<![CDATA[Children’s voice is central to participatory early childhood education, yet little is known about how conversational artificial intelligence (AI) may mediate children’s expression and participation within STEAM learning. This qualitative descriptive study examined how ChatGPT Voice mediated children’s voice during AI-supported STEAM activities in an early childhood education setting in Central Java, Indonesia. Participants were 20 children aged 5–6 years and two classroom teachers. Data were generated from May to July 2026 through eight classroom observation sessions lasting approximately 45–60 minutes each across six AI-mediated STEAM activities, semi-structured interviews with the two teachers, and documentation, and were analyzed thematically. Five qualitative patterns were identified: AI opened an additional space for expression; functioned as a responsive interactional audience; accompanied task-level agency; provided shared prompts for collaborative voice; and exposed developmental and communicative barriers requiring teacher mediation. These findings represent context-bound qualitative patterns rather than evidence that ChatGPT Voice directly increased children’s participation. Conversational AI may therefore complement, rather than replace, teacher-led participatory pedagogy when interaction is developmentally appropriate, critically mediated, and embedded in hands-on and peer-collaborative activity.]]>
Copyrights © 2027
