With the rapid development of generative artificial intelligence (GenAI), agents based on foundation models (FMs) have gradually acquired capabilities, such as multimodal perception, retrieval-augmented generation, reasoning and planning, interaction, and evolution. This study proposed a fundamental concept and framework for FM-based pedagogical agents. Centering on the FM as the technological core, the framework primarily consisted of several functional modules, including educational task setting, educational task planning, realization and expansion of educational capabilities, memory and reflection on educational content, and interactive collaboration and dynamic evolution. These modules support interaction with diverse entities and facilitate dynamic evolution, encompassing human-computer interaction, multi-agent interaction, and environment interaction. Building upon the proposed framework, this study adopted project-based learning tasks as an application scenario. It elaborated on the roles of pedagogical agents functioning as assistant agents and peer agents across various phases, such as personalized driving question formulation, collaborative project plan design, collaborative project work completion, and multi-role evaluation of project works, along with the associated supporting technologies. Furthermore, this study discussed the future development and prospects of pedagogical agents.

Generative artificial intelligence (GenAI) tools have brought new impetus to the digital transformation in education, empowering talent cultivation. This study conducted a questionnaire survey of 1,781 college students to explore the impact of using GenAI tools on their critical thinking and autonomous learning ability. The study identified four key findings: (1) Students who use GenAI tools have significantly higher levels of critical thinking and autonomous learning ability than those who do not use such tools. (2)Students who consistently use GenAI tools exhibit superior performance in cirticalthinking and autonomous learning ability than those who do not consistently usesuch tools. (3) Compared to undergraduates and students from regular universities, postgraduates and students enrolled in universities under the Double First-Class Initiativedemonstrate stronger critical thinking and autonomous learning ability through the use of GenAI tools. (4) Students predominantly use GenAI tools to pose closed questions.Moreover, in-depth analysis is required to determine the boundaries of the personalizedassistance that such tools can provide to students.

Generative artificial intelligence (GenAI) has demonstrated significant effectiveness in supporting students’ metacognitive regulation. However, current research has yet to examine the patterns of interaction behaviors between students and GenAI in learning contexts. This study developed a GenAI-supported metacognitive regulation learning system and implemented it in teaching practice. Through cluster analysis to model the interactions between students and GenAI across multiple learning tasks, this study identified four distinct interaction patterns, namely, active, balanced, disengaged, and detached patterns. Students exhibiting the active pattern and the balanced pattern demonstrated higher self-efficacy and better academic achievement, as well as richer and more complete behavioral sequences of metacognitive regulation, compared to those with the disengaged pattern and the detached pattern. These findings provide valuable insights for the design and implementation of the GenAI-supported learning environment.

In the era of digital transformation, generative artificial intelligence (GenAI) has become an important tool for enhancing research efficiency. However, its specific impact on research output remains underexplored. Based on the data from the 2024 national doctoral graduate survey, this study investigated the impact of GenAI on doctoral students’ research output using propensity score matching and inverse probability weighted regression adjustment methods. The findings showed that GenAI use increased doctoral students’ total research output by 6.5%, international publications by 6.9%, and top-tier journal publications by 16.5%. However, the technological dividends were not equally shared, with factors such as gender and age constituting substantial barriers to using GenAI. Moreover, heterogeneity analysis revealed that the benefits of GenAI use varied significantly across different disciplines and academic environments. For doctoral students with insufficient mentor guidance, although GenAI contributed to an increase in the total number of papers and international journal publications, it failed to yield significant benefits for top-tier journal publications. Accordingly, this study recommends the systematic integration of GenAI into doctoral training systems, the development of intelligent resource-sharing platforms, and the strengthening of ethical norms and fairness safeguards for GenAI use. These measures aim to promote the rational application of the technology and the equitable sharing of digital dividends, thereby fostering the high quality and sustainable development of doctoral education.

The advent of generative artificial intelligence (GenAI), represented by ChatGPT, has challenged traditional learning mode. How students learn through GenAI tends to be an urgent problem to be explored in the current education and teaching reform in China. This study analyzed the in-class dialogues between undergraduates and GenAI by coding discourse types, questioning levels, questioning strategies, and self-reports of students to explore the learning model of human-AI collaboration. The findings showed that in student-GenAI dialogues dominated by students, there were more single-round dialogue units and fewer continuous discussions around a topic. The main types of students’ discourse were initial questioning, extended questioning, and rephrasing questioning, while evaluation and continuing prompt discourses were less prevalent. Moreover, students’ cognitive level of questioning was low, focusing on remembering-level questioning and understanding-level questioning. The use of questioning strategies was unfamiliar, and students seldom used role questioning, material questioning, and solution questioning. In addition, it was discovered that different task stages and different utilization experiences both led to different dialogues between students and GenAI. With the development of task solving, there were more frequent and sustained dialogues, along with a higher cognitive level and more proficient use of questioning strategies. Students with more experience of using GenAI generated more dialogues with a high cognitive level. In human-AI dialogues representing different characteristics, though there were different opinions on using GenAI in classroom teaching, most of the students held a positive attitude. In the students’ perception, GenAI has the advantage in generating responses, furnishing valuable information, handling various types of tasks, and fostering the development of student abilities, thereby assisting students in learning. Meanwhile, GenAI encounters challenges related to technical limitations, raising concerns about student development, learning assessment, and the overall educational ecosystem. On this basis, this study provides effective suggestions for further using GenAI in classroom teaching from three aspects: providing question guidance, enriching question scenarios, and strengthening reflection on GenAI responses.

Enpowering education with artificial intelligence (AI) has established new pathways for high-quality development in education, bringing new opportunities while fostering innovative educational concepts, models, and methodologies. This study introduces an integrated experimental platform, namely, the teacher-student-AI collaborative classroom, for implementing these educational innovations. Grounded in macro disciplinary, grand-scenario, and big-idea-based education, this classroom is designed to organize instruction for each knowledge unit with a three-phase method: introductory sessions, seminar sessions, and summary sessions. Introductory sessions and summary sessions are teacher-guided, while seminar sessions emphasize the self-directed inquiry based learning of students with AI assistance. During introductory sessions, teachers adopt a problem-analyzing approach to establish learning tasks and contexts. Seminar sessions enable students to engage in collaborative discussions and self-directed learning centered on problem analysis, facilitated by AI Large Language Models (AI-LLMs). Summary sessions, in addition to showcasing the outcomes of student representatives, entail teachers synthesizing knowledge through a problem-solving approach, highlighting the significance of knowledge units and future development. The teacher-student-AI collaborative classroom synergizes the strengths of teachers and AI technology, fostering active student participation, personalized learning, and social-emotional competence development. It serves as a practical platform for innovative, problem-centered education. Applied to the Probability Theory and Mathematical Statistics course in universities, this platform marks improvements in student motivation, initiative, and comprehensive competencies, validating its feasibility and broad applicability.

Higher education, serving as the primary source of talent, the principal platform for research, and the chief engine of innovation, is a key force in accelerating the development of new quality productive forces. As a new strategy for empowering the new quality development of higher education through digital and intelligent technologies,convergence innovation combines theoretical thinking, research paradigms, and practical innovations. This study argues that the convergence innovation research and convergence innovation education should be recognized as a new frontier of interdisciplinary research and a new paradigm for educational innovation enabled by digital and intelligent technologies. In addition, this study explores the connotative elements, logical mechanisms, and implementation pathways of convergence innovation research and convergence innovation education. Furthermore, it proposes strategies and recommendations to advance convergence innovation in Chinese higher education. These include cultivating convergence innovation talent through convergence innovation research and convergence innovation education, fostering a science of team science (SciTS)-based convergence innovation culture, and improving the governance framework for convergence innovation in higher education.