Exploration of AI-driven innovation and entrepreneurship training models in vocational education: A design-based framework and evaluation protocol

Artificial intelligence is reshaping vocational innovation and entrepreneurship education through learning analytics, virtual simulation, and generative AI support, yet its large-scale implementation is constrained by infrastructure costs, limited teacher preparedness, risks of AI dependency, and equity concerns. This study proposes and evaluates the Vocational AI Venture Studio (VAVS), a four-layer framework integrating adaptive learning pathways, simulation-based venture practice, generative AI venture co-pilots with critical-use scaffolds, and a governance layer emphasizing teacher capacity, privacy protection, infrastructure equity (bandwidth- and device-adaptive access), and educational equity. A propensity score–matched quasi-experimental study was conducted with 610 students across eight vocational colleges in China sampled to span high-, medium-, and low-infrastructure settings (differing in internet bandwidth and hardware availability). Results show that VAVS significantly improved skill mastery, entrepreneurial self-efficacy, innovation competency, and venture performance. Generative AI effects were substantially stronger when guided by critical-use scaffolds, while unscaffolded use led to signs of AI dependency. The intervention also enhanced teacher AI preparedness and reduced outcome disparities across demographic and infrastructure subgroups, indicating improved equity. Overall, the findings demonstrate that VAVS is an effective and modular model for AI-enhanced vocational entrepreneurship education, highlighting that sustainable learning gains and fairness depend not only on technology adoption but also on pedagogical scaffolding and governance infrastructure.