The ethics of educational artificial intelligence (AIED) has become a prominent topic of debate in recent years. However, many prospective AIED tools are still under development and may change before they are widely deployed in schools. Researchers analyzing the prospective ethical impacts of AIED cannot rely on actual use cases and empirical evidence concerning the experiences of teachers and students affected by AI. To address such limitations, this paper uses futures studies methods to develop four anticipatory scenarios that can enable ethical deliberation on the future of AIED. By employing a 2 × 2 grid, I forecast scenarios in which AIED is used to deliver content, used to manage students, used for both these purposes, or not used at all. I argue that futures studies methods can enrich the approaches used to investigate the possible futures of education. Just like educational fiction, scenario planning helps construct rich narratives, but produces distinct and original results while being more explicit about the factors and assumptions that influence them. Moreover, while previous anticipatory ethical work highlighted only individual issues or principles, my scenarios are ethically ambiguous and multifaceted – they present value conflicts and moral dilemmas that may be at play in the forecasted futures.