Adoption of artificial intelligence in educational assessment: A mixed-methods study in Uruguayan secondary education and implications for in-service teacher development

Abstract

Objective. To analyse the adoption of artificial intelligence (AI) tools in the assessment of learning in Uruguayan secondary education and its implications for in-service teacher development. Methodology. A convergent mixed-methods study was conducted, integrating 187 responses to a self-administered questionnaire and 11 focus groups involving 52 secondary teachers. The study examined levels of familiarity and use of AI, pedagogical purposes, perceived barriers, and teachers’ views of the available training offer. Results. The findings show an early, uneven, and predominantly instrumental adoption. Use is concentrated on general-purpose tools, especially ChatGPT, and in planning and material production tasks. By contrast, the use of AI for assessment and personalised feedback remains limited due to pedagogical and ethical concerns related to the relational, contextual, and subjective nature of evaluation. Teaching experience did not produce significant differences in adoption, whereas specific training was associated with higher levels of familiarity and integration. In addition, the training offer was perceived as abundant but fragmented and weakly connected to classroom needs. Conclusions. The incorporation of AI into assessment in Uruguayan secondary education remains incipient and selective. Its development requires situated training that integrates technological, didactic, and ethical dimensions and supports pedagogically relevant and professionally responsible use.