At the heart of many developments in educational technology (edtech) are products utilising artificial intelligence technologies. Welcomed by some as a solution to problems in schools (e.g., low student motivation, insufficient personalisation, socio-emotional challenges, burnt-out teachers), AI, in particular ‘generative AI’, is critiqued by others as a threat to the human interactions at the heart of education (e.g. increasing bias, blocking explorative learning, reducing critical thinking and thus enabling fascism). Biases are particularly strong when legitimation is based on notions of community and humanity, for instance, in arts and music education. In this binary discussion (solutions/threats), a ‘third position’ is crucial, which combines design and situated analysis. An exciting body of work is emerging in this area, but there remains a lack of research that combines social, cultural and technical sciences to co-develop novel AI systems and investigate their use in pedagogical settings. 

RECOMPOSE develops this third position by developing an AI-powered platform architecture specifically to foster students’ and teachers’ critical AI literacies. Here, ‘critical AI literacies’ refers to an increased understanding of AI technologies that (i) demystifies the mechanics, (ii) identifies societal (political, ethical, mental health, etc.) implications, and (iii) thus enables students and teachers to make informed decisions about how to use these technologies for which purposes. Put simply: If young people have fun with AI while learning how, e.g., bias is generated as music is generated, they develop a sense of how decisions are made, and can explore the implications of these decisions for societal cohesion, e.g., gender, race, class, and other intersecting dimensions of diversity. The key is for young people to transfer this knowledge to other areas, asking how high-stakes decisions (e.g., grading, policing, journalism) are shaped by AI.