1. The Construction of Innovative Theories and Systematic Practical Exploration of New Happy Education

• Participants: 

  Junjie Shang, Yu Liu, Peng Zhang, Junyi Zhou, Ruonan Hu, Zhuo Li, Lan Hou, Qiuping Hu, Zhu Shi, Jialing Zeng, Wenli Yang, Leisi Pei, Lu Zhang, Qi Xia, Yuanyuan Zhang, Lixiang Gao, Wendan Huang, Shufang Tan, Yueying Zhao, Yulong Huo

• Keywords: 

  New Happy Education, Game-based Learning, Learning Sciences, Construction of Theories, Systematic Practice

• Executive Summary: 

  New Happy Education aims to integrate learning sciences, game-based learning, artificial intelligence, and other emerging technologies and learning approaches to make learning more scientific, more enjoyable, and more effective. As educational reform enters a deep-water zone, it is essential to recognize that learning lies at the heart of education. Transforming learning is therefore key to driving deeper changes in education. Only through such transformation can we address persistent issues, such as low student motivation and increasing psychological challenges among learners.

• Project Description

  This project was inspired by the applicant’s long-term exploration of how to make learning more engaging and meaningful. Since 2004, the applicant has researched game-based learning, aiming to improve students’ learning experiences and motivation under exam-oriented education. However, it became evident that relying solely on games risked turning “happy education” into entertainment, overlooking learning effectiveness. In response, the applicant integrated learning sciences with game-based learning, later leading a national project titled Game-Based Learning from the Perspective of Learning Sciences. With the rapid rise of artificial intelligence, AI was gradually incorporated into the framework.  This evolution led to the theory of New Happy Education—an approach grounded in learning sciences, characterized by game-based learning, and supported by AI. It also draws on brain science, virtual reality, and other emerging technologies to promote learning that is scientific, joyful, and effective, helping every child grow up in a healthy, balanced way. The project seeks to break the dilemma of “sacrificing joy for achievement” or“choosing fun over growth.” Through thoughtful learning design, students can experience real happiness in learning—not just fun in playing—and achieve holistic development through genuine engagement.

2. Development and Validation of the PA-SDA Scale for AI-Integrated Self-Directed Language Learning

• Participants: 

  Belle Li, Zhuo Zhang, Victoria Lowell, Chaoran Wang, Curtis J. Bonk

• Keywords: 

  Artificial intelligence; Self-directed learning; Scale development; Personal attributes

• Executive Summary: 

  The Personal Attributes for Self-Directed AI Learning (PA-SDA) Scale is the first validated quantitative instrument measuring key personal attributes of language learners in AI-integrated self-directed learning contexts. Grounded in the AI-Integrated SDL Framework (Li et al., 2024b), the 44-item scale was developed and validated with 699 global language learners using generative AI tools such as ChatGPT. It offers educators, researchers, and policymakers a robust diagnostic tool to assess learners’ readiness for AI-enhanced education, enabling targeted interventions, equitable access, and strategic pedagogical design. The PA-SDA Scale bridges theory and practice, providing a scalable and adaptable solution for AI-era education worldwide.

• Project Description:

  This project was inspired by a pressing gap in education: the absence of validated tools to assess learners’ readiness for AI-integrated self-directed learning (SDL). While generative AI tools are rapidly transforming language education, success depends not only on technology but on learners’ ability to navigate, adapt, and critically engage with AI systems. Existing SDL readiness scales were developed before the advent of advanced AI, overlooking new competencies such as AI-specific strategy use, resource management, and critical evaluation of AI-generated content. Building on Li et al.’s (2024b) AI-Integrated SDL Framework, we operationalized five personal attributes (Attitude, Strategy Use, Motivation, Self-Efficacy, and Resource Use) into measurable items. Following rigorous scale development protocols, we conducted exploratory and confirmatory factor analyses (EFA, CFA) to validate the construct structure. The final 44-item PA-SDA Scale reflects nine subconstructs, offering fine-grained insight into learner readiness in AI-mediated contexts. The PA-SDA addresses the UN Summit of the Future 2024 agenda by advancing inclusive, quality digital education. It empowers educators to design interventions that bridge the AI readiness gap, supports equitable access to AI-enhanced learning, and informs policy on AI integration in curricula. Its adaptability allows application across languages, age groups, and educational levels. Ultimately, this project contributes a theoretically grounded, empirically validated innovation that will help education systems worldwide prepare learners for the future of AI-enabled learning.

3. RAG-Tutor

• Participants: 

  Daniel Burgos, Enrique Frías, José Carlos San José

• Keywords: 

  Dialogic learning, On-demand learning, Generative AI, RAG
  

• Executive Summary: 

  Provide a summary of your project, why it is important and how it will address a specific issue/problem. This project generates a tool that enhances DIALOGIC LEARNING, what is the process through which a student learns based on the conversation that he or she makes with data sources (the corpus). As a complement to memorization, or procedural learning (where a series of steps that must be followed are listed), dialogic learning focuses on inquiry, exploration, and the curiosity of the student. It is, therefore, a tool that enhances the student's capacity for constructive thinking and critical analysis, as a complement to other transversal competences. Dialogic learning also supports functional diversity, with real-time adaptation to learning styles, ways of expression, along with competence and cognitive levels.

• Project Description:

  This project designs a privacy-preserving multilingual assistant that acts as a tutor for students. The system is implemented using a RAG architecture (Retrieval Augmented Generation) that solves students’ doubts and questions using the educational content provided by the professor for the course, in a private environment, both for sources and for user interaction. The main objective is to create a tool, RAG-Tutor, that adapts content to student needs and as a result enhances and facilitates the learning process by personalizing the answers and providing mechanisms for human-machine collaboration. This project is built on recent advances in LLMs (Large Language Models), Retrieval-Augmented Generation (RAG), and privacy-preserving approaches, to build a tutor architecture that provides relevant answers within the context of the documentation of a course. In general, the goal is to design tutoring systems that can facilitate the learning process for online students.

  Key objectives:

  1. Designing a multilingual privacy-preserving RAG-based architecture that allows students to ask questions and clarify doubts without compromising sensitive information.

  2. Ensuring scalability and adaptability of RAG-Tutor to different courses

  The main contributions to Dialogic and On-demand learning are:

  (1) Contextualized and Adaptive Content Delivery: RAG-Tutor dynamically contextualizes learning content for  specific needs. By applying Retrieval-Augmented Generation, RAG-Tutor can present just-in-time information  adapted to each student's progress enhancing on-demand learning.

  (2) Student-Centered Approach: RAG-Tutor is designed with a student-centered approach, putting learners' goals, interests, and challenges in the core of the system. It identifies knowledge gaps while presenting content in a personalized and accessible way.

  (3) Self-Directed Exploration and Lifelong Learning: RAG-Tutor allows students to engage with educational  material on their own and as per their needs and interests. It enables self-directed learning with real-time and  context-aware responses.