Qingchun Bai

Large language models (LLMs) are reshaping intelligent tutoring systems (ITSs) by reducing authoring costs and enabling flexible interaction. However, effective tutoring requires more than fluent responses. It also depends on pedagogical control, domain grounding, learner modeling, interaction design, orchestration, and reliable evaluation. We examine these requirements through a component-centric framework that organizes LLM-based tutoring systems into four functional layers: pedagogical policy, domain and grounding, learner state, and interaction/interface. A cross-cutting orchestration layer coordinates these components, while agentic tutoring systems are treated as an emerging subset of the broader field. Drawing on representative literature on LLM-based and agentic tutoring systems, published mostly between late 2022 and mid-2026, we analyze how tutoring capabilities are implemented across system architecture, learner modeling, interaction, orchestration, and evaluation. The reviewed literature suggests that evaluation remains concentrated on response-level and model-based proxies, while direct evidence of learning gains remains limited. Learner-state modeling and interaction design also receive less systematic attention. We conclude by identifying challenges in answer leakage, pedagogy–solving trade-offs, learner-state calibration, interaction validity, and proxy validity.

The rapid advancement of artificial intelligence (AI) has increasingly permeated various sectors, with generative AI technologies, such as ChatGPT and ERNIE Bot, creating unprecedented opportunities and challenges for educational practices. Dialogic learning, an enduring method of instruction, is on the brink of transformation in forms and models with the integration of these cutting-edge technologies. This paper examines existing literature on the nature of dialogic learning and the types of dialogues while highlighting the advantages and characteristics of generative AI. It explores the value implications of human-machine dialogic learning, proposes a strategic framework for crafting effective prompts to engage with generative AI, and introduces a foundational model for human-machine dialogic interaction. Furthermore, the paper outlines strategies and methodologies to promote this model, with a focus on fostering problem awareness, identifying linguistic risks, enhancing digital literacy, and mitigating the risks of technology dependency.

Multimodal large language models (MLLMs) have attracted considerable attention for their impressive capabilities in understanding and generating visual-language content, particularly in tasks such as visual question answering (VQA). However, the rapid evolution of knowledge in real-world applications poses challenges for these models: offline training becomes increasingly costly, and exposure to non-stationary data streams often leads to catastrophic forgetting. In this paper, we propose CL-MoE+, a dual-momentum Mixture-of-Experts (MoE) framework based on MLLMs for continual VQA. Our method integrates continual learning into MLLMs to leverage the rich commonsense knowledge embedded in large language models.We introduce a Dual-Router MoE (RMoE) module that selects both global and local experts through task-level and instance-level routers, enabling robust and context-aware expert allocation. Furthermore, we design an adaptive Momentum MoE (MMoE) to update experts’ parameters based on the knowledge drift degree and their relevance to specific tasks, thereby facilitating knowledge integration without forgetting. Extensive experiments on a 10-task split of the VQA v2 benchmark demonstrate that CL-MoE+ achieves state-of-the-art performance, validating its effectiveness in both retaining historical knowledge and learning new information in the continual learning setting.

Powered by artificial intelligence (AI), AI Tutoring Assistant holds significant potential in simulating human tutors, instructional behaviors and functions. To explore the practical demands of AI Tutoring assistants in online learning scenarios, this study analyzed social platform user reviews and questionnaire data to evaluate their acceptance of online learning tools and influencing factors. The research identified three core learner needs: dynamic learning support, precise resource recommendations, and emotionally resonant interactive experiences. Regression analysis validated the Technology Acceptance Model (TAM), revealing that perceived usefulness and ease of use significantly influenced usage intention. This study also proposed the application of AI Tutoring Assistant, providing a new theoretical perspective for personalized learning support and empirical evidence for optimizing the user experience of artificial intelligence education products.