Education Futures in the Making

In February 2025, the Estonian government launched the educational innovation programme AI Leap (TI-Hüpe), a state–industry partnership aimed at introducing generative AI into upper-secondary education. Initially conceived to provide all teachers and students in grades 10 and 11 with access to OpenAI ChatGPT Edu, the programme was soon developed with a large-scale teacher training component. The initiative was explicitly framed as a continuation of the 1990s Tiger Leap, a national programme that equipped schools with computers and internet access and that has become central in the narrative of Estonia’s transformation from a post-Soviet state into a digital forerunner. This study explores how the expert group behind the early stages of the AI Leap were formed and how their work contributed to legitimising particular visions of education futures. Theoretically, the paper draws on the concepts of education futures and sociotechnical vanguards to analyse how visions of what constitutes legitimate knowledge and expertise shape the conditions for educational transformation. Education futures are understood here as practices of anticipation through which actors seek to render the future governable in the present, while sociotechnical vanguards refer to collectives of experts that mobilise social and material resources to steer change in particular directions. The analysis is based on twelve semi-structured interviews with actors involved in AI Leap, including government officials, researchers, consultants, and school leaders, as well as contextual material from media and official documents. The findings indicate that the formation of the expert group responsible for implementing the AI Leap was contingent and network-based rather than the result of formal institutional procedures. The group was dominated by representatives with a background in psychology and neuroscience rather than, as in previous educational reforms in Estonia, by entrepreneurs from the private sector. This configuration shaped what counted as legitimate expertise within the programme, privileging scientific and evidence-oriented approaches over other forms of knowledge. The analysis further shows how this sociotechnical vanguard was embedded within a broader network of policy actors, technologies and institutions through which particular postdigital education futures were stabilised and promoted. AI was framed as an already present and disruptive condition that could neither be ignored nor blindly trusted, positioning governance and pedagogical steering as necessary responses. While generative AI was widely used by students, pedagogical responses remain unsettled, giving rise to what the paper describes as proto-habits, where established practices and anticipated futures coexist.