eaSEL: Promoting Social-Emotional Learning and Parent-Child Interaction through AI-Mediated Content Consumption

Research Background and Problem

• As children increasingly use electronic devices for video content consumption, the authors have identified a lack of digital media support for children's social-emotional learning (SEL). The use of these devices is predominantly passive content consumption rather than active learning or reflection.
• Social-emotional skills, such as emotion recognition, emotion regulation, and self-management, are widely regarded as crucial for children's academic success and long-term well-being. However, existing SEL programs are primarily confined to school environments, with limited application in family settings.
• Traditional media consumption solutions often rely on parent-child co-viewing ("joint media engagement"), which demands significant time and energy from parents, making widespread implementation challenging.

Solution

• The authors developed a prototype system called eaSEL, which aims to (1) integrate SEL activities into children's video consumption and promote independent reflection and learning through these activities; (2) provide discussion prompts for parents to facilitate parent-child interaction without requiring joint video viewing.
• Innovations:
  • Utilizing large language models (LLMs, such as GPT-4) to automatically detect SEL moments in videos.
  • Generating learning activities for children based on SEL content, such as drawing, personal storytelling, or role-playing.
  • Creating personalized parent-child conversation prompts to help parents discuss social-emotional skills with their children without co-viewing the content.
• Implementation Steps:
  • Analyzing video transcripts using LLMs to detect "teachable moments" related to social-emotional skills.
  • Generating three types of outputs: (1) learning activities for children; (2) interaction prompts for parents based on children's activities; (3) brief summaries of video content.
  • Implementing the system interface on a tablet application, with dual interfaces for children and parents.

Research Outcomes

• Specific Findings:

  • Technical evaluations showed that LLMs performed well in detecting SEL moments and generating activities for children and parents, though certain areas require improvement (e.g., language complexity for younger children).
  • User studies revealed that after completing activities, children used more emotion-related language when recounting video content. For instance, the use of words like "kindness" and "anger" increased, indicating deeper resonance with the emotional themes of the videos.
  • Parents reported that the tool effectively enhanced children's reflective abilities and facilitated deeper parent-child conversations through activity design.

• Advantages Compared to Existing Solutions:

  • Provides activities suitable for independent use by children, addressing the high time investment required by traditional "joint media engagement."
  • Directly links activity content to the narrative of the videos, enhancing the practical relevance of reflection.
  • Generates actionable conversation prompts, enabling parents to interact with their children without co-viewing the videos.

• Experimental or Evaluation Results:

  • Most of the system-generated children's activities and parent-child conversation prompts were of good quality, successfully capturing key SEL skills and fostering meaningful exchanges.
  • User studies found that children improved their use of emotional language through the activity process, while parents felt the activities helped them better understand and engage in their children's SEL development.

• Limitations and Future Directions:

  • The language complexity of LLM-generated activities sometimes exceeds the capabilities of younger children. Additionally, certain video content may include themes unsuitable for reflection.
  • Long-term learning outcomes have not been verified, as the study focused on short-term experiments.
  • Parents found the video summaries in the system overly lengthy, leading to information overload; future designs may simplify the summaries.
  • Future research should explore broader SEL application scenarios, including use in diverse cultural or family environments, and investigate how to tailor activities to align with different parental educational goals.