In a Wenzhou Polytechnic University course on machine vision and its applications, students aren't working with virtual simulations. They're tackling genuine defect detection challenges from an actual circuit breaker production line. Working in teams, they take on the full development of an AI visual inspection system—from data labeling through model training to system deployment—achieving detection speeds several times faster than manual methods and earning some of them direct job offers from the participating companies.

"The first time you see real defect images from a production line, they're completely different from the typical examples in textbooks. Lighting, angles, reflections—everything is a variable," said student Hou Renjie. He and his teammates spent two days refining their labeling strategy before the model produced its first usable results. This hands-on training approach, honed in real industrial settings, is rapidly spreading across multiple vocational colleges in Zhejiang.

As technical skills face ever-shortening "shelf lives," Zhejiang is deepening its "AI + vocational education" reforms. The province is comprehensively rebuilding its skilled talent development framework, spanning program offerings, teaching environments, and educational objectives.

Program Innovation: From Following to Leading

For years, vocational education program updates lagged behind the pace of industrial transformation, putting schools in an awkward position. Zhejiang is now challenging this pattern. According to the Zhejiang Provincial Department of Human Resources and Social Security, the province has added more than 460 new higher vocational program offerings over the past five years in fields like intelligent robotics and microelectronics technology, while phasing out over 250 programs in areas such as business administration. By 2025, Zhejiang had established 100 provincial-level higher vocational program clusters aligned with its "415X" advanced manufacturing initiative, plus 180 high-quality secondary vocational program clusters.

The machine vision course at Wenzhou Polytechnic University exemplifies this forward-thinking approach. The curriculum transforms real corporate cases into nine projects and 41 tasks, partnering with local companies including Chint and Tianzheng Electric. "Deep industry-education integration lets students build real project experience while still in school, dramatically shortening their on-the-job ramp-up time," said course instructor Zou Pengda.

At the same institution, the "Integrated Product Innovation Design" course has been implementing AI tool-chain teaching reform since 2023. Using wearable product design as its central theme, it progresses through three phases: first, using conversational AI to master concept generation; then embedding AI across requirements analysis, concept development, and 3D modeling; and finally, taking on actual corporate projects. Student Wu Wencong worked on a local smart eyewear design project, generating more than 60 concepts with AI assistance and completing the full design cycle in four weeks. Some of his structural innovations, including a modular battery compartment, are now being considered for the company's new product pipeline.

"AI produces a sketch in three seconds. The real challenge is judgment—developing the ability to weigh requirements, manufacturing processes, and costs through repeated refinement under instructor guidance," Wu explained. This forward planning also extends to micro-majors. Under the Ministry of Education's "Double Thousand" employment initiative, 708 micro-majors at Zhejiang universities have been selected, with cumulative approvals reaching 881 over two years. Characterized by small credit loads, focused themes, streamlined courses, interdisciplinary content, and flexibility, these micro-majors target specific cutting-edge skill gaps. Students can pursue them based on their interests and career plans, with completion recorded in the national academic information database.

By shifting from reactive adjustment to proactive planning, Zhejiang is closing the timing gap between curriculum and industry.

Teaching Environments: From Apprenticeship to Human-Machine Collaboration

At Jinhua Polytechnic University, robotics student Meng Shuo found his training experience different from what he expected. He and his classmates first master the complete workflow using an online virtual simulation system, and only after passing assessments do they move to the physical lab to operate real equipment. "Practicing in a virtual environment lets you make mistakes without consequences. When you finally work with the real equipment, you feel much more confident," he said.

This "virtual trial, physical verification" model is quietly reshaping traditional master-apprentice teaching methods. Medical students at Jinhua Polytechnic don VR headsets to complete full procedures from virtual blood collection to microscopy, with the system capturing details in real time and generating reports that let instructors dynamically adjust case complexity. Pharmaceutical engineering students can complete hazardous chemical spill emergency drills in a virtual plant in five minutes, overcoming the classic challenges of high-cost, high-risk, hard-to-replicate training scenarios. In the new energy vehicle program, the university partnered with Leapmotor to develop an "AI process mentor" that identifies operational steps in real time and corrects errors, cutting student error rates by nearly 60 percent.

The university has also built a smart management platform centered on "digital work cards," converting each training session into quantitative "skill profiles." This has reduced average skill mastery timelines by about one-third. Meanwhile, Tourism College of Zhejiang developed its own "Research China" immersive remote training platform, integrating resources from over 300 research bases nationwide. Using 3D mapping, VR simulation, and AI annotation, it creates a fully digital teaching ecosystem where students can see, learn, practice, and be evaluated.

From manual supervision to data-driven guidance, from purely physical training to blended virtual-physical environments, the intelligent transformation of training spaces is bringing precise measurement to hands-on skill development.

Educational Goals: From Operating Machines to Creating Value with AI

When AI can generate better parameter combinations than seasoned workers' experience, where does human advantage lie? Zhejiang's vocational schools have a clear answer: shift from memorizing experience to evaluating whether AI suggestions make sense, and from operating machines to maximizing AI's value.

According to Huang He, Dean of Academic Affairs at Wenzhou Polytechnic University, students' core task isn't memorizing specific detection parameters. It's 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 said.

This educational shift is underpinned by systematic faculty development. Jinhua Polytechnic established a three-tier AI empowerment framework, supporting over 20 faculty teams annually in intelligent training research and building 34 AI teaching scenarios. Xu Xiaofei, an instructor in the School of Information Engineering, noted that traditional digital circuit labs consumed significant student time on wiring and troubleshooting loose connections—frustrated by physical faults rather than learning digital logic. "Our remote practice platform lets students perform real debugging on programmable chips and expand as needed. The platform has even been exported to overseas institutions in Rwanda and elsewhere, bringing quality practice resources across borders."

From launching 120 provincial-level first-class AI vocational courses in 2025 to the March 2026 provincial conference on vocational education high-quality development that affirmed the "AI + vocational education" direction, Zhejiang is building a comprehensive AI-empowered education system for the digital age, starting with curriculum and classrooms. The end goal remains consistent throughout: not producing people who cannot be replaced by AI, but producing people who can maximize what AI has to offer. In the intelligent era, this may be the most durable advantage skilled workers can build.