Designing ethical AI characters for children’s early learning experiences

Abstract This qualitative study investigated the ethical design of AI characters for children ages 3–8 by addressing this research question: what factors contribute to ethical AI character design? Data were collected from public documentation of 20 AI characters, including websites, demo videos, and user-generated social media videos ( N = 60). The study utilized Chen and Lin’s (Contemp Issues Early Child, 25:146–153, 2024) POWER principles (purposeful, optimal, wise, ethical, responsible) as a theoretical framework, and Tiwari’s (Assessing AI characters as fascilitators of children's learning experience. In: Proceedings of the MIT AI + Education Summit, 2025) AI Character Assessment (AIC-A) as an analysis framework to understand data across three factor considerations: child, AI character, and the interactions between them. Findings showed that the most educationally effective AI characters had clear learning goals, developmentally appropriate content, semi-structured interactions, opportunities for both individual and joint engagement, and transparency about being AI. An important implication of the study is operationalizing the findings as a planning tool for learning designers to develop ethical AI characters for early learning.