Who Pilots the Copilots?

Generative AI (GenAI) is praised as a transformative force for education, with the potential to significantly alter teaching and learning. Despite its promise, debates persist regarding GenAI impacts, with critical voices highlighting the necessity for thorough ethical scrutiny. While traditional ethical evaluations of GenAI tend to focus on the opacity of AI decision-making, we argue that the true challenge for ethical evaluation extends beyond the models themselves, and to the socio-technical networks shaping GenAI development and training. To address this limitation, we present an evaluation method, called Ethical Network Evaluation for AI (ENEA), which combines Latour’s Actor-Network Theory—used to map network dynamics by tracing actors’ interests and values—with Brusseau’s AI Human Impact framework, which identifies ethical indicators for evaluating AI systems. By applying ENEA to GenAI “copilots” in education, we show how making Actor-Networks visible lets us unveil a great variety of dilemmas, guiding ethical auditing and stakeholder discussions.