A chatbot-building environment helps children investigate unreliable AI output through confidence cues, fact-checking, and comparisons between models. A study with 48 learners aged 10–14 found gains in AI knowledge and hallucination awareness, alongside challenges in combining these supports effectively.
