Article Details

Abstract

With the rapid penetration of Generative AI (GenAI) in programming education, the traditional “code-text” single-modal tutoring mode is difficult to meet the cognitive needs of high visualization scenarios such as web development. Aiming at the learning pain points of students in technician colleges, this paper constructs a Multi-modal Visual Programming Generative AI Platform (MVG) and implements three key functions based on large models: real-time code-visual synchronization, automatic mind map generation, and combined deployment of domestic large models and localization. This study adopts the spiral iterative method of “demand diagnosis-rapid prototyping-expert review-classroom verification”. Results show that the System Usability Scale (SUS) score of the platform increases from 62.7 in the prototype stage to 79.8 (students)/76.3 (experts), both of which are significantly higher than the industry benchmark value of 68, and approach the “excellent” threshold of 80; the average score of expert heuristic evaluation increases from 3.4 to 4.3 (out of 5). The study shows that the design combining multi-modal visualization and semantic mapping significantly improves beginners' overall understanding of web structure and coding logic, and also provides a replicable GenAI application paradigm for technician education scenarios.

Keywords

multi-modal; GenAI; technician education; visual programming; SUS

References

1. [1] Opinions of the Ministry of Education and Nine Other Departments on Accelerating the Digitization of Education [J]. Bulletin of the Ministry of Education of the People's Republic of China, 2025 (4): 16-20
2. [2] Ling Jie, Zhang Lili, Chen Hongguang, etc Application of Generative Artificial Intelligence in Computer Experiment Teaching [J]. Computer Education, 2025 (5): 28-32
3. [3] LI H，XU T L，ZHANG C L，et al. Bringing Generative AI to Adaptive Learning in Education [J/OL].arXiv:2402.14601 [cs.CY].[2025-07-03].http://arxiv.org/abs/2402.14601.
4. [4] LEE G，SHI L，LATIF E，et al. Multimodality of AI for Education: Toward Artificial General Intelligence [J].IEEE Transactions on Learning Technologies，2025，18：666-683.
5. [5] Mao Yu, Qin Zizhen, Fu Zhaomin Blended teaching practice of programming based on the dual encoding theory of human brain knowledge [J]. Journal of Higher Education, 2024, 10 (27): 17-20
6. [6] Cao Haofeng Construction of Adult Online Learning Support Service System in Fragmented Learning Environment [J]. Continuing Education Research, 2025 (9): 23-29
7. [7] SHIHAB M I H，HUNDHAUSEN C，TARIQ A，et al. The Effects of GitHub Copilot on Computing Students' Programming Effectiveness, Efficiency, and Processes in Brownfield Coding Tasks [C]//ICER' 25: Proceedings of the 2025 ACM Conference on International Computing Education Research V.1.Charlottesville：ACM，2025：407-420.
8. [8] PANKIEWICZ M，BAKER R S. Navigating Compiler Errors with AI Assistance - A Study of GPT Hints in an Introductory Programming Course [C]//ITiCSE 2024: Proceedings of the 2024 on Innovation and Technology in Computer Science Education V.1.Milan：ACM，2024：94-100.
9. [9] LEE J，CHEN S-S，LIANG P P. Interactive Sketchpad: A Multimodal Tutoring System for Collaborative, Visual Problem-Solving [J/OL].arXiv:2503.16434 [cs.HC].[2025-07-20].https://doi.org/10.48550/arXiv.2503.16434.
10. [10] Zhang Mingxin A study on the sustained willingness of college students to use social reading apps: the mediating effect of immersive experience [J]. Journal of University Libraries, 2021, 39 (1): 100-109

About the authors

Liang Weijian (1994-), male, Han nationality, born in Guangning, Guangdong Province, is an intermediate engineer and a master's candidate. His research interests include artificial intelligence, Internet development and vocational education.