In a classroom at Wenzhou Vocational and Technical University, students aren't working with hypothetical scenarios but tackling a real-world problem: detecting defects on a company's circuit breaker production line. From data annotation and model training to system deployment, student teams take full ownership of developing an AI-powered vision inspection system. The resulting solution operates several times faster than manual checks, and some students have earned direct job offers from the enterprise based on their project's success.
"The first time I saw actual defect images from a production line, I realized they looked nothing like the textbook examples. Lighting, angles, and reflections all introduce variables," said student Hou Renjie. He and his teammates spent two days refining their annotation strategy before the model produced usable results for the first time.
This kind of hands-on learning within real industrial settings is rapidly expanding across vocational colleges in Zhejiang Province. As the shelf life of technical skills shortens, Zhejiang is deepening its "AI + vocational education" reform, restructuring its talent development framework across curriculum design, teaching environments, and educational objectives.
Evolving Programs: From Following Trends to Setting Them
For years, vocational education program updates lagged behind industrial shifts, leaving graduates ill-prepared for evolving job demands. Zhejiang is now systematically addressing this timing gap. According to the Zhejiang Provincial Department of Human Resources and Social Security, the province has added over 460 new higher vocational program offerings in fields like intelligent robotics and microelectronics over the past five years, while phasing out more than 250 outdated programs in areas such as business administration. In 2025, Zhejiang is focusing on its "415X" advanced manufacturing clusters, establishing 100 high-level provincial specialty groups in higher vocational institutions and cultivating 180 quality specialty groups in secondary vocational schools.
The Machine Vision and Applications course at Wenzhou Vocational and Technical University exemplifies this forward-thinking approach. The curriculum breaks down real corporate case studies into 9 projects and 41 tasks, partnering with local enterprises like Chint and Tianzheng Electric. "Deep industry-education integration lets students accumulate real project experience while still in school, dramatically shortening their adaptation period after employment," said course instructor Zou Pengda.
Also at Wenzhou Vocational and Technical University, the Product Integrated Innovation Design course has been advancing AI toolchain teaching reforms since 2023. Structured around wearable product design, the course progresses in three stages: first using conversational AI to grasp solution generation techniques, then embedding AI throughout the entire process from requirements analysis and solution design to 3D modeling, and finally taking on real corporate projects. Student Wu Wencong handled a local smart glasses design project during the course, using AI to rapidly generate over 60 design options and completing the full workflow in four weeks. Several features from his proposal, including a modular battery compartment, are now being evaluated for inclusion in the company's new product pipeline.
"AI can produce a draft sketch in three seconds, but the real challenge lies in judgment—which requires refining considerations of requirements, manufacturing processes, and costs under teacher guidance," Wu said.
This proactive planning also extends to micro-majors. Under the Ministry of Education's "Double Thousand" plan for enhancing graduate employability, 708 micro-majors at Zhejiang institutions have been selected, bringing the cumulative total to 881 over the past two years. Micro-majors feature "small credit loads, high focus, streamlined courses, interdisciplinary scope, and flexibility," targeting specific and cutting-edge skill gaps. Students can select them based on personal interests and career plans, with completion recorded in the CHSI academic record system as part of their educational profile.
By moving from reactive adjustment to proactive design, the lag between educational offerings and industry needs is being minimized.
Transforming Learning Environments: From Master-Apprentice to Teacher-Student-Machine Collaboration
At Jinhua Vocational and Technical University, student Meng Shuo from the robotics program found his training experience different from what he expected. In the robot system integration lab, he and his classmates first familiarize themselves with the complete workflow using online virtual simulation, and only after passing assessments do they move to the physical lab to operate real equipment.
"Practicing in the virtual environment means no fear of making mistakes. When you then work with real machinery, you feel much more confident," Meng explained.
This "virtual trial and error, physical validation" model is quietly reshaping the traditional master-apprentice teaching format. At the university's medical school, students use VR headsets to complete full procedures from virtual blood draws to microscopy examination. The system captures details in real time and generates reports, allowing instructors to dynamically adjust case difficulty. In the pharmaceutical engineering program, students can complete hazardous chemical spill emergency drills in a virtual workshop within five minutes, solving the longstanding challenges of "high investment, high loss, high risk, difficult implementation, difficult observation, and difficult reproduction" in conventional teaching.
In the new energy vehicle program group, Jinhua Vocational and Technical University collaborated with Leapmotor to develop an "AI process mentor" that identifies operational steps in real time and corrects errors, reducing student mistake rates by nearly 60 percent. The university has also built a smart management platform centered on "digital work cards," converting every hands-on operation into a "skill profile." The average cycle time for achieving skill benchmarks has been shortened by nearly one-third.
Zhejiang Tourism College has independently developed the "Study China" immersive remote training platform, integrating resources from over 300 research and study bases nationwide. Using 3D mapping, VR simulation, and AI annotation technology, the platform creates a digital teaching ecosystem that is "visible, learnable, practical, and assessable."
From manual oversight to data-driven guidance, from purely physical training to integrated virtual-physical environments, the digital transformation of training scenarios is bringing new precision to skill development.
Shifting Educational Goals: From Operating Machines to Maximizing AI's Value
When AI can generate more optimal parameter combinations than seasoned technicians, what remains the human advantage? Zhejiang's vocational institutions answer: shifting from "memorizing experience" to "evaluating whether AI suggestions are correct," and from "operating machines" to "maximizing AI's value."
According to Huang He, Dean of Academic Affairs at Wenzhou Vocational and Technical University, the core task for students is not memorizing specific detection parameters but understanding how data quality determines model performance, how to adjust network structures to reduce false detection rates, and how to adapt laboratory solutions to real production lines. "These are precisely the judgments AI cannot make independently," he noted.
This shift in educational objectives is underpinned by systematic faculty development. Jinhua Vocational and Technical University has established a "three-tier progression" AI enablement framework, supporting more than 20 faculty teams annually in intelligent training research, with 34 typical AI teaching scenarios now in place. Xu Xiaofei, a faculty member in the School of Information Engineering, observed that in traditional digital circuit experiments, students spent excessive time on wiring and troubleshooting poor contacts—struggling with physical faults rather than digital logic. "Our remote lab platform allows students to perform real debugging based on programmable chips and expand as needed. This platform has now been deployed at overseas institutions including in Rwanda, allowing quality practical resources to cross borders."
From the 120 provincial first-class AI courses in vocational education approved in 2025 to the provincial vocational education high-quality development conference in March 2026 that confirmed the "AI + vocational education" direction, Zhejiang is building a comprehensive AI-enabled vocational training system starting from curriculum and classroom practice.
The end goal of this entire chain remains constant: not to cultivate people who "won't be replaced by AI," but to develop people who "can maximize AI's value." In the intelligent age, that may be the most durable moat for skilled talent.
Comments