
Graz University of Technology
Austria
martin.ebner@tugraz.at
Sandra Schön
Universitas Negeri Malang
Indonesia
mail@sandra-schoen.de
Abstract. The usage of AI, more precisely ChatGPT, is almost embedded in our everyday routine.
It reduces the time, effort, and cost of performing boring and now also very delicate tasks. It appears
that ChatGPT knows everything and knows how to explain everything in a personalized and peculiar
way, adjusted to the user’s preferences. In this paper, we will explore how ChatGPT can be used in
higher education and what its limitations are, if any at all. We shall endeavor to examine both the
advantages and disadvantages of ChatGPT in education and state of the art. We will use a literature
review as our research method to explore and find the necessary answers. We will explore and use
various scientific articles, newspaper articles, blog posts, and other relevant literature from many
authors. Our results and conclusion will show how to use ChatGPT in the best way in education and
what to look out for now that ChatGPT is still available to everyone.
1. Introduction
Education has already undergone quite some changes over the years due to technological advancements. We
observed some visible changes in education, such as in the design and elements of a classroom - blackboards,
notebooks, and pencils are almost completely extinct and are now replaced by projectors, laptops, keyboards, and
even some new digital classrooms. For example, we observed intensively the change of our beginners’ for more
than 10 years now (Ebner et al., 2023; Nagler et al., 2022; Ebner et. al., 2012). Many changes that are not observable
visually in education also occurred with the help of technological advancements. The quality of teaching and
learning has been significantly improved by, e.g., using online tools or just a faster way to reach the professor.
Today, we our research is strongly concentrated to digital transformation of teaching (Thoring et al., 2018)
(Alhubaishy & Aljuhani, 2021) and digital skills (Jørgensen, 2019). With the emergence of ChatGPT - is education
now again a target for yet another change, maybe one of the most significant changes ever?
The idea of having some machines or tools as well as intelligent algorithms that either completely replaces
humans in performing some tasks or helps perform human tasks, is not new (Ebner & Schiefner, 2008). People
have been fantasizing about this for a long time, and therefore the oldest mention of such an entity in a myth about
Talos, a huge bronze protector of Crete. We somehow went from writing myths about artificial intelligence to now
finally having such an artificial intelligence model that can write such myths by simply asking it to. ChatGPT is
now at a stage where it, similarly to how Talos protected Crete, protects artificial intelligence and its usage in
education - hopefully not with the same result.
The term artificial intelligence (AI) was first mentioned by John McCarthy in 1956 and defined as ”the
science and engineering of making intelligent machines.” (Joshi, A. & Mishra, 2010). AI now has a slightly
Draft, finally published in: Jahic, I., Ebner, M. & Schön, S. (2023). Harnessing the power of artificial intelligence and
ChatGPT in education – a first rapid literature review. In T. Bastiaens (Ed.), Proceedings of EdMedia + Innovate Learning
(pp. 1462-1470). Vienna, Austria: Association for the Advancement of Computing in Education (AACE). Retrieved July 13,
2023 from https://www.learntechlib.org/primary/p/222670/
different definition since the Cambridge dictionary today defines it as ”the study of how to produce machines that
have some of the qualities that the human mind has, such as the ability to understand language, recognize pictures,
solve problems, and learn.”1 We now consider AI a broad field with many subsets, such as machine learning, deep
learning, expert systems, machine vision, etc. Many of these subsets of AI have already found their purpose and
application in education. First universities already use AI tools such as plagiarism detectors, chatbots on their
homepages (Kühnelt et al., 2020), and tools to help transcribe lectures.
One of the subsets of AI is generative AI which denotes AI that can create data such as images, texts, music, videos,
etc. Many generative AIs are already available such as Jasper Art2, Musenet3, Midjourney4, and ChatGPT5. The
prime example of generative AI and a very hot topic is of course ChatGPT. ChatGPT may exist for only few
months, but the impact it has already had on education is immense and remarkable.
ChatGPT is an advanced generative AI model based on a generative pre-trained trans-former (GPT)
architecture that can create human-like responses after the user’s text prompt. This is accomplished through natural
language processing, and thanks to that process, the outcome is moreless grammatically correct, contextually
relevant in most cases, and is almost always indistinguishable from the human text. The current version of ChatGPT
uses GPT-3.5 architecture, and the GPT-4 version is already ready to be tested out and, according to OpenAI,
performs a lot better than its predecessor.
The emergence of ChatGPT has caused a lot of mess in the academic technology enhanced learning
community. Many debated the benefits and challenges of such a tool, whether it should be allowed in education,
or whether one should use it in general. To explore and find out more about the potential of ChatGPT in education,
we will propose three research questions within this publication:
• How can students and teachers use ChatGPT in education?
• What are the disadvantages of using ChatGPT in education?
• What are the advantages of using ChatGPT in education?
We will perform a first rapid literature review to help answer these questions following the approach of
Wu et al. (2021). We will discuss our research methodology and approach in detail in the methods section.
Additionally, we will go through our findings and results in the findings section. Then, we will provide concrete
and most precise answers to our three research questions, discuss our findings, and give more context and meaning
to them in the discussion section. Finally, we will conclude our paper with some final remarks and some
recommendations for the future.
2. Research method
To accomplish our goal, we performed a literature review by doing the six steps described by Machi and
McEvoy (2016) and the approach of Wu et al. (2012). Figure 1 shows those six steps. For our literature review, we
used the keywords “ChatGPT” and “ChatGPT in education” to search for relevant papers across different databases.
We gathered articles and records from different sources such as Google Scholar6, IEEE Explore7, ScienceDirect8,
and WebOfScience9 and since the subject is still in the early stages of development, we also took newspaper articles
and blog posts in to consideration that addressed our topic and questions and popped up during our search. We
initially found 112 records from the aforementioned sources, where the most amount of the records came from
Google Scholar. We found additional ten blog posts and newspaper articles, which we also took into consideration.
1
https://dictionary.cambridge.org/dictionary/english/artificial-intelligence (last accessed May 2023)
2
https://www.jasper.ai (last accessed May 203)
3
https://openai.com/research/musenet (last accessed May 2023)
4
https://www.midjourney.com/ (last accessed May 2023)
5
https://chat.openai.com/ (last accessed May 2023)
6
https://scholar.google.com/ (last accessed May 2023)
7
https://ieeexplore.ieee.org (last accessed May 2023)
8
https://www.sciencedirect.com/ (last accessed May 2033)
9
https://www.webofscience.com/wos/woscc/basic-search (last accessed May 2023)
Draft, finally published in: Jahic, I., Ebner, M. & Schön, S. (2023). Harnessing the power of artificial intelligence and
ChatGPT in education – a first rapid literature review. In T. Bastiaens (Ed.), Proceedings of EdMedia + Innovate Learning
(pp. 1462-1470). Vienna, Austria: Association for the Advancement of Computing in Education (AACE). Retrieved July 13,
2023 from https://www.learntechlib.org/primary/p/222670/
We exported all our findings to a .txt file, and with the help of Excel and some Python scripting, we managed
to have all our data processed, sorted, and ready for further inspection. Since we gathered all our initial results in a
spreadsheet, we could easily spot the duplicates and discard them as well as the articles that we could not access
online. After that, we read the abstract, introduction, and conclusion of all the remaining articles to decide whether
that record was appropriate for our topic. In total, we excluded 71 articles. The most common reason for discarding
a record was that it simply did not address the usage of ChatGPT in education but rather in some other environment.
We carefully reviewed a total of 41 articles, categorizing and describing the prevailing trends and ideas they
presented. All the articles are listed in the chapter “References for the literature review”. The graphic representation
of the entire record selection process, in our case according to the PRISMA flow chart, is depicted in Figure 2:
Findings
In this section, we aim to provide an overview of the main findings of the review, summarizing the key
themes, trends, and insights that have emerged from our analysis of the literature. After reading and analyzing all
41 papers, we decided to categorize the records and report some of the peculiarities that we spotted.
In Figure 3, we show where our references were published from, i.e., from which continent. We can see that
most records come from North America (n=16), and a close second comes from Asia (n=11). It is important to
mention that there are some records that were published by institutions from two different continents. In that
case, we increased the count for both continents by one.
Figure 4 illustrates the year of publishing of all papers. We notice that all of the papers we took into
consideration were published either in 2022 or in 2023. This comes as no surprise since ChatGPT was published
on November 30th, 2022. Even though GPT language models existed before 2022, it was only after GPT-3,
GPT-3.5, and ChatGPT specifically were released that they gained much popularity. We observed an increasing
trend of mentioning ChatGPT in scientific publications and records.
Our goal is to examine ChatGPT in education in general, but we also focused on exploring how ChatGPT is
used in some specific fields of education. There have been 24 records that talk about the utilization of ChatGPT
in education in general and 17 records that mentioned and demonstrated the usage of ChatGPT in some specific
branch of education - this ratio is illustrated by Figure 5.
Fig. 5 Number of records that address general education and field-specific education.
Table 1 includes every specific field mentioned by these 17 records and we will include some quotes from
those papers. As can be seen from the table the most popular education was medical education and some closely
related fields such as healthcare in general, neurology or nursing.
Table 1: Specific fields of education mentioned by the records, and their representative quotes.
Field Studies Direct quotation
Medicine [10, 16, 20, 23, 26] “To acquire correct and in-depth information from ChatGPT
requires the active learning capability of users because only those
who can identify their knowledge gaps and understand what they
do not know can use ChatGPT efficiently.” [20]
Healthcare [36] “If properly addressed, these technologies can have the potential
to expedite the research and innovation in healthcare and can aid
to promote equity in research by overcoming the language
barriers” [38]
Statistical [27] “In this section, we ask ChatGPT to provide descriptions that can
Process be used in reporting SPC concepts in practice” [28]
Control
Software [21] “In this paper, we examined the potential applicability of
testing ChatGPT to a popular software testing curriculum. We found that
ChatGPT is able to provide correct or partially correct answers to
55.6% of questions” [21]
Neurology [29] “On the educator’s side, it might also be tempting to use a system
like this to write the basis for a new course.” [30]
Hospitality [7] “Hospitality and tourism schools may choose to allow the use of
& Tourism ChatGPT in specific contexts while also implementing safeguards
and guidelines to ensure that its use aligns with their values and
educational goals.” [7]
Draft, finally published in: Jahic, I., Ebner, M. & Schön, S. (2023). Harnessing the power of artificial intelligence and
ChatGPT in education – a first rapid literature review. In T. Bastiaens (Ed.), Proceedings of EdMedia + Innovate Learning
(pp. 1462-1470). Vienna, Austria: Association for the Advancement of Computing in Education (AACE). Retrieved July 13,
2023 from https://www.learntechlib.org/primary/p/222670/
On the one hand, many of these records mentioned the benefits of using ChatGPT and how good it can be
when used in education. On the other hand, many of these publications also inform on the bad sides and
disadvantages of using ChatGPT in education. We identified that the most popular advantage mentioned was
personalized learning. Many papers described how ChatGPT could be used to focus on teaching a subject
tailored to the learner’s needs and environment. The most popular disadvantage was, by far plagiarism. Authors
argued that our current inability to spot the text generated by ChatGPT is a great challenge to our current
assessment. Furthermore, we do not even possess a reliable tool that can confirm with certainty whether the text
was AI-generated or not. It is unlikely that we will ever have such a tool. We illustrate in Figure 6 how many
papers mentioned plagiarism (n=31) and how many did not (n=10).
Fig. 6. Number of records that did and did not mentioned plagiarism.
In this section, we will provide short, clear, and concise answers to the three research questions we
mentioned at the beginning of the paper according to our findings in this first rapid literature review.
Research question 1: How can students and teachers use ChatGPT in education?
ChatGPT is used in education by both students and teachers to accomplish and fulfill learning goals, speed
up the learning process, and improve that process altogether. We compiled, based on our rapid literature review,
a list of how teachers and students can use ChatGPT in education from the records we read. According to the
existing research work ChatGPT can be used:
• as a great helper during research
• to produce exams on a specific topic
• to produce curriculums and plan courses
• to help with assessments
• to encourage critical thinking
• to summarize large chunks of text, e.g. whole chapters
Furthermore, we were able to observe scenarios where ChatGPT is helpful and advantageous. It proved to
be an exceptional tool that is extremely useful with a lot of potential in education. Some of the advantages of
using ChatGPT, based on the literature review, in education are:
Most of the listed papers focused on both benefits and limitations. Nevertheless, it must be noticed that
some papers focused entirely on the disadvantages of ChatGPT. We compiled this list of mentioned
disadvantages of using ChatGPT in education based on our first rapid literature review:
• Plagiarism concerns
• Bias, because of the restriction to only few sources
• AI hallucinations
• Privacy concerns
Discussion
In this part of the publication, we try to summarize our findings and collect some main outcomes of this first
rapid literature review according to their relevance to education.
We cannot deny the fact that ChatGPT has a huge impact on education and that it is already widely used in
everyday activities surrounding education, whether it is in the classroom, writing homework, an essay, or a
scientific paper. If it remains open to the public, the impact will be more significant, and this tool will be used in
more ways that are yet to be discovered. The improvements to ChatGPT will accomplish the same effect -
improvements such as GPT-4 version or training data that is up to date.
ChatGPT can make many tasks much more manageable, enabling substantial time-saving. It is helpful for
teachers as well as students. It helps students access information faster, and it can provide concrete solutions and
examples to their prompts. For teachers, it can play a prominent role in grading papers, constructing exams, and
providing ideas for new curriculums and plans.
As Sok Sarin and Heng Kimkong commented in their paper [37], ChatGPT can be treated as an assistant
that helps to produce essays, translate languages, and summarize texts. It can also recommend and explain topics
to specific readers’ needs. In one exam-ple, it produced a tailored answer for a reader with dyslexia.
Furthermore, as shown by Han Zhiyong et al. [20] ChatGPT produced multiple-choice exams in seconds.
ChatGPT can understand and write in many different languages, which is useful when learning languages. Not
only can ChatGPT communicate in many different world languages, but it is also fluent in programming
languages and can translate from one programming language to another. Additionally, it can produce code in a
desired programming language, as shown by Opara Emmanuel et al. [30]. There are many other cases where
ChatGPT is exceptionally good at executing a particular task that would benefit each learner or teacher. All of
these perks combined are what make this tool unbelievably brilliant. It is hard to imagine that there exists just
one tool that can satisfy more than just a typical user.
ChatGPT-generated text is hard to distinguish from human work which represents a threat to academic
integrity, possible plagiarism problems, and academic misconduct. Additionally, it can also produce incorrect
statements that it believes to be true with high confidence. That characteristic is known as an AI hallucination,
which is not rare in similar AI tools. Another limitation mentioned in most papers, like in [30], for example, is
the bias of the training data that ChatGPT used. Another critical thing to consider if ChatGPT is allowed to be
used is to grant permission to all students and inform them about it to avoid any inequalities or inequities. Last
but not least, there are always privacy concerns for every user.
Since the benefits and drawbacks are pretty much well-known already, it is important to react properly and
accordingly. The end goal is to use the advantages to the best and fullest while eliminating pitfalls of
disadvantages. The plagiarism issue is probably the biggest issue and concern in the community, which affects
all levels of education. It is important to educate students to write essays and develop critical thinking and
problem-solving skills. The traditional homework essay might not be the best task to give to students in future. A
possible solution would be to give students enough time to educate themselves on a presented topic and to ask
them to write an essay in class with a supervisor present. Other flaws and limitations are easier to handle by
simply informing users about them. It is essential to understand bias, how it occurs in such models, and what
hallucinations are in AI. Bias in AI refers to systematic and unjustified assumptions that may be reflected in an
AI system’s data, algorithms, or decision-making processes. AI hallucinations refer to the phenomenon where an
AI system generates unexpected or meaningless outputs that appear coherent and plausible to human observers.
A simple dose of skepticism and the need to proof-check concrete information is enough to handle those
problems.
Education is due to change, and everybody has got their eyes on the next move of educational institutions.
However, banning ChatGPT and similar technologies is not a long-term solution as long as ChatGPT or even
next AI based tools remains available. Traditional assessment will inevitably change, and formal incorporation
of such tools in class is only a matter of time. In our opinion, ChatGPT brings more good than bad, and change
is almost always good and welcomed.
Conclusion
Draft, finally published in: Jahic, I., Ebner, M. & Schön, S. (2023). Harnessing the power of artificial intelligence and
ChatGPT in education – a first rapid literature review. In T. Bastiaens (Ed.), Proceedings of EdMedia + Innovate Learning
(pp. 1462-1470). Vienna, Austria: Association for the Advancement of Computing in Education (AACE). Retrieved July 13,
2023 from https://www.learntechlib.org/primary/p/222670/
ChatGPT is, without a doubt, a powerful and valuable tool that brings exciting expectations for the future of
generative AI and other GPT and NLP tools. It brings many benefits and challenges to education that need to be
adequately addressed and defined, timely. This AI tool has the potential to transform education with promises of
personalization of education, creating a more interactive learning environment, and many more.
However, the full integration of ChatGPT into education should be carefully planned and executed.
Challenges such as plagiarism, bias, data privacy, and similar should be tackled. One of the first steps should be
defining rules for ChatGPT usage in education. In summary, we firmly believe that ChatGPT has more pros than
cons and a bright future in education. It has the potential to change education even more significantly than a
calculator, or Google did.
[28] Mhlanga, D., (2023) Open ai in education, the responsible and ethical use of chatgpt towards lifelong learning.
Education, the Responsible and Ethical Use of ChatGPT Towards Lifelong Learning (February 11, 2023)
[29] Moisset, X. & Ciampi de Andrade, D. (2023) Neuro-chatgpt? potential threats and certain opportunities. Revue
Neurologique, pp S0035–3787.
[30] Opara, E., Adalikwu, M.-E. T. & Tolorunleke C. A. (2023) Chatgpt for teaching, learning and research: Prospects and
challenges. Opara Emmanuel Chinonso, Adalikwu Mfon-Ette Theresa, Tolorunleke Caroline Aduke (2023). ChatGPT for
Teaching, Learning and Research: Prospects and Challenges. Glob Acad J Humanit Soc Sci, 5.
[31] Perkins, M. (2023). Academic integrity considerations of ai large language models in the post-pandemic era: Chatgpt and
beyond. Journal of University Teaching & Learning Practice, 20(2):07
[32] Qadir, J. (2023, May). Engineering education in the era of ChatGPT: Promise and pitfalls of generative AI for education.
In 2023 IEEE Global Engineering Education Conference (EDUCON) (pp. 1-9). IEEE.
[33] Qi, X., Zhu, Z., & Wu, B. (2023). The Promise and Peril of ChatGPT in Geriatric Nursing Education: What We Know
and Do Not Know. Aging and Health Research, 100136.
[34] Rospigliosi, P. A. (2023). Artificial intelligence in teaching and learning: what questions should we ask of ChatGPT?.
Interactive Learning Environments, 31(1), 1-3.
[35] Rudolph, J., Tan, S., & Tan, S. (2023). ChatGPT: Bullshit spewer or the end of traditional assessments in higher
education?. Journal of Applied Learning and Teaching, 6(1).
[36] Sallam, M. (2023). The utility of ChatGPT as an example of large language models in healthcare education, research and
practice: Systematic review on the future perspectives and potential limitations. medRxiv, 2023-02.
[37] Sok, S., & Heng, K. (2023), ChatGPT for Education and Research: A Review of Benefits and Risks. Available at SSRN:
https://ssrn.com/abstract=4378735 or http://dx.doi.org/10.2139/ssrn.4378735
[38] Sun, G. H., & Hoelscher, S. H. (2023). The ChatGPT storm and what faculty can do. Nurse Educator, 48(3), 119-124.
[39] Thorp, H. H. (2023). ChatGPT is fun, but not an author. Science, 379(6630), 313-313.
[40] Zhai, X. (2022). ChatGPT User Experience: Implications for Education. Available at SSRN:
https://ssrn.com/abstract=4312418 or http://dx.doi.org/10.2139/ssrn.4312418
[41] Zhai, X. (2023). ChatGPTfor next generation science learning. XRDS: Crossroads, The ACM Magazine for Students,
29(3), 42-46.
References
Alhubaishy,A.,& Aljuhani,A.(2021).The challenges of instructors’ and students’ attitudes in digital transformation: A case
study of Saudi Universities. Education and Information Technologies, 26(4), 4647-4662.
Ebner, M., Schiefner, M. (2008) Will E-Learning die?; in: E-Learning: 21st Century Issues and Challenges, Audrey R.
Lipshitz and Steven P. Parsons (Ed.), Nova Publishers, pp. 69-82, 2008, ISBN: 978-1-60456-156-2
Ebner, M., Nagler, W. & Schön, M. (2012). Have They Changed? Five Years of Survey on Academic Net- Generation. In
Proceedings of World Conference on Educational Multimedia, Hypermedia and Telecommunications 2012 (pp. 343-353).
Chesapeake, VA: AACE
Ebner, M., Mair., B., De Marinis, C., Müller, H, Nagler, W., Schön, S., Thurner,S. (2023). Digital Transformation of
Teaching and Perception at TU Graz from the Students’ Perspective: Developments from the Last 17 Years. In: Auer, M.E.,
Pachatz, W., Rüütmann, T. (eds) Learning in the Age of Digital and Green Transition. ICL 2022. Lecture Notes in Networks
and Systems, vol 633. Springer, Cham, pp 366–377. doi.org/10.1007/978-3-031-26876-2_34
Jørgensen, T.(2019).DigitalSkills: Where Universities Matter. Learning and Teaching Paper #7, European University
Association, Retrieved May 2023 from URL: https://eua.eu/downloads/publications/dig-
ital%20skills%20%20where%20universities%20matter.pdf
Joshi, A. & Mishra, G. (2010) Artificial intelligence. In Proceedings of the International Conference and Workshop on
Emerging Trends in Technology (ICWET '10). Association for Computing Machinery, New York, NY, USA, 1023.
https://doi.org/10.1145/1741906.1742236
Kühnel, J., Ebner, M., Ebner, M. (2020) Chatbots for Brand Representation in Comparison with Traditional Websites. In:
International Journal of Interactive Mobile Technologies (iJIM) 14(18):18-33
Machi, L., A. & McEvoy, B. T. (2016) The literature review: Six steps to success. Corwin, Ed. 3, ISBN: 9781506336244
Nagler, W., Schön, S., Mair, B., Ebner, M. & Edelsbrunner, S. (2022). A decade of first-semester students surveys concerning
IT equipment and communication applications and effect of Covid-19 related experiences for first-year students in 2021. In T.
Bastiaens (Ed.), Proceedings of EdMedia + Innovate Learning (pp. 1060-1067). New York City, NY, United States:
Association for the Advancement of Computing in Education (AACE). Retrieved May, 2023 from
https://www.learntechlib.org/primary/p/221412
Draft, finally published in: Jahic, I., Ebner, M. & Schön, S. (2023). Harnessing the power of artificial intelligence and
ChatGPT in education – a first rapid literature review. In T. Bastiaens (Ed.), Proceedings of EdMedia + Innovate Learning
(pp. 1462-1470). Vienna, Austria: Association for the Advancement of Computing in Education (AACE). Retrieved July 13,
2023 from https://www.learntechlib.org/primary/p/222670/
Thoring, A., Rudolph, D., & Vogl, R. (2018). The digital transformation of teaching in higher education from an academic’s
point of view: An explorative study. In International Conference on Learning and Collaboration Technologies (pp. 294-309).
Springer, Cham.
Wu, W.-H., Chiou, W.-B., Kao, H.-Y., Hu, C.-H.A. & Huang, S.-H. (2012) Re-exploring game-assisted learning research:
The perspective of learning theoretical bases. Comput. Educ. 59, 1153–1161.