Abstract: The transition from general large models to educational large models is an important trend in the deepening development of artificial intelligence large model technology. Based on an analysis of the current status, typical cases, and potential challenges of educational large models, this article posits that educational large models are artificial intelligence models suitable for educational scenarios, characterized by ultra-large scale parameters, integrating general and specialized knowledge training. They are an integration of large model technology, knowledge base technology, and various intelligent educational technologies, capable of promoting the bidirectional construction of human learning and machine learning. Furthermore, it proposes an innovation architecture driven by applications and co-construction and sharing, as well as future application scenarios centered on learners, aiming to establish open interfaces between artificial intelligence large models and various digital educational applications. This will continuously train and improve educational scene models that can better address professional educational issues, forming a cluster of intelligent educational open models and knowledge bases that can be routinely used by a wide range of teachers and students, while simultaneously refining and extracting deep educational knowledge and addressing the risks and challenges in artificial intelligence educational applications.
Keywords: Educational Large Model; Generative Artificial Intelligence; Intelligent Education; Educational Big Data
Currently, artificial intelligence large model technologies represented by ChatGPT, Gemini, Wenxin Yiyan, and iFLYTEK Spark are developing rapidly and have attracted widespread attention globally. With their strong natural language processing capabilities, large models can perform complex tasks such as question answering, content creation, and code generation, showcasing their tremendous potential to liberate social productivity, which may profoundly impact human information acquisition methods, knowledge structures, and educational models. However, these general large models are not adept at solving specialized educational problems. The transition from general large models to specialized large models in the educational field is an inevitable trend in the deepening development of artificial intelligence large model technology. Educational large models are not merely fine-tuning and optimizing general large models; rather, they represent a systematic transformation aimed at reconstructing the future educational landscape, based on an open algorithm model architecture, with innovative educational application scenarios at their core. Clarifying the conceptual connotations of educational large models and designing a system architecture based on the essence of technology to further create new educational application scenarios has become a critical issue related to the digital transformation and intelligent upgrading of education.
// 1 Current Status of Educational Large Models
1 Definition and Analysis of Related Core Concepts
As an emerging research field, large models have generated many related concepts in academia, such as AIGC, generative artificial intelligence, and large models. Clarifying the connotations of these concepts is of great significance for deepening the understanding of educational large models and constructing high-quality educational large models.
① AIGC (Artificial Intelligence Generated Content) can be directly translated as “artificial intelligence generated content,” which is a concept proposed in contrast to professionally generated content (PGC) and user-generated content (UGC). AIGC is based on intelligent technologies such as supervised learning, reinforcement learning, pre-trained models, and natural language processing, which automatically generate various forms of content, such as text, images, audio and video, and 3D interactive content through learning and training on existing data[1].
② Generative artificial intelligence is an artificial intelligence technology that automatically generates response content based on natural language dialogue prompts (Prompt)[2]. The technical implementation process of generative artificial intelligence typically consists of two steps: first, training or learning based on existing data (pre-training); then, when new instructions or commands are input, automatically generating new content based on the learned intentions.
③ Large models refer to artificial intelligence models with billions to hundreds of billions or even more trainable parameters. They are the product of the joint development of deep learning, GPU hardware, large-scale datasets, and other intelligent technologies. The powerful capabilities exhibited by large models are essentially the result of the “quantitative change leading to qualitative change” in artificial intelligence algorithms, a process vividly referred to as “intelligent emergence ability,” which is the capacity to automatically learn and discover new, higher-level features and patterns from the original training data[3]. These capabilities are prominently manifested in strong and general user intention understanding, contextually continuous dialogue ability, intelligent interaction correction ability, and new content generation ability.
In summary, the concepts of AIGC, generative artificial intelligence, and large models are closely related, all emphasizing the leap of the new generation of artificial intelligence technology from passive design to active production, representing an evolutionary trend of a new round of technological revolution. However, each of the three has its focus on technical characteristics: AIGC emphasizes the diversity of generated content types, generative artificial intelligence emphasizes autonomous generation and creativity, and large models emphasize the parameter characteristics of algorithm models.
Based on this, this study believes that educational large models are artificial intelligence models suitable for educational scenarios, characterized by ultra-large scale parameters, integrating general and specialized knowledge training. They are an integration of large model technology, knowledge base technology, and various intelligent educational technologies, capable of promoting the bidirectional construction of human learning and machine learning. They not only contain the educational knowledge necessary for the inheritance of human civilization but also refine educational experiences and methods that previously existed only in teachers’ minds, guiding learners to think deeply in human-machine dialogue interactions, providing guidance and support for learners’ independent exploration, and continuously updating and upgrading in the process to achieve a higher level of professionalism.
2 Research Dynamics of Educational Large Models
Educational large models have immense transformative potential, not only prompting new understandings and reflections on education and teaching but also profoundly changing educational concepts, content, and models, potentially even reshaping the form of school education[4]. Some scholars believe that generative artificial intelligence, through human-machine co-teaching, inclusive intelligence, and interactive evaluation, can promote high-quality educational development[5], giving rise to educational teaching concepts that are “more open and inclusive, interdisciplinary integration,” forming a diversified AI teaching system that is “human-centered, education-assisted”[6]; others believe that educational large models will reconstruct the structure of school education, gradually shifting standardized evaluation methods towards personalized evaluation standards, forming a lifelong “credit bank”[7].
At the same time, educational large models may also bring potential risks such as intellectual property disputes, data usage biases, and algorithm abuses. The quality and safety of generated content still need improvement, with common issues of lack of coherence and logic. Therefore, not all educational scenarios are suitable for using large models[8]. Some scholars point out that educational large models may lead to imbalances in human-machine relationships, bringing technology dependence, data security, and other ethical risks[9][10].
In summary, the rapid development of artificial intelligence presents significant opportunities for education but also poses a series of unknown risks. How to effectively support educational reform and innovation with educational large models, developing warm intelligent education, has become a common challenge faced by global education.
3 Current Applications of Educational Large Models
Currently, countries around the world attach great importance to the exploratory application of educational large models. The United States released “Artificial Intelligence and the Future of Teaching: Insights and Recommendations,” summarizing the opportunities and risks of artificial intelligence in teaching, learning, assessment, and research, and proposing seven action recommendations for the application of the next generation of artificial intelligence in teaching and learning[11]. The United Kingdom released “The Application of Generative Artificial Intelligence in Education,” suggesting that the education sector should fully utilize various new technologies to provide high-quality education for learners, equipping them with the abilities needed to adapt to social development[12]. Meanwhile, schools have also attempted corresponding practical explorations. In September 2023, Hong Kong, China, developed an artificial intelligence curriculum for junior high school students, requiring public schools to offer 10-14 hours of AI courses covering topics such as ChatGPT, artificial intelligence ethics, and the social impact of artificial intelligence. In October 2023, Japan’s Ministry of Education, Culture, Sports, Science and Technology announced 53 elementary and middle schools as pilot schools for generative artificial intelligence, using new technologies to improve the efficiency of educational activities and school management[13]. Australia announced that starting in 2024, artificial intelligence, including ChatGPT, will be allowed in all schools[14]. These practical explorations highlight the important role of the new generation of artificial intelligence technology in education, underscoring the unstoppable trend of digital transformation and intelligent upgrading in education.
4 Analysis of Typical Cases of Educational Large Models
Globally, educational large models are undergoing extensive and in-depth exploratory development, having formed solutions in areas such as oral practice, mathematics learning, emotional analysis, and personalized recommendations. This study outlines five typical application cases of educational large models (as shown in Table 1), analyzing their application scenarios, technical progress, and existing shortcomings.
Table 1 Typical Application Cases of Educational Large Models
From the application scenarios, Spark Language Partner is mainly used for language learning, supporting real-time translation of multilingual text, speech, and images, as well as correcting grammar errors and providing oral practice. EmoGPT is used for psychological counseling, capable of recognizing and responding to user emotions, providing ongoing psychological support. MathGPT is aimed at global math enthusiasts and research institutions, providing algorithms for problem-solving and explanations, supporting users in mathematical problem-solving and practice. Zhihai-Sanle is used for AI knowledge learning, providing functions such as search engines, calculation engines, and local knowledge bases, supporting intelligent Q&A and question generation. Khanmigo provides personalized learning plans for learners through a conversational AI chatbot, covering subjects such as mathematics and science.
In terms of technical progress, educational large models show significant advantages in model performance, application scenarios, and technical characteristics, covering most subject content, primarily focusing on autonomous learning scenarios, including knowledge Q&A, language learning, learning guidance, and teaching assistance. In terms of technical routes, the “general + fine-tuning” path has proven effective, with many technical solutions based on general large models achieving effective responses to specific subject knowledge through instruction fine-tuning.
In terms of existing shortcomings, current educational large models are limited in accuracy, diversity of teaching content, support for core educational scenarios, and inclusivity for diverse learners, facing issues such as high error rates and a lack of empathetic understanding abilities. They primarily focus on subject knowledge teaching and exam-oriented education contexts, still lacking in cross-disciplinary learning and the cultivation of learners’ comprehensive abilities and higher-order thinking. They mainly focus on supporting autonomous learning, and effective exploration of how to fully leverage large models in real classroom settings, peer collaboration, and blended teaching scenarios has yet to be conducted.
In summary, the application of large models in the field of education has made significant progress, but still faces practical issues. There is a need to further improve the quality and scale of training data, especially embedding advanced educational concepts, deep educational knowledge, and the real needs of core educational scenarios into the technical design, combining user feedback for multiple iterations to form a more intelligent and flexible educational large model.
// 2 Major Challenges Faced by Educational Large Models
The continuous updating and upgrading of artificial intelligence technology is driving the large-scale application of large models. Deploying and establishing vertical domain large models has already demonstrated significant effectiveness in scenarios such as intelligent customer service, digital assistants, and multimodal retrieval, and will further deeply integrate into various fields and links of economic and social development, empowering intelligent upgrades across industries and aiding the leap in social productivity. However, compared to other fields, achieving true automation and intelligence in education often faces higher requirements because most educational tasks are “non-programmable,” making automation more challenging[15], while also facing a series of severe challenges in terms of capabilities, values, data, and algorithms.
At the capability level, educational large models possess strong content generation and creativity capabilities, capable of directly providing answers to questions. However, over-reliance on large models can lead to cognitive inertia among teachers and students, weakening their problem-solving abilities and further exacerbating the passive, superficial, and fragmented nature of knowledge acquisition. Over time, this will lead to the degradation of human cognitive abilities. In reality, educational large models are merely simulations of human cognitive abilities and do not possess true wisdom or the ability to “solve the unknown” or “innovate knowledge.” It is crucial to guide teachers and students to develop the capability and literacy to harness educational large models. At the value level, the value deviations brought by educational large models may lead to the emergence of “hallucination” phenomena, generating erroneous or non-existent content. If the training data carries a certain value orientation, then the corpus that aligns with that value will be repeated, thus being recognized by the converter as key text and output as standard answers. Some large models based on Western corpus may use content products such as text, images, and videos to obscure and subtly project “value and cultural standard answers,” imperceptibly infiltrating values into adolescents and leading to the upgrading of “digital colonialism” for disadvantaged groups[16]. Therefore, it is essential to strengthen educational goals and value guidance, emphasize “value alignment,” and establish corresponding risk prevention and intervention mechanisms. At the data level, educational large models require massive amounts of training data, which further expands the risks of data security and privacy protection. Privacy issues for teachers and students will become an unprecedented challenge. We must strengthen the security guarantees for educational data, encrypt and decrypt raw data without compromising the quality of content generated by large models to prevent the leakage of privacy data from teachers and students, while also establishing effective mechanisms for co-building and sharing educational data, expanding high-quality public training data resources, and promoting the healthy and sustainable development of educational large models. At the algorithm level, educational large models based on deep learning are often regarded as “black boxes,” making it difficult to explain their decision-making processes, potentially leading to educational behaviors that are difficult to understand and accept, even undermining the learner’s subject position. From a certain perspective, personalized learning algorithms seem to improve the accuracy of information delivery, but they may also trap learners in “information cocoons,” where they can only see information that aligns with their existing viewpoints, leading to increasingly narrow perspectives and further impacting learners’ comprehensive development.
Currently, educational large models are at a critical stage of research and application, requiring further improvement in the quality and scale of training data, especially embedding advanced educational concepts, deep educational knowledge, and the real needs of core educational scenarios into the underlying architecture of algorithm models, and updating and iterating based on learner needs to achieve the practical implementation of educational large models.
//3 Innovative Architecture of Educational Large Models
Currently, the research and development of educational large models mainly adopts two technical routes: one is to directly call general large models, enabling them to possess certain professional capabilities through fine-tuning or prompt learning; the other is to use specialized data in the education field to train large models specifically designed to solve educational tasks. However, although both technical routes have made certain progress, the effectiveness still needs to be improved. The issue lies in the lack of sufficient professional data for training, coupled with insufficient deep knowledge in the education field, resulting in the current large models lacking intelligence and being unable to flexibly handle complex and variable educational tasks. The key to developing educational large models lies in integrating these two technical routes. This is not a simple addition, but rather a method of continuously obtaining data from normalized educational applications through open data interfaces, achieving an organic combination of “large models” and “small models,” “big data” and “small data,” to meet the actual needs of teachers and students in daily teaching, breaking down “data silos”[17]; at the same time, using expert knowledge bases as a supplement to large models[18], consciously “teaching” educational knowledge and pedagogy to large models, and integrating various intelligent educational technologies to form dedicated large models capable of flexibly handling various complex educational tasks.
1 Underlying Logic
The core competitiveness of educational large models lies not in technology or data, but in a deep understanding of education. The “learner-centered” concept should be regarded as the underlying logic for developing educational large models, integrated into the entire process of algorithm model architecture design and prototype development. “Learner-centered” means aiming at the proactive and creative development of learners, configuring educational resources around learners’ needs, interests, and abilities, designing learning activities, and planning growth paths to achieve large-scale differentiated instruction. Guided by this concept, educational large models are no longer cold machines or tools, but important assistants and collaborative entities that promote learning, optimize learning, and stimulate learning, helping learners transition from passive recipients of knowledge to active seekers, explorers, and collaborators of knowledge. However, this “learner-centered” approach should not turn into “precise question answering” and must not fall into the trap of “efficient exam-oriented” approaches. Instead, it should adhere to the idea that “every learner is a person of comprehensive development, a multi-dimensional person,” utilizing educational large models to understand learners’ growth states, providing personalized, adaptive, and warm learning support and teaching guidance services that promote learners’ comprehensive and individualized development.
2 Open Innovative Architecture
The educational large model is based on general large models and connects various digital educational applications, continuously training educational scene models, and constantly improving the ability to solve educational professional tasks. The open innovative architecture of educational large models is divided into three layers: basic capability layer (L0), professional capability layer (L1), and application service layer (L2), as shown in Figure 1.
Figure 1 Open Innovative Architecture of Educational Large Models
L0: Basic Capability Layer. This layer includes large language models, video analysis models, subject large models, and emotional computing models. Among them, large language models are responsible for processing text data; video analysis models handle video data, such as classroom recordings; subject large models deal with subject-specific tasks; and emotional computing models handle indicators related to physical and mental well-being, involving tasks such as psychological health, emotional monitoring during the learning process, and emotional analysis of interpersonal interactions. During task completion, multiple large models work collaboratively and supportively, with the task center integrating and processing the outputs from different models.
L1: Professional Capability Layer. This layer consists of two parts: ① Educational Scene Model Library. The educational scene model library mainly includes models for learning behavior analysis, classroom interaction analysis, ability assessment, academic prediction, emotional computing, and decision support, with a portion of commonly used models pre-configured and continuously optimized and expanded during application. ② Expert Knowledge Base. The expert knowledge base contains two types of knowledge: subject content knowledge and pedagogical knowledge. The two types of knowledge are integrated and stored and presented in the form of a multi-dimensional dynamic knowledge graph. As the teaching process continues to develop, teachers and students are both users of the knowledge graph and co-editors and creators, ultimately forming personal knowledge graphs for learners as well as shared knowledge graphs at different levels such as classes, schools, and regions.
L2: Application Service Layer. The most important innovative concept of educational large models is “application-driven,” meaning that various digital educational applications are connected to the large models. While the large models empower the applications, application data is continuously fed back into the large models, continuously enhancing the educational professional capabilities of the large models. These applications cover various educational scenarios such as teaching, learning, assessment, and management, forming high-quality training data with unified standards through open data interfaces. Meanwhile, teacher and student users can issue task instructions through a unified usage portal, and the large models automatically call the corresponding functional modules based on the nature of the tasks, forming a learner-centered application model that allows for seamless use of large models even without any knowledge of artificial intelligence.
3 Construction and Deployment Ideas
Educational large models are not a singular, closed model but rather a process in which developers and users participate and continuously improve. Professional teams develop the foundational architecture and core components of the models, while various users contribute to optimization during application. Teachers, students, and various developers of educational digital applications are both users and contributors/builders of educational large models, thereby forming a co-constructed and shared innovative ecosystem for intelligent education. This ecosystem will undergo two important phases: ① In the foundational construction phase, based on a “data + knowledge” dual-driven artificial intelligence technology route, various educational AI technologies will be integrated to establish a model system, application system, and data system, forming a continuous training mechanism for models. From a technical implementation perspective, this phase includes seven steps: large-scale multi-dimensional educational data collection, data preprocessing, feature engineering, model design, model pre-training, fine-tuning and transfer learning, and model evaluation and optimization. ② In the application improvement phase, educational large models will continuously innovate algorithms and models, innovate data applications, and innovate application development.
//4 Application Prospects of Educational Large Models
Educational large models will promote the digital transformation and intelligent upgrading of education from three aspects: learning spaces, learning resources, and the roles of teachers, forming a new ecology of education characterized by human-machine collaboration and coexistence.
1 Interactive Generation of Learning Spaces
With the support of educational large models, learners obtain learning support and create learning outcomes through human-machine interaction, constructing personal and collective learning spaces, forming learning scenarios that integrate physical and online spaces, allowing all learners to access any information they need at any place and time[19]. On one hand, learners utilize tools such as knowledge graphs and digital textbooks to organize and create collections of learning outcomes, establish personal and team knowledge bases, and collaboratively write “digital learning cases,” forming a learning community based on the normalized co-construction and sharing of learning resources, while developing user evaluation and identification mechanisms for learning resources, transitioning from knowledge consumption learning to knowledge creation learning. On the other hand, learners’ learning experiences are extracted from learning behavior data by intelligent algorithms, summarized into new learning methods, continually optimizing and refining educational teaching strategy models and knowledge bases, achieving mutual empowerment of human learning and machine learning.
2 On-Demand Supply of Learning Resources
Leveraging the learning analytics capabilities of educational large models, the gap between the demand and supply sides of educational resources can be narrowed, providing personalized learning resources for learners and addressing the issue of matching high-quality educational resource supply with learning needs. On one hand, a classification system for learning needs in the resource application process should be established. Based on known resource classifications, learners can label resource types, continuously expanding and enriching the resource classification framework, making resource labeling closer to real learning needs. From a technical perspective, this process is essentially a “alignment” between educational large models and human value orientations, ensuring that large models adhere to human values, preferences, and ethical principles, thereby providing learners with massive and appropriate learning resource support. On the other hand, a user-centered evaluation mechanism for educational resources should be established, promoting the survival of the fittest among educational resources based on feedback from teachers and students, stimulating the enthusiasm of teachers and students to apply resources, and discovering and cultivating a group of quality educational resource builders, promoting the transition of “educational specific resources” to “educational large resources.”
3 Transformation and Upgrade of Teacher Roles
Educational large models will gradually replace repetitive and inefficient educational labor, enhancing the scientific and creative nature of educational work, and pushing teachers to transition from “experts in teaching” to “experts in learning”[20], providing personalized support for every learner through creative teaching design. On one hand, relying on intelligent knowledge bases and application sets, large-scale and precise knowledge transmission will be conducted, freeing teachers’ time, energy, and creativity, allowing them to focus on organizing learning and guiding activities, fostering deeper teacher-student dialogues. On the other hand, teachers will gradually become learning guides and teaching researchers, likely understanding learners’ thought processes during learning through human-machine collaborative learning data analysis and diagnosis, and designing targeted teaching activities based on learners’ actual learning states, promoting experiential teaching to gradually transition to evidence-based educational professional practice activities. In the future, teachers will be both experienced and wise practitioners and researchers adept at utilizing big data analysis and intelligent teaching research tools, further exploring new educational laws and teaching methods under the conditions of artificial intelligence, ultimately forming a new knowledge system for intelligent education.
// 5 Conclusion
Currently, there is still a gap between the educational large model technology in our country and the internationally leading level, but the development momentum is good, having accumulated a certain technological strength, and possessing significant advantages in massive data, capable of accurately capturing and deeply understanding the learning process based on large-scale, multimodal, and long-cycle educational data, further clarifying the underlying mechanisms of teaching and learning, promoting the rapid iteration of educational algorithms, and establishing large language models that are more targeted, professional, and accurate, achieving a leapfrog development. At the same time, educational large models will bring a series of new challenges and unknown risks, and it is necessary to clarify their development principles and usage scope as soon as possible, strengthen ethical risk assessment and review, and formulate targeted usage guidelines for teachers and students to ensure that the principles of educational equity, inclusiveness, and sustainable development are evident throughout the entire life cycle of the development and application of educational large models.
References
[1] Jiang Sha, Zhao Mingfeng, Zhang Gaoyi. A Brief Analysis of the Application Progress of Generative Artificial Intelligence (AIGC) [J]. Mobile Communication, 2023, (12): 71-78.
[2] Miao F C, Holmes W. Guidance for generative AI in education and research [OL]. <https://unesdoc.unesco.org/ark:/48223/pf0000386693>
[3] Wei J, Tay Y, Bommasani R, et al. Emergent abilities of large language models [OL]. <https://arxiv.org/pdf/2206.07682.pdf>
[4] Li Yongzhi. Opening New Paths for Educational Development through Digitization [N]. People’s Daily, 2023-10-13(9).
[5] Li Yanyan, Zheng Yafeng. Educational Applications of Generative Artificial Intelligence [J]. People’s Forum, 2023, (23): 69-72.
[6] Chen Xiaohong, Yang Ningyi, Zhou Yanju, et al. A Review of the Impact of AIGC Technology on Education and Employment Markets in the Digital Economy Era—Taking ChatGPT as an Example [J]. Systems Engineering Theory and Practice, 2024, (1): 1-13.
[7] Zhu Yongxin, Yang Fan. ChatGPT/Generative Artificial Intelligence and Educational Innovation: Opportunities, Challenges, and the Future [J]. Journal of East China Normal University (Education Science Edition), 2023, (7): 1-14.
[8] Lu Yu, Yu Jinglei, Chen Penghe, et al. Educational Applications and Prospects of Generative Artificial Intelligence—Taking the ChatGPT System as an Example [J]. China Distance Education, 2023, (4): 24-31.
[9] Yang Junfeng, Shen Zhongqi, Chen Ruining. Educational Applications and Ethical Risks of Generative Artificial Intelligence [J]. Huzhou Normal University Journal, 2024, (1): 1-8.
[10] Chen Qianqian, Zhang Lixin. Ethical Reflections on Educational Artificial Intelligence: Phenomenon Analysis and Vision Construction—Based on the Perspective of “Human-Machine Collaboration” [J]. Journal of Distance Education, 2023, (3): 104-112.
[11] US Department of Education. Artificial intelligence and the future of teaching and learning: Insights and recommendations [OL]. <https://tech.ed.gov/ai-future-of-teaching-and-learning/>
[12] UK Government. Generative artificial intelligence (AI) in education [OL]. <https://www.gov.uk/government/publications/generative-artificial-intelligence-in-education>
[13] Jia Yun. Japan Establishes Pilot Schools for Generative Artificial Intelligence [J]. International Educational Exchange, 2023, (6): 78-79.
[14] Australian Government Department of Education. Australian framework for generative artificial intelligence (AI) in schools [OL]. <https://www.education.gov.au/schooling/resources/australian-framework-generative-artificial-intelligence-ai-schools>
[15] Autor D H, Levy F, Murnane R J. The skill content of recent technological change: An empirical exploration [J]. The Quarterly Journal of Economics, 2023, (4): 1279-1333.
[16] Miao Fengchun. Principles of Generative Artificial Intelligence Technology and Its Educational Applicability [J]. Modern Educational Technology, 2023, (11): 5-18.
[17] Sang Xinmin, Xie Yangbin, Yu Zhong, et al. System Engineering Discussion on Educational Digital Transformation [J]. Modern Educational Technology, 2023, (1): 5-16.
[18] Wei Bin. Analysis of the Integration Path of Symbolism and Connectionism in Artificial Intelligence [J]. Research on Dialectical Materialism, 2022, (2): 23-29.
[19] Cao Peijie. Smart Education: Educational Reform in the Era of Artificial Intelligence [J]. Educational Research, 2018, (8): 121-128.
[20] Cao Peijie. The Triple Realms of Educational Reform through Artificial Intelligence [J]. Educational Research, 2020, (2): 143-150.
Article cited from: Cao Peijie, Xie Yangbin, Wu Huizi, Yang Yuanyuan, Shen Yuan, Zuo Xiaomei, Huang Baozhong. Current Status, Innovative Architecture and Application Prospects of Educational Large Models [J]. Modern Educational Technology, 2024, 34(02): 5-12.
