Revolutionizing educational landscapes: A systematic review of Metaverse applications, paradigms and emerging technologies

Abstract This paper addresses the existing gap in the literature on Metaverse applications in education by providing a comprehensive analysis of associated paradigms, platforms, hardware, and software, which have not been systematically reviewed yet. Notably, this study is among the first, to the best of our knowledge, to introduce the pioneering concept of Meta-AI in education and propose its potential to revolutionize the educational landscape by synergistically combining generative AI technologies, such as ChatGPT, with Metaverse education. Through this in-depth examination, our research makes a significant contribution to the field, establishing a solid foundation for the development of effective Metaverse teaching applications and guiding future scholarly endeavours.


Introduction
The integration of technology in education has been a transformative force, reshaping pedagogical approaches, learning experiences, and educational infrastructures.From the early days of computer-assisted learning (Razik, 1971) to the rise of e-learning platforms (Ally, 2004), technology has consistently expanded the horizons of what's possible in education.Immersive technologies, which include virtual reality (VR), augmented reality (AR), and mixed reality (MR), have emerged as powerful tools in this evolution, offering learners immersive, interactive, and engaging experiences (Bailenson, 2018;Merchant et al., 2014).For instance, VR has been employed in medical education for surgical simulations (Nicholson et al., 2006), while AR has found applications in history classes, allowing students to virtually experience historical events (Dunleavy et al., 2009).
In the evolving technological landscape, the term "Metaverse" distinctly emerges.While "immersive technologies" refer to tools enhancing digital immersion, the Metaverse signifies a fusion of these tools with cutting-edge technologies like blockchain and generative AI (Baidoo-Anu & Owusu Ansah, 2023).This fusion herald a digital experience that's not just immersive, but also interconnected, persistent, and comprehensive.The term "Metaverse" encapsulates this vision, emphasizing continuity, community, and complexity (Bell, 2008).As such, the Metaverse, bolstered by advancements like ChatGPT, has transcended mere entertainment, becoming a pivotal educational trend.In recent years, the Metaverse has been hailed as a promising educational trend with a thriving global research landscape (Dwivedi et al., 2022;Guo & Gao, 2022;Lee, 2020), leading many educators and academics to incorporate various Metaverse-related implementation scenarios into their teaching practices.
In 2007, the Acceleration Studies Foundation, a Metaverse research organization, took the initial step in developing the Metaverse roadmap, arguing that the Metaverse is a synthesis of both virtually enhanced physical reality and physically persistent virtual space (Zhang et al., 2022).The roadmap was built based on two axes: augmented reality versus simulation and intimate versus external (Smart et al., 2007), with scenarios including augmented reality, lifelogging, virtual worlds, and mirror worlds.Building upon the foundation laid by the Acceleration Studies Foundation and the Metaverse roadmap, Suzuki et al. (2020) note that the Metaverse has the potential to serve as a groundbreaking learning environment that can redefine educational experiences for learners worldwide.The Metaverse technology transcends the inherent limitations of traditional web-based, two-dimensional e-learning tools, such as lack of immersion, engagement, and motivation, and enables learners to access and participate in educational settings without time and location constraints (Sarıtaş & Topraklıkoğlu, 2022).In addition, the Metaverse also supports a more transformative learning experience by leveraging information and communication technologies as integral components of the educational process, in alignment with emerging paradigms (Demirer, 2013).This shift towards immersive, collaborative, and experiential learning environments enables educators to design and implement innovative instructional strategies that cater to a wide array of learning needs, preferences, and styles.
The rest of the paper is structured as follows.Section 2 recognizes research gaps and presents our research questions.Section 3 provides an outline of the research method, followed by the presentation of results in Section 4. Section 5 discusses the principal findings, limitations, and future research directions, and the final section offers a conclusion.

Research gap and our research questions
There is a dearth of comprehensive review articles on the use of the Metaverse in education.However, four influential articles have provided valuable insights into the application of immersive technologies in education.Wang et al. (2018) examines the evolution and implementation of Virtual Reality (VR) in construction engineering education and training, highlighting its progression from desktop-based to BIM-enabled VR, its diverse applications, and potential future directions, offering insights for enhancing training performance.The second paper is published by Ibanez and Delgado-Kloos (2018), summarizes the application of augmented AR in education.The third benchmark article, written by Radianti et al. (2020), provides a comprehensive review of the use of VR in education, while Tlili et al. (2022) focus specifically on summarizing the findings related to the Metaverse in education.Table 1 offers a comparative analysis of these four benchmark articles and the present study, examining the article type, time span, database, keywords, papers reviewed, Metaverse types, and educational levels included.The selection of these four articles as benchmark papers is grounded in their high impact and representativeness within the field.Each has been highly cited, indicating their significance and influence in shaping discussions around the Metaverse and immersive technologies in education.Moreover, while each article has its unique focus, collectively they provide a comprehensive overview of the landscape.Notably, the time spans covered by these articles cumulatively span from 1997 to 2022, ensuring a thorough and holistic review of the literature over this 25-year period.This comprehensive coverage further underscores the importance of these benchmark papers in our analysis.
However, these four benchmark articles do not provide a comprehensive review of factor research, paradigms, platforms and advancing technologies in Metaverse education, indicating a necessity for further exploration in these areas.Therefore, we raise five research questions: (1) What factors influence learners' willingness, efficiency and improvement of skills to adopt the Metaverse in education?
(2) What strategies, frameworks, and ecosystems support Metaverse teaching, and how can a comprehensive teaching paradigm be developed?
(3) What Metaverse platforms (commercial or self-developed) are applicable for teaching purposes?
(4) What are the various types of software used in Metaverse education, and what supportive hardware is necessary to facilitate their use?
(5) What are the potentials of generative AI and metaverse synergy in education?

Research method
This research aims to address specific questions by employing a systematic, explicit, and replicable search strategy.To identify relevant articles, a set of exclusion criteria was established, encompassing factors such as time span, article type, content, and topic.The study then conducts a statistical examination of these articles to identify common themes and emerging trends related to the application of the Metaverse in education, thereby providing a comprehensive understanding of this burgeoning field.
In this study, Web of Science and Scopus serve as the search databases.The initial search string, displayed in Table 2, is devised by combining Boolean operators.The keyword of the first search string is "Metaverse", while the second search string features three education-related terms: "education", 'teaching', 'learning'.Upon implementing the first two keyword groups, it was observed that a significant number of articles were focused on medical practice, health care, review and meta-analysis.To address this, the NOT operator shown in Table 2 was incorporated to exclude such articles, refining the search results to better align with the study's objectives.
In December 2020, our search process commenced, yielding an initial pool of 192 records from Web of Science and Scopus.To refine this selection, we set criteria to exclude articles published before 2013, those classified as reviews, and those not written in English.The intricate details of this selection process are depicted in Figure 1.
Upon acquiring the preliminary list, every abstract was meticulously examined by two independent researchers.This step was crucial to gauge the relevance of each article to our study's objectives.This dual-researcher approach was designed to ensure both a comprehensive and unbiased assessment.The researchers simultaneously read and assessed the abstracts, marking articles for inclusion or exclusion.In instances where there were discrepancies in their individual  selections, a mutual discussion was held to reach a consensus.On the rare occasion where a consensus couldn't be achieved through discussion, a third researcher was brought in for arbitration.This rigorous multi-layered scrutiny ensured the utmost precision in our selection process, ultimately leading us to a refined list of 90 articles that were most aligned with our research questions.

Results
In this section, we begin with a Bibliometric analysis, followed by addressing the five research questions outlined in Section 2. Our goal is to bridge the existing literature gap on the applications of the Metaverse in education, thereby promoting enhanced implementation and innovation in the domain.

Bibliometric analysis
In this section, we present a statistical analysis of the 90 articles included in our study, focusing on the geographical distribution of authors and the variety of publishers.

Countries
In our analysis of the geographical distribution of articles, we considered the country of origin for the first author.The study encompasses a total of 90 articles from various countries, as depicted in Figure 2.China and South Korea stand out as leading contributors in the field, with 30% and 20% of the articles, respectively.Conversely, France and Canada have the lowest representation among the included papers, with only 1% and 3%, respectively.The remaining countries, including Taiwan, Germany, Colombia, Spain, Singapore, Brazil, and the USA, demonstrate a moderate level of engagement with the topic, each contributing between 5% and 9% of the articles.This distribution signifies a growing global interest in exploring the potential of Metaverse applications in education.

Publisher
The sample of papers included in our study spans a diverse array of journals, showcasing the widespread interest in Metaverse applications.As illustrated in Figure 3, MDPI stands out as the most prolific publisher, with 24 articles related to the topic.IEEE follows with a considerable contribution of 12 articles, while Springer adds another nine articles to the mix.IGI Global has also made a notable impact with six articles on the subject.Elsevier and HINDAWI LTD each contribute five articles, demonstrating their engagement with the evolving educational landscape.WILEY and Frontiers, both well-regarded publishers, have published three articles each, further enriching the discourse on Metaverse education.The remaining 23 articles are spread across 22 other publishers, indicating that the topic has garnered attention from a wide range of sources within the academic community.

The connection and evolution relationship between key words
Figure 4 provides an in-depth visual representation of the primary keywords featured across the analyzed articles, revealing the complex interconnections and progressive associations among these terms.The most frequently mentioned keywords-'model,' "system," "student," "research," and "virtual world"-serve as the central pillars within the body of literature on Metaverse education.The prominence of these terms demonstrates the critical role of the Metaverse as both an abstract model and a tangible system, specifically designed for implementation within the virtual world.Moreover, this keyword analysis serves as a compass for navigating the ever-growing landscape of Metaverse education research, helping scholars identify key areas of interest and guiding future studies in this rapidly expanding field.

Factor research about education in Metaverse
Prior studies have highlighted the profound influence of Metaverse applications on multiple facets of students' educational journeys.Notably, themes such as students' willingness, efficiency, and their enhanced ability to embrace the Metaverse in education consistently surfaced in our review.These themes not only recurred across numerous studies but also stood out as pivotal areas of exploration, underscoring their importance in understanding the integration of the Metaverse in educational settings.In this section, we will delve into the factors influencing these research objectives by examining both experimental and questionnaire research.

Research by experiment
Table 3 provides an overview of the reviewed articles that employed experimental research methods, summarizing their research objectives, factors affected, influencing factors, and experimental design.
In the detailed review showcased in Table 3, the transformative potential of Metaverse applications on student learning experiences becomes palpably evident.Starting with the willingness domain, Guo and Gao (2022) explore the adoption of Metaverse in education, anchored in the theory that experiential situational teaching can significantly amplify interactivity, immersion, and cognition.Their utilization of the CNNs-RNNs model exemplified this theory, emphasizing its capability in identifying students' emotional EEGs during experiential English teaching.Complementing this, Gim et al. (2022) tackle willingness through a robust theoretical triad: the Self-determination theory, the Information systems success model, and the Technology acceptance model, providing insights into how intrinsic factors shape learning satisfaction in metaverse education.
Turning attention to learning efficiency, Chang et al. (2022) theorize that an amalgamation of intelligence network, situational learning, and project-based learning can enhance learning outcomes, as evidenced by their PBL virtual reality course.In a pivotal addition, Pigultong and Ieee (2022) spotlight the overarching influence of network speed on cognitive efficiency.Their one-way ANOVA experiment was not merely a technical endeavor; it underscored a foundational aspect of digital education-that the fluidity and speed of connectivity can significantly sway the efficacy of learning in the Metaverse.Their research resonates with the emerging consensus that as digital learning environments like the Metaverse gain prominence, foundational infrastructural elements like network speed can no longer be overlooked.
Lastly, in the sphere of ability improvement, J. Lee et al. (2022) postulate the transformative potential of biometric information, especially for those with Autism Spectrum Disorder (ASD).Their emphasis on the Metaverse-based social skills training program showcased the applicability of this This section summaries the articles conduct experiments on the impact of Metaverse applications on learning and education.According to the studies mentioned, our factor research reveals that factors influencing learners' willingness, efficiency, and skill improvement in adopting Metaverse education include interactivity, immersion, cognition, and technology acceptance.Studies show that experiential situational teaching, such as the ELAINE framework, enhances interpersonal effectiveness and learning effectiveness.Network speed also significantly impacts cognitive effectiveness, while self-determination and information system quality influence student satisfaction in VR education.

Research by questionnaire
This section presents the research objectives, factors affected and influencing factors of the reviewed articles that utilize questionnaire research methods (see Table 4).Some researchers focus on investigating users' intention to use Metaverse system (IU) in education (Akour et al., 2022;Alawadhi et al., 2022;Almarzouqi et al., 2022;Kim et al., 2022).However, there appear to be discrepancies between the reported observations.Akour et al. ( 2022) identify Perceived Usefulness (PU) to be an essential predictor of the factor of Users' Intention to Use the Metaverse System (MS) in Gulf area.However, in the United Arab Emirates (UAE) for medical-educational purposes, Alawadhi et al. (2022) document Perceived Ease of Use (PEOU) and PU significantly influenced the participants' intention to use the Metaverse technology, while Kim et al. (2022) report that perceived usefulness (PU), perceived enjoyment (PE) and perceived ease to use (PEU) impact IU in the Metaverse-based learning environment.Conversely, Almarzouqi et al. (2022) reveal the User Satisfaction (US) as the most vital determinant of IU.
Other scholars have focused on augmented reality (AR)-based applications in education.(Celik & Ersanli, 2022;Marini et al., 2022;Sofianidis, 2022;Sunardi et al., 2022).They consistently find that AR contributes positively to learning outcomes, learner motivation, and the overall learning experience, even extending to its acceptance in video conferencing environments.In addition, according to Lee and Hwang (2022), VR-Making experiences for instructional contents can also help pre-service teachers foster 4C abilities, including critical thinking, creativity, collaboration, and communication, as well as perceive pedagogical benefits.
Notably, Khansulivong et al. (2022) and Yang et al. (2022) find that technology adoption in online learning enhances user satisfaction and success, with crucial factors including habits and attitudes.Additionally, Teng et al. (2022) identify positive effects on satisfaction from performance and effort expectancy, social influence, and facilitating conditions in educational Metaverse platforms.Immersive media benefits include increased engagement and motivation (Erturk & Reynolds, 2020).Similarly, Social media also play a vital role in building interpersonal relationships among students (Frania et al., 2022).Furthermore, Yue (2022) suggests that popularizing Metaverse can enhance people's future usage confidence.Summarily, factors influencing learners' willingness, efficiency, and skill improvement when adopting Metaverse in education include perceived usefulness, ease of use, enjoyment, and user satisfaction.Other considerations are performance and effort expectancy, social influence, and facilitating conditions, all of which significantly impact learners' satisfaction and adoption of educational Metaverse platforms.These elements enhance motivation, engagement, and interaction within the learning experience.

Paradigms
The paradigms of Metaverse education are experiencing rapid growth, with an expanding array of strategies and frameworks emerging to enrich the learning experience, as summarized in Table 5. Ahmad et al. (2022) advocate for MOOCs 5.0, integrating Industry 4.0 technologies like the Metaverse for enhanced engagement and collaboration, while The EEDI paradigm emphasizes inter-disciplinary teaching and practical skills (Bijlani, 2022).Besides, Diaz et al. (2020) explore the use of Metaverse in education through Scrum, offering flexible knowledge acquisition.
Regarding the framework, Dahan et al. (2022) propose the E-Learning Environment based on the Metaverse (ELEM), enhancing the functionality of virtual learning environments.Liu et al. (2022) offer the Four-Layer Metaverse Architecture (FLMA) for developing virtual representations of learning scenarios.Additionally, Wang et al. (2022) present the IKRT framework, with hubs dedicated to instructional design, knowledge, research, technology, and training.Furthermore, Zhou and Kim (2022) proposes a smart education ecosystem-"four ecology integration" (RISC) -for the seamless integration of Metaverse and smart education.
To summarize, MOOCs 5.0, ELEM, FLMA, IKRT, and RISC are proposed strategies for the application of the Metaverse in education, emphasizing improved engagement, collaboration, and organic integration of Metaverse and smart education.

Platforms
Educational Metaverse platforms are generally divided into non-profit/institutional platforms and commercial platforms.Table 6 summarizes the way of use for all the platforms mentioned in this section.

Non-profit/institutional platforms
OpenSim software facilitates dynamic simulations of movement and is utilized for student evaluation in a Metaverse connected to Moodle and Sloodle (Alvaro-Farfan et al., 2019;Delp et al., 2007;Sghaier et al., 2022).Second Life, widely used in universities for virtual collaboration, also offers educational benefits but lacks goal-oriented learning (Cheong, 2010;Gregory, 2011;Zhu et al., 2007).VoRtex and iU-GJ provide collaborative learning environments (Jeong et al., 2022;Jovanovic & Milosavljevic, 2022), while the Erasmus + XR4Ped project promotes XR-based VR and AR in education (Mangina et al., 2022).USALSIM and VWFSEUC platforms are developed by the University of Salamanca and the University of Cundinamarca respectively to simulate 3D virtual environments and promote digital skills (Diaz, 2020;Lucas et al., 2013).

Commercial
Minecraft is a sandbox game that enhances social interaction and serves as a tool for projectbased learning, with successful integration relying on pedagogical approaches (Baek et al., 2020;Callaghan, 2016;Short, 2012).Zepeto improves satisfaction in online education through Metaverse classrooms (Lee et al., 2022), while Meta Horizon Workrooms boost engagement by integrating video conferencing and spatial audio for virtual instruction (Hedrick et al., 2022).
In summary, platforms for Metaverse education include non-profit/institutional ones, such as OpenSim, Second Life, VoRtex, Erasmus+ XR4Ped, iU-GJ, USALSIM, and VWFSEUC, and commercial platforms like Minecraft, Zepeto, and Meta Horizon Workrooms.These platforms support virtual collaboration, communication, 3-D content creation, and avatar-based interaction.Minecraft is used for project-based learning, while Zepeto and Meta Horizon Workrooms focus on virtual instruction and remote collaboration.OpenSim, Second Life, and VoRtex emphasize collaborative learning in virtual environments.

Hardware and software
This section is to provide a snapshot of the technological infrastructure that supports Metaverse learning environments, thereby helping readers gain a comprehensive understanding of the hardware and software that enable effective educational experiences in the Metaverse.Employing Metaverse teaching necessitates the use of hardware capable of presenting immersive environments while gathering data from learners.Four novel software are discussed in this section, and their functions and supportive hardware are summarized in Table 7.
In order to conduct water resources education, a unique 3D virtual reality architecture in the Metaverse (VRAM) is developed by Lo and Tsai (2022) through quasi-experimental research comparing learning that involves VRAM versus learning that does not.Students enter the immersive environment while wearing VRAM helmets to run VRAM Android applications and engage with 3D VRAM multimedia material in the Metaverse.Moreover, Lombeyda et al. (2022) present an Enhanced Reality Teaching Concierge (ERTC), an open networking hub which provides quick and simple connection across a broad range of services or applications for a wide range of users.It is intended to demonstrate 3D for educational purposes spanning web technologies, virtual reality, and even virtual worlds.Additionally, the concept of Distributed Interactive Whiteboard (DIWB) is proposed by Grandi et al. (2013) to describe an interactive whiteboard in the context of allowing instructors, students, and other potential community members to interact with the whiteboard whether they are physically there or not.The local classroom is transformed into a Metaverse, which is considered as a communal virtual shared space by interactions with telepresence rooms, mixed reality, and the Internet.Besides, an eye blinking system for avatars is introduced for a virtual problem-based learning (PBL) class in the Metaverse (Barry et al., 2015).It connects the way they blink their eyes with how they feel when they are asked to talk about different issues on behalf of their users.According to the findings, the challenging question would upset students' emotions and cause more eye blinks.With the help of this technology, teachers could thoroughly examine student replies to improve the results of virtual PBL.

Meta-AI
The advent of generative AI, represented by cutting-edge technologies like ChatGPT, calls for a thorough reevaluation of the educational landscape, urging us to investigate the role of AI in fostering transformative changes within teaching and learning practices, especially in relation to Metaverse education.In this context, we propose the novel concept of Meta-AI in education for the first time to the best of our knowledge and its potential to revolutionize education.Figure 5 demonstrates three generations of education and their specific characteristics.
Meta-AI in education refers to the integration of advanced AI technologies, such as generative language models, into the Metaverse to create highly immersive, interactive, and personalized learning environments.This fusion of AI and the Metaverse has the potential to reshape education through the following key aspects:

Augmented instruction
Recognizing the transformative impact of artificial intelligence in education, Popenici and Kerr (2017) argue that its integration into the Metaverse can significantly augment teachers' instructional capabilities, enabling them to concentrate on fostering meaningful connections with their students and facilitating higher-order thinking skills.By leveraging the power of AI, educators can receive invaluable support in various aspects of their work, such as lesson planning, content generation, assessment, and real-time feedback (Luckin et al., 2016).This shift in focus enables teachers to spend more time designing creative, impactful, and engaging learning experiences tailored to their students' diverse needs and interests.

Intelligent tutoring
Generative AI has been shown to deliver bespoke learning experiences, catering to the unique needs of individual students (Woolf, 2010).By analyzing learning profiles, preferences, performance, and interactions, AI can create tailored content, activities, and feedback.This approach not only promotes inclusivity but also enhances effectiveness by catering to diverse learning styles and abilities.Functioning as a virtual tutor, generative AI provides immediate assistance, guidance, and feedback, improving subject matter understanding and boosting students' confidence and motivation (Roll & Wylie, 2016).By addressing learners' specific requirements, AI-driven tutoring can accurately identify and fill knowledge gaps, leading to superior learning outcomes.

Self-directed learning
Generative AI can act as a comprehensive knowledge repository within the Metaverse, granting students immediate access to a vast array of information and resources.This instant, contextually relevant information provision aligns with the concept discussed by Roll and Wylie (2016), where AI encourages students to independently explore topics, fostering a self-directed learning approach.Such an approach nurtures essential research skills, a core component of selfregulated learning as outlined by Zimmerman (2002).This empowers students to take ownership of their educational journey, igniting a lifelong passion for learning.Furthermore, generative AI can assist students in developing critical 21st-century workforce skills, such as problem-solving, critical thinking, and collaboration.By cultivating these skills within the Metaverse's immersive and interactive environment, students can better prepare for future challenges and adapt to an everevolving world.

Principle findings
Factors influencing learners' willingness, efficiency, and skill improvement in adopting the Metaverse in education are multifaceted.Willingness is shaped by perceived usefulness, ease of use, enjoyment, user satisfaction, technology acceptance, and social influence (Akour et al., 2022;Alawadhi et al., 2022;Almarzouqi et al., 2022;Kim et al., 2015).Efficiency is shaped by internal drivers such as personal habits, attitudes (Yang et al., 2022), performance expectancy, and effort expectancy (Teng et al., 2022), while also being affected by external elements like network speed, the quality of information systems, social influence and facilitating conditions (Pigultong & Ieee, 2022;Z. Q.;Teng et al., 2022).Skill improvement is fostered by interactivity, immersion, cognition, augmented reality applications, immersive experiences, and pedagogical benefits that promote critical thinking and collaboration (Erturk & Reynolds, 2020;Frania et al., 2022;Lee & Hwang, 2022).These factors collectively contribute to the successful integration of the Metaverse in educational settings.
Traditional teaching strategies and methodologies such as MOOC 5.0 (Ahmad et al., 2022), EEDI (Bijlani, 2022), and SCRUM (Diaz et al., 2020) can be applied in Metaverse teaching practices to enhance students' engagement, collaboration, and learning efficiency.Moreover, researchers have specifically proposed ELEM (Dahan et al., 2022) and FLMA (Liu et al., 2022) frameworks for constructing enjoyable learning experiences, and the vision of the IKRT four-hub system (Wang et al., 2022) and RISC model (Zhou & Kim, 2022) provides guidance for building a sustainable teaching ecosystem.
Hardware devices that enable immersive experiences and facilitate the collection of user data are fundamental components in the realm of Metaverse education (Lo & Tsai, 2022).The capacity to foster more dynamic and engaging interactions and communication among users is a pivotal challenge in both software development and hardware design (Grandi et al., 2013).Furthermore, the Metaverse imposes stringent demands on information transmission (Barry et al., 2015).The implementation of suitable algorithms for anticipating user requirements and efficiently allocating computational resources will play a vital role in maintaining learner retention within the Metaverse educational environment.
Lastly, Meta-AI combines AI technologies like ChatGPT with the Metaverse to create immersive, personalized learning environments.Meta-AI revolutionizes education through augmented instruction (Luckin et al., 2016;Popenici & Kerr, 2017) and intelligent tutoring (Roll & Wylie, 2016;Woolf, 2010).Enhancing teacher capabilities and fostering self-directed learning, Meta-AI develops essential skills for students' success in the workforce.The fusion of generative AI and Metaverse education reshapes the educational landscape for more inclusive, engaging experiences.

Limitations
While our research provides a solid foundation for understanding Metaverse applications in education, we recognize certain limitations.The study relies on restricted databases (Web of Science and Scopus) and keyword searches, which might not capture all relevant Metaverse teaching papers.Researchers could use combinations of keywords like "Metaverse," "Augmented Reality," "External Reality," and "Virtual Reality" for a more comprehensive review.The study only includes English papers, potentially overlooking valuable research in other languages.Additionally, most papers reviewed are from computer-related journals, with few from top education journals.This technologyfocused trend highlights the need for further investigation into Metaverse teaching's impact on various education areas.Despite these limitations, our study offers valuable insights and potential research directions for scholars and professionals exploring Metaverse in education.

Future research direction
As the Metaverse continues to develop, the expansion of teaching applications within it parallels its growth.This includes the integration of different Metaverse scenes and the deepening incorporation of AI technologies, such as ChatGPT.This convergence will profoundly impact education, transforming teaching methods and learning experiences in the Metaverse.To successfully integrate these trends, educators and computer scientists must collaborate, experiment with new ideas, and develop innovative approaches, ultimately unlocking new teaching and learning opportunities.Three possible research directions to consider are:

Combining immersive Metaverse environments and generative AI
The integration of generative AI, like ChatGPT, with Metaverse education creates opportunities for personalized and adaptive learning experiences.This requires ongoing exploration of innovative teaching methods, effective strategies, and collaborative efforts between educators and computer scientists to ensure learners can fully benefit from their Metaverse-based education.The potential impact of generative AI on education is immense, and as technology continues to advance, the possibilities for transformative teaching and learning experiences in the Metaverse are limitless.

Merging virtual worlds and mirror worlds in education
Connecting these worlds demands the development of complex models, efficient algorithms, and accurate rendering of the real environment, including visual, auditory, olfactory, tactile, and gustatory perspectives.This combination offers great potential for immersive educational experiences that closely simulate the real world, enabling learners to intuitively integrate fragmented concepts and knowledge, thereby improving learning efficiency.As technology evolves, educators and computer scientists should work together to create innovative and effective ways of integrating these worlds, enriching the learning experience for students.

Integrating mirror worlds and augmented reality in education
The integration of mirror worlds and augmented reality offers remarkable opportunities for immersive education in the Metaverse.This fusion demands advanced sensors for collecting realworld information and synchronizing it with the mirror world, posing challenges for information synchronization, interaction, and data storage.This integration fosters a constructionist learning approach, enhancing learning experiences and enabling intuitive interactions within the virtual world.As these technologies progress, the potential for transformative Metaverse education grows increasingly substantial.

Conclusion
In conclusion, this paper has analyzed 90 articles focusing on the applications of the Metaverse in education.Our study underscores the complex factors influencing learners' willingness, efficiency, and skill development when adopting Metaverse education, as well as the potential of both traditional and innovative teaching strategies.Furthermore, the significance of creating robust Metaverse platforms and the crucial role of hardware and software are examined.Notably, we also propose the groundbreaking concept of Meta-AI and its potential impact on transforming education.Despite certain limitations, our research offers valuable insights into the current state of Metaverse applications in education and suggests future research directions.As the Metaverse continues to evolve, it is essential for educators to collaborate, innovate, and unlock new teaching and learning opportunities, ultimately reshaping the educational landscape and fostering more inclusive, engaging experiences for learners worldwide.

Figure 1 .
Figure 1.Three steps of removal of articles.

Figure 2 .
Figure 2. Number of included articles by country.
Figure 3. Number of included articles (n = 90) by publisher.

Figure 4 .
Figure 4. Map of keywords contained in the articles.

Table 3 . Summary of articles conducts empirical analysis (experiment) Research objective Affected factors Influencing factors Experiment design & model Paper
Nagendran et al. (2022)is realm,Nagendran et al. (2022)introduce the ground-breaking ELAINE framework.This framework doesn't merely merge Virtual Reality and AI; it heralds a new era of interactive learning.The theoretical underpinning of the ELAINE framework suggests that a wellorchestrated blend of VR and AI can create immersive learning landscapes, maximizing interpersonal effectiveness.Their research champions the idea that as we transition into more sophisticated digital learning paradigms, frameworks like ELAINE will become instrumental in shaping outcomes.