GPTs and the Choice Architecture of Pedagogies in Vocational Education

Generative pre-trained transformers (GPTs) have rapidly entered educational contexts, raising questions about their impact on pedagogy, workload, and professional practice. While their potential to automate resource creation, planning, and administrative tasks is widely discussed, little empirical evidence exists regarding their use in vocational education (VE). This study explores how VE educators in England are currently engaging with AI tools and the implications for workload and teaching practice. Data were collected through a survey of 60 vocational teachers from diverse subject areas, combining quantitative measures of frequency, perceived usefulness, and delegated tasks with open qualitative reflections. Descriptive statistics, cross-tabulations, and thematic analyses were used to interpret responses about the application and allocation of work given by teachers to GPTs. Findings indicate cautious but positive adoption, with most educators using AI tools infrequently (0–10 times per month), yet rating them highly useful (average 4/5) for supporting workload. Resource and assessment creation dominated reported uses, while administrative applications were less common. The choice architecture framing indicates that some GPTs guide teachers to certain resources over others and the potential implications of this are discussed. Qualitative insights highlighted concerns around quality, overreliance, and the risk of diminishing professional agency. The study concludes that GPTs offer meaningful workload support but require careful integration, critical evaluation, and professional development to ensure they enhance rather than constrain VE pedagogy.