EmbeddingGemma 2: an open, lightweight multimodal embedding model
deepmind.google·6d ago
AI Engineering
Hacker News·Jul 22·5 min read
Anthropomorphism in Children's Interactions with LLM Chatbots
TL;DR
A systematic review of 35 studies reveals that children often anthropomorphize LLM chatbots, attributing human traits to them, which influences their interactions and outcomes. Key drivers include human-like design and supportive companionship, leading to both positive and negative social responses.
✦ Why It Matters
Design LLM chatbots with human-like traits to enhance children's engagement while being mindful of potential social confusion.
Key Takeaways
- Incorporate human-like traits in chatbot design for better engagement.
- Monitor children's social responses to chatbot interactions.
- Explore adaptive scaffolding techniques in chatbot interactions.
- Consider the implications of children's narratives around chatbots.
Full Summary
Research has shown that children frequently interact with large language model (LLM) chatbots by attributing human characteristics to them, a phenomenon known as anthropomorphism. This systematic review analyzed 35 empirical studies published between 2022 and 2025 to identify the factors driving this behavior and the resulting outcomes.
Key drivers include the construction of human-like personas, adaptive scaffolding that supports learning, and the chatbots' design as supportive companions. The review found five main outcomes of these interactions: children display complex social and moral responses, develop dual consciousness regarding the chatbots, form varying social ties, explore social boundaries, and create human narratives around conversation failures.
These findings highlight both the potential benefits and risks of such interactions, providing insights for designing LLM chatbots that prioritize children's developmental needs and well-being.