Historical expert commentary, excerpt edition. The Chair first carried this text on 22 September 2024. References to school access concern the review of intelligent products for use in schools. The review arrangements, oversight mechanisms and standards discussed are the author's proposals in their 2024 context.
Author: Huang Ronghuai. Roles stated in the historical image caption: Director of the Smart Learning Institute of Beijing Normal University and Director of the National Engineering Research Center of Cyberlearning and Intelligent Technology.
Article contents
Large AI models exhibit scale, generalisation, emergence and versatility. Their powerful capabilities in generating multimodal text, images, audio and video, conducting contextual dialogue, reasoning logically, calculating with data and handling tasks are accelerating intelligent transformation across industries and creating new opportunities for deeper educational digital transformation. Understanding the changes in educational conceptions and forms arising from these models, and enabling their reasonable, orderly and safe integration into the education ecosystem, is a key issue for education in the intelligent era.
The iterative evolution of large AI models is awakening education systems to systemic transformation
Global scientific and technological innovation has entered an exceptionally intensive and active period. A new technological revolution and industrial transformation are redrawing the global innovation landscape and reshaping the world economy. Continually evolving disruptive technologies profoundly affect countries' futures and people's well-being.
Education is actively adapting to the new wave brought by large AI models. One clear trend in the current transformation is that education systems are at a critical stage, moving from passively accepting external change towards actively adapting and driving internal transformation. This marks education's self-awakening: the system is becoming an active leader of change as well as responding to external forces such as technology and capital. Education authorities need to understand the educational logic behind technology, alongside its direct applications, to prepare people for future development.
First, higher education is at the forefront of change. Institutions are rapidly integrating new technologies, using AI teaching assistants and intelligently connected learning environments to change how content is delivered and to innovate in teaching organisation. They are also promoting interdisciplinary research and learning, using AI to renew research paradigms and advance the integration and innovative application of knowledge.
Second, the need to transform vocational education is even more pressing. As demand for skilled workers continues to grow, vocational education is a crucial bridge between education and employment. Institutions and government departments are increasing investment and strengthening links with industry so that curricula respond to market needs.
Finally, basic education is steadily advancing its digital transformation under strong policy guidance. Governments at different levels are providing supportive policies and funding, including building and using smart education platforms, using big data and AI to improve teaching and management, and strengthening digital literacy training for teachers and students.
Deeper application of large AI models creates new forms of human–AI collaborative education
From the perspective of human–AI collaboration, large AI models create opportunities for systemic changes in educational forms. Their contribution is particularly evident in collaborative teaching, collaborative learning and collaborative decision-making between people and machines.
Develop diverse, cross-domain human–AI collaborative teaching.
Large AI models will further accelerate the transition from the traditional teacher–student structure towards a teacher–student–machine structure. Relationships between people and machines in educational practice are becoming more complex, with coexistence and multidimensional interaction changing teaching towards human–AI co-teaching.
First, teaching scenarios expand. Large AI models can bring the advantages of machine intelligence to particular teaching contexts and address bottlenecks in educational applications. For example, combining humanoid robots with large models creates embodied intelligence that can improve machines' multimodal perception of the physical world, their understanding and adaptation, and their autonomous planning and decision-making, enabling intelligent interaction with real environments. In teaching agents, model-driven intelligent learning platforms and tutoring systems can overcome constraints of time, place and teaching, providing efficient, immediate and adaptive learning support.
Second, collaboration becomes more advanced. As technology can handle a wider range of tasks, teaching in the intelligent era is entering a collaborative field in which teachers' and machines' intelligence coexist and jointly guide and control activity. This is changing teachers' roles, their authority in relation to knowledge and the working paradigm of instructional design.
Develop mutually empowering human–AI collaborative learning.
Educational applications of large AI models will gradually change how learners learn. Alongside their own cognitive processing, learners will call on machine intelligence through cognitive offloading and improve that intelligence through iterative feedback, creating a form of human–AI co-learning.
First, learning experiences improve. Large models raise the intelligence of human–machine interaction. Machines can act as tutors, learning companions and assistants, naturally collecting and analysing multimodal learning-process data and identifying implicit emotional changes and cognitive states. This can support immediate emotional companionship, precisely targeted learning assistance and personalised learning pathways.
Second, learning mechanisms change. When learners and machines complete activities or tasks together, machines begin to share cognitive work previously performed entirely by the human brain. Learners can use large AI models as an external cognitive resource to understand and process learning content collaboratively, achieving complementary enhancement of human and machine intelligence.
Develop safe, trustworthy human–AI collaborative decision-making.
Successive breakthroughs in large AI models will further upgrade educational governance, moving it towards higher levels of evidence-based decision-making, data-driven practice and human–AI collaboration. Governance processes become more scientific, precise and intelligent, taking on the characteristics of collaborative decision-making.
First, governance methods change. Large AI models can perceive and extract vast amounts of educational information and process it across modalities, identifying patterns hidden in educational data. This supports dynamic monitoring, evaluation and feedback across education systems and gives decision-makers timely, comprehensive and objective analysis.
Second, governance thinking changes. Drawing on large models' advantages, training on and analysing extensive educational data can support predictions about educational elements and processes and guide actual practice. This can help shift governance from passively responding to events towards anticipating them proactively.
Key initiatives for building a smart education ecosystem with large AI models
Continue strengthening teachers' and students' competencies for the intelligent era.
To respond to technology's effects on employment and career prospects, priority should be given to developing these competencies and improving teachers' and students' digital creativity, adaptability and ability to navigate the digital world.
For students:
First, improve mechanisms for developing digital literacy and skills. Deliver information technology courses effectively and establish systems to assess and monitor digital literacy, tracing students' progress over time and accurately describing the development of their literacy and skills.
Second, strengthen self-directed learning. Encourage teachers to introduce active learning approaches in class and develop students' habits of planning, monitoring and evaluating their own learning.
Third, improve learning with intelligent technologies. Large AI models can help students acquire knowledge, learn independently and benefit from learning companionship. They can support course review and new learning, provide positive experiences, and help manage learning time, tasks and processes.
For teachers:
First, encourage active exploration and use of intelligent technologies. Teachers' engagement can move through four stages: awareness, experience, practice and dissemination. Teachers should be encouraged to learn about AI principles, use technology to support their own learning and improve teaching, and exchange practical experience.
Second, develop awareness of changing professional roles. Schools should provide digital literacy and skills training suited to teachers' individual needs, strengthen their capacity for human–AI collaborative teaching, and guide them to take responsibility for students' emotional support and character development.
Third, use targeted teaching research to support professional development. Intelligent technologies can collect and analyse educational data of different types, sources and dimensions to identify needs in pedagogy, subject knowledge and technology use. Summaries and analytical reports on teaching processes can help teachers conduct focused human–AI collaborative inquiry, understand classroom practice, reflect on educational design and improve teaching competence.
Improve mechanisms for approving intelligent products for use in schools.
First, establish comprehensive, differentiated reviews of AI educational technology products. This includes product classification and risk grading, detailed identification of products and subject areas, and analysis of the connected times, places, people, resources, functions and disciplines involved in their use. Rigorous review and adaptation should help products fit diverse educational settings and the individual teaching and learning needs of teachers and students at different stages.
Second, explore comprehensive, continuous dynamic oversight. Establish evaluations of product safety and educational suitability, conduct spot checks and routine inspections, operate dynamic black, grey and white lists, and explore scenario-based risk warnings and emergency mechanisms. These measures support assessment of safety and effectiveness.
Third, explore education-specific large AI models with greater accuracy, logical coherence and educational suitability. Strengthen ethical testing to reduce erroneous, biased or unsuitable content and build safe digital teaching environments that meet applicable requirements and ethical principles.
Fourth, improve practical guidance for using intelligent products on campus. Education authorities, practitioners and researchers should jointly develop guidance addressing student growth, teacher development and school transformation. Differentiated, enhanced and collaborative teaching can provide starting points for guiding reasonable use in accordance with applicable requirements.
Conduct human–AI collaborative educational practice in an orderly way.
First, use digital pedagogy as the theoretical foundation of human–AI collaborative teaching. Anchor practice in effective teaching and deep learning, supported by improved digital environments, digital resources and new technologies. The aim is effective learning and teaching, focusing on technological empowerment and learner-centred learning in trustworthy digital environments.
Second, deepen research into collaborative teaching mechanisms. Use multidisciplinary and interdisciplinary approaches to study teaching behaviour and complex teacher–machine–student interactions, enriching the theoretical framework. Systematically assess the combined effects of collaboration through an evidence-oriented approach and iteratively improve its structures and procedures.
Third, publish practical guidance and teaching standards for human–AI collaboration. Support teachers in using generative AI to analyse learning needs, find resources, generate test questions, design instruction, and provide assessment and feedback, while guiding students towards reasonable use of intelligent technologies.
Fourth, disseminate experience from collaborative teaching. Identify cases in which collaboration improves educational quality and draw on exemplary teachers' leadership. Organise training around teachers' actual needs and spread advanced, transferable, high-quality practice.
Develop standards for educational applications of generative AI.
To avoid unforeseen technological risks and protect learners whose ability to make judgements is still developing, educational uses of technology need ethical principles, service requirements and relevant standards.
First, technical standards. These are foundational, broadly applicable and forward-looking. They should address, among other matters, stable and scalable platform architecture, convenient and secure access, and clear arrangements for content ownership and usage permissions.
Second, quality standards. A quality framework should measure generative AI platforms' value, efficiency and benefits. It can cover regular testing of model accuracy and stability, reliable and diverse data sources, monitoring of generated content's accuracy, relevance and suitability, and evaluation of whether applications effectively support teaching objectives.
Third, service standards. These should cover technical services, confirmation of content rights, practical guidance and feedback. They should define the scope and guidance for educational applications of generative AI, clarify users and scenarios, and provide teachers and students with detailed operating instructions and good practice.
Build a framework for large-model security and ethics.
First, strengthen oversight and external supervision. Develop and implement policies to guide model design, development and use effectively. Protect data and privacy against breaches, unauthorised access and other violations. Implement data governance and trusted computing, establish rigorous encryption standards, and carry out routine security audits and risk assessments.
Second, strengthen the responsibility of model developers and their institutions. Technically, improve traceability and the explainability of decision-making, design transparent algorithmic frameworks, record decision processes and provide detailed explanations. In social applications, strengthen industry self-regulation, embed trustworthy principles, prevent, monitor and eliminate discrimination and bias, establish industry standards and self-regulatory commitments, and publish transparency reports regularly.
Third, strengthen responsible conduct and usage practices among model users. They should follow data and cybersecurity laws and rules, use technologies reasonably and responsibly, avoid unlawful or unethical conduct, respect privacy and prevent data misuse. Internal training and communication can strengthen legal awareness and ethical responsibility, alongside oversight and complaint mechanisms.
Sources and further reading
Source credited in the historical copy: the WeChat account of the Zhongguancun Internet Education Innovation Center.
The historical Chair copy credits the WeChat account of the Zhongguancun Internet Education Innovation Center. This edition retains the scope of that excerpt. The related full article by the same author appeared in Academic Frontiers, issue 14, 2024, and was posted by its publisher on 5 August 2024 at 15:38. The excerpt and the full article have different scopes and dates.
Original link: https://aiedchair.bnu.edu.cn/news-7989-en/
