Patterns of generative AI Use in Chinese university Spanish education: A two-sample descriptive study of student and teacher perspectives

Generative AI (GenAI) has proliferated in language education, yet empirical research on its real-world use remains scarce, especially beyond mainstream EFL settings. This parallel cross-sectional survey, partnered with an online research team, recruited 404 college Spanish as a foreign language (SFL) students and 406 SFL teachers across 17–18 Chinese provinces. Two custom questionnaires, built on cognitive offloading theory and critical AI application studies, measured GenAI usage, cognitive attitudes, participation tendencies, and teaching practices. Results show prevalent multi-GenAI adoption among students (only 14.1% rarely use such tools), with ChatGPT, DeepSeek, Kimi, Gemini, and Doubao mainly used for grammar, translation, and vocabulary learning. Teachers widely integrate GenAI into instruction but lack formal training: 44.6% learned it independently, and 9.9% had zero formal training. Provincial data shows a negative correlation between student self-reported critical engagement and teacher-perceived student critical engagement (r = –.53). Confirmatory and exploratory analyses found four theoretical subconstructs (cognitive offloading, critical engagement, beliefs, learning outcomes) converged into one dimension, reflecting severe acquiescence and common method bias in Chinese online survey data. This study establishes baseline GenAI adoption data for non-EFL Asian higher education and offers methodological warnings for relevant scholarship.