Discussing ChatGPT in education: A literature review and bibliometric analysis

Abstract This article aims to provide a review of existing research on the use of OpenAI ChatGPT in education using bibliometric analysis and systematic literature review. We explored published articles to observe the leading contributors to this field, the important subtopics, and the potential study in the future. We discovered that there is already a rise in the quantity of articles published on this subject between 2022 and 2023. We also map the research clusters using network analysis. The results obtained found a total of 93 articles. The analysis conclude that keywords such as “challenge,” “teaching,” and “knowledge,” have not been thoroughly researched This article contributes to the literature discussing the use of artificial intelligence in education and can be useful for academics and policymakers in education.


PUBLIC INTEREST STATEMENT
The goal of this study is to provide a review of existing research on the use of OpenAI ChatGPT in education using bibliometric analysis and systematic literature review.To assess contributions to marketing research and to graphically explain the findings of bibliometric approaches, we used visualization software.This article contributes to the literature discussing the use of artificial intelligence in education and can be useful for academics and policymakers in education.

Introduction
Concerns regarding what to educate and how to teach the future generation have been raised as a result of the major changes artificial intelligence (AI) is making to the labor market, which is one of education's main functions.Such worries underscore the necessity of educating future citizens to provide them with the necessary skills and competence to live in the society that is changing quickly (Zhai, 2021).It is crucial to comprehend what AI is capable of in order to handle this issue.To save time and money, AI is anticipated to replace human labor in everyday tasks (OECD, 2021).Such progress has been made in many facets of life, including the grading of homework (Zhai et al., 2020).Since the introduction of technology and artificial intelligence, the educational landscape has experienced a considerable upheaval (AI).The way that students and teachers communicate, study, and share knowledge has been completely transformed by AI.One such AI model is OpenAI's ChatGPT, which has been gaining popularity in the education sector for its ability to generate human-like responses and summarize complex information.
The study focuses on the role of ChatGPT, a powerful AI tool, in conducting various tasks that require both knowledge and creative intelligence (Bozkurt et al., 2023).ChatGPT has the potential to automate tasks such as exam-style questions, homework assignments, academic essays, and contract generation (O'Connor, 2022).According to Stokel-Walker (2022), ChatGPT is a gamechanger that may eliminate the need for certain traditional forms of assignments and assessments like essay writing.
Although there is already existing research on the topic of ChatGPT in education sector (Pavlik, 2023), to the best of the authors' knowledge (Rudolph et al., 2023), there have been only few studies that have specifically reviewed and evaluated the subject area of ChatGPT in education sector using bibliometric and tissue analysis techniques (Thorp, 2023).This study aims to fill this gap by reviewing the available literature in the broad field of ChatGPT in education sector, identifying the main contributing authors, journals, and keywords through bibliometric analysis and suggesting future research directions related to different ChatGPT in education sector.The goal of conducting a bibliometric analysis is to analyze research in a systematic, reproducible, and transparent manner, in order to inform current and future researchers about the trends and evolution of the field of study and minimize research bias through mining/audit trails of exhaustive literature databases.The next section of the study reviews the relevant literature to gain deeper insight into various aspects of ChatGPT in education sector and the methodology used to conduct this research.This article is organised as follows: an introduction is followed by a quick survey of the literature.After that, we talk about the bibliometric approach before talking about the outcomes.The final section will go through the study's key findings and suggestions for the future.

Literature review
By making the different management procedures nimble and precise, artificial intelligence (AI) plays a part in improving an organization's function's efficiency and effectiveness (Palos-Sanchez et al., 2022).AI will make it possible for educational institutions to comprehend and manage a data collecting procedure so that it may be incorporated into an educational efficiency plan (Mariño-Romero et al., 2022).One of the tools which has been an interesting topic is ChatGPT.Based on the GPT language model technology, it is a very sophisticated chatbot that can handle a variety of text-based requests, including simple question answering and more difficult tasks like writing essays and assisting people in difficult conversations about productivity problems (Lund & Wang, 2023;Tlili et al., 2023).By utilising its enormous data stores and effective architecture, ChatGPT is able to comprehend and interpret user requests before producing suitable replies in almost natural human language (Hasbi et al., 2022).
Since we are still in the early stages of discussing the use of OpenAI in education, particularly about ChatGPT, we would like to contribute by showing the trending topics related to this discussion.Rojas-Sánchez et al. ( 2022) started analyzing the development of using technology in education by conducting a bibliometric analysis, thus motivated us to conduct the same methodology for this article.
According to Pradana et al. (2023), bibliometric analysis is a method for quantifying, tracking, and analysing academic publications.It lists the authors' publications, the top journals, the methodology employed, and the findings reached (Susanty et al., 2022).Any field of study may be summarised using metadata (Wardhana et al., 2022).The examination of various themes, journals, and countries has all been done using bibliometric approaches, which require a significant amount of bibliographic data (Palos-Sanchez et al., 2022).

Recent publications about the use of OpenAI in education
An artificial intelligence research facility named OpenAI released a chatbot called ChatGPT (Generative Pre-trained Transformer) in November 2022.Natural language processing (NLP) is used by ChatGPT, a conversational artificial intelligence interface that converses realistically and even "answers follow-up questions, admits its mistakes, challenges incorrect premises, and rejects inappropriate requests" (van Dis et al., 2023).While ChatGPT's main purpose was to mimic human speech, it can invent new drafts, such as poems, stories, or novels, as well as act in whichever orders given to it (Firat, 2023).
The benefits of ChatGPT in education have been mentioned in several preprints of papers, blogs, and media sites (Firat, 2023).However, both policymakers in education and users' opinions of ChatGPT are split concerning the adoption of the technology in education (Rojas-Sánchez et al., 2022).When we tried to explore more about it, the use of technologies such as ChatGPT was still not fully explored in the field of education.Therefore, we would like to go deeper on whether academics see it as threat or opportunity.A bibliometric analysis of the available research on ChatGPT in the education sector is the main objective of this study.A research technique called bibliometric analysis uses quantitative analysis and statistics to find patterns and trends in publications.We put the phases of the analysis using PRISMA procedure which is shown in Figure 1 (Moher et al., 2015;Page et al., 2021).
This study makes use of Google Scholar databases, Microsoft Excel for statistical analysis, and VOSviewer as a network map-making program for bibliometric analysis.This approach can also be used to get a general understanding of the subject, spot knowledge gaps, and guide future research (Pradana et al., 2023).The bibliometric analysis conducted in this study adheres to the recommendations made by Donthu et al. (2021).

Result and discussion
As explained in Figure 1, this study reviews 93 articles related to ChatGPT in education, this research has appeared since 2022.During 2022, there are 19 articles related to ChatGPT in education research.In 2023 there were 74 publications resulting in a significant increase in the number of publications.Our data is still temporary because 2023 will not be completed until December 2023, so it is still possible to add and add to the collection of papers in this research (Figure 2).This study intends to investigate the ChatGPT in education in greater detail.The explanation that follows displays the results of the analysis that was performed using VOSviewer.Co-authorship, coaccuracy, and citations were presented as three distinct characteristics of the data analysis..

Co-accuracy
With a minimum of 1 co-occurrence, the author employed a co-occurrence analysis kind of phrase to obtain 50 keywords utilising the complete counting technique.The co-occurrence analysis of keywords (Gaviria-Marin et al., 2018).This is shown in the depiction of figure 4 regarding the cooccurrence data for terms.
Figure 3 is used to discover which phrases are used frequently based on the circle size.The combined mapping and grouping in Figure 4 gives an overview of the structure in ChatGPT research subjects related to education.Each group is represented by a different color, signifying the importance, affinity, and relatedness of each group differently.This indicates that "the density of elements reduces with the distance between them and the point of interest and increases with the number of adjacent elements" (van Eck & Waltman, 2010).According to the co-occurrence analysis (co-occurrence keywords) displayed in Figure 4, "ChatGPT," "education," and "artificial intelligence" are the phrases that appear the most frequently.The visualization overlay in Figure 5 can illustrate the extent to which keywords are growing from year to year, the more yellow the cluster means the newer the research.From Figure 4 it can be seen that the development of keywords is growing, this can be seen in most of the keywords that appear in yellow.These findings can be used as input for further research to further develop other keywords related to the use of ChatGPT in the education sector.
Figure 6 shows the density of related terms and the number of nodes that are close to one another (Aribowo, 2019).A yellow node showing the level of saturation indicates the number of phrases that appear frequently.The yellow node, where the keyword "ChatGPT" appears most frequently, represents the area where the greatest study has been done.Different topics, such 2. Number of paper."challenging," "teaching," and "knowledge," are covered in the green nodes.Green nodes show that a subject has not been thoroughly researched.
Based on the size of the yellow nodes, which suggest a frequently occurring issue, the editor will assess this subject to see if it is still relevant to the discipline mentioned by the reviewer.The review's outcomes won't be ideal, especially in terms of substance, if each reviewer's scholarly interests are not adequate.The inclusion of this topic gives editors evaluative information they may use to help the journal focus and become more specialized.Additionally, the editors choose Editors and Reviewers whose scientific specialization matches the issue shown in Figure 5 using this density visualization as a guide.

Citation
This analysis is used to determine which articles have the greatest impact on ChatGPT in the education sector.Table 1 provides further details on the top 10 cited on the topic of ChatGPT in education.
Table 1 shows the top 10 most cited articles from the time of its publication to the time of writing this paper, most citations are dominated by papers published in 2022.According to Table 1, the paper with the most citations, "OpenAI ChatGPT generated literature review: Digital twin in healthcare," was written by Aydın and Karaarslan and had a total of 13 citations.The purpose of Aydın and Karaarslan, the study was to investigate how Artificial Intelligence, in particular OpenAI's ChatGPT model, may be used to summarize and produce a literature review article on the use of Digital Twins in healthcare.The goal of the study was to illustrate the stage of OpenAI ChatGPT's use in producing a literature review article and to illustrate how AI might be able to hasten the gathering and expression of knowledge in academic publishing.The study's findings were encouraging, but the researchers cautioned that such advancements were still in their early stages and that additional research would be required to assess the academic validity of text generated by AI.
Numerous past research have examined or made predictions about potential future citation patterns.Machine learning approaches have been used in numerous studies, including Gradient Boosting Decision Tree (Sandulescu & Chiru, 2016), Support Vector Machine (Adankon et al., 2009), and XGBoost (Chen & Guestrin, 2016).The idea that works published for longer periods of time generally earn more citations has been portrayed by generative models in a number of citation prediction studies (Newman, 2009).
In addition, the reviewed papers are also classified based on the authors who have contributed to publishing their papers on the theme of ChatGPT in the education sector, as shown in Table 2. Table 3 shows the top most productive writers.
Based on the above table, there are 2 writers who wrote 3 paper and 2 paper titles published on the theme of ChatGPT in the education sector.The top writer, Benzon, WL., wrote 3 papers, the paper was titled "A Note about Story Grammars in ChatGPT," "ChatGPT Intimates a Tantalizing  Future; Its core LLM is Organized on Multiple Levels; and it has Broken the Idea of Thinking," and "Discursive Competence in ChatGPT, Part 1: Talking with Dragons".Second top writer, Azaria, A., wrote 2 papers, was titled "ChatGPT Usage and Limitations," and "ChatGPT: More Human-Like Than Computer-Like, but Not Necessarily in a Good Way".
ChatGPT is a language model for artificial intelligence that was released in November 2022 and offers conversational responses to inquiry prompts.Over 150 billion parameters are used to train the model using a combination of reinforcement learning algorithms and human input (Dowling & Lucey, 2023).The platform gained a million users in just its first week of public use, and because of the model's perceived output quality, it has been dubbed "the industry's next big disrupter" (Metz & Grant, 2022).ChatGPT has the potential to offer tailored learning experiences, create educational content, grade assignments, and even offer student counseling thanks to its capacity to produce human-like responses to text inputs.However, there are difficulties associated with integrating AI in education, including the concern that it would replace teachers and the need to guarantee the precision and dependability of AI-generated answers.Despite these challenges, implementing ChatGPT and other AI technologies has the potential to revolutionize the way education is provided and support teachers in giving students more tailored and efficient learning experiences.Education must work to strike a balance between enhancing education using AI's capability while also retaining the human touch and interpersonal communication that are so crucial to the transmission of knowledge.The use of ChatGPT in education has the potential to completely change how instruction is provided, but it must be done with care to ensure that it improves rather than hinders the learning process.

Conclusion
Co-authorship, co-occurrence, and citation are the three crucial factors in this study that were identified utilising a bibliometric analytical strategy.Research on the ChatGPT theme in the education sector has only appeared since 2022 and was taken from the Google Scholar database.We eliminated all irrelevant or repetitive studies; whose keywords are not in the abstract, keywords, or title; the correct document type or not (for example, a book chapter).Of the provisions regulated through keywords and other prohibitions, 93 articles were obtained.
ChatGPT is a powerful AI tool that has the potential to transform the education sector.Its ability to conduct tasks that require knowledge and creative intelligence, such as grading assignments and offering student counseling, has the potential to revolutionize the way education is provided.However, there are also challenges associated with integrating AI in education, including the need to guarantee the precision and reliability of AI-generated answers and concerns about replacing teachers.
Education must find a balance between employing AI to improve education and preserving the human touch and interpersonal communication that are so important to the transfer of knowledge.
It is advised that future research assess ChatGPT efficacy in enhancing student learning outcomes and consider the moral and legal ramifications of deploying AI in education.Furthermore, it's critical to consider how to incorporate ChatGPT and other AI technologies into education in a way that complements teachers rather than supplants them in order to ensure that learning is advanced rather than hampered.Assuring the accuracy and precision of AI-generated answers and investigating strategies to make sure that AI technologies are used in a way that is consistent with ethical and legal considerations should also be the focus of future research on the challenges associated with integrating AI in education.
It should be mentioned that this study has several limitations that need to be recognised and investigated further.Without using quantitative analysis, we depended on bibliometric analysis with several drawbacks.The first one was when we decided to rely on Google Scholar database instead of Scopus or Web or Science, with the consideration of collecting bigger samples.The second drawback is that the results were restricted to a certain time period and limited to keywords in English.Nevertheless, despite these limitations, this study offered a decent foundation for discussing adopters' doubts about employing chatbots, notably ChatGPT, in education.
Future research could concentrate on taking teaching practises "one step forward" by integrating ChatGPT, examining how humans and computers could collaborate to accomplish an educational goal, and examining the changes and outcomes brought to the education field.Academics should also carry out more formal study on this subject by extensively relying on bibliometric data, such as popular keywords and choosing references based on the authors, journals, and citations that have received the most citations.Researchers may build a more diverse bibliometric map to look at the most recent trends, the issues the industry is facing, and potential solutions by looking at more specialist sources like Web of Science or Scopus indexes.

Figure 3
Figure3has been shown in relation to the writers who are connected to a network.It is believed that more researchers would collaborate in the future to generate more high-quality study outcomes because there are still a few authors who are associated to one another.The authors focus on the ChatGPT in education issue in Figure3.