Mowa eksplanacyjna w rozmowach dzieci z chatbotami GPT: szanse, ryzyka i implikacje pedagogiczne

Abstract

The article addresses the problem of default “closure” in child–AI conversations produced by large language models (LLMs), which often provide coherence that is fluent yet cognitively shallow. Using opportunistic data (voice-mode transcripts of a six-year-old interacting with a GPT chatbot), it analyzes the conditions that support explanatory talk and propose a short protocol of pre-dispositions/meta-prompts together with a polyphonic response mode (multiple explanatory variants).These solutions increase the number of accurate explanations and meaningful follow-up questions while reducing the adult time required to scaffold dialogue. It discusses the limitations and risks of relying on explanations generated by chatbots. The text offers tools for consciously designing child-AI dialogue that can be replicated in both research and classroom practice.

Ähnliche Arbeiten