Generative AI and Creative Mediums for Youth’s Emotion Regulation: An Interview Study with Clinicians

Paper Title

Generative AI and Creative Mediums for Youth’s Emotion Regulation: An Interview Study with Clinicians

Publication Info

• Topic area: Exploring the use of Generative AI (GenAI) in supporting youth emotion regulation (ER) through creative mediums.
• Keywords: Generative AI, emotion regulation, cognitive behavioral therapy, youth mental health, creative mediums, visual narratives, personalization, clinician perspectives, risks, design implications.

Background and Problem

• Problem / challenge: Youth face barriers in accessing and engaging with emotion regulation (ER) tools, such as limited personalization and interactivity in traditional methods like worksheets. While creative mediums like visuals and narratives can enhance ER, their analog nature restricts accessibility and adaptability. The potential of Generative AI (GenAI) in addressing these gaps remains underexplored.
• Significance: Effective ER is critical for youth mental health and lifelong well-being, as poor ER skills are linked to depression, anxiety, and other adverse outcomes. Addressing ER challenges through accessible, engaging, and personalized tools can significantly improve youth mental health support.
• Motivation and related work: Prior work has shown that visuals and narratives enhance ER by making abstract emotions tangible and engaging. However, existing tools lack personalization and interactivity, and technology-supported ER solutions often fail to address youth-specific needs. This study builds on these insights by investigating how GenAI can enhance ER through creative mediums while addressing associated risks.

Solution

• Proposed approach: The study explores the use of GenAI to generate personalized visuals and narratives for supporting youth ER, grounded in cognitive behavioral therapy (CBT) principles.
• Novelty:
  1. Examines clinicians’ perspectives on GenAI’s potential to enhance youth ER through creative mediums.
  2. Identifies benefits, risks, and design implications for GenAI-assisted ER tools.
  3. Proposes a technological probe based on Gross’s extended process model of ER to explore GenAI’s role in emotional identification, strategy selection, and implementation.
  4. Highlights specific use cases and safeguards for safe and effective GenAI integration into youth ER practices.
• Procedure and key techniques:
  • Conducted semi-structured interviews with 20 U.S.-based pediatric clinicians (psychotherapists, art therapists, psychiatrists).
  • Developed a GenAI-powered technological probe using GPT-4 and DALL·E to generate visuals and narratives aligned with CBT principles.
  • Analyzed data using reflexive thematic analysis to identify benefits, risks, and recommendations for GenAI-assisted ER.

Results

• Concrete findings:
  • GenAI-generated visuals and narratives can make abstract emotions more vivid, support personalized coping strategies, and bridge ER practices between clinics and homes.
  • Risks include GenAI’s unpredictability, potential to trigger trauma, and reinforcement of maladaptive thinking.
  • Clinicians emphasized the need for safeguards, such as contextual awareness, adult supervision, and symptom-specific adaptations.
• Advantage over baselines:
  • Compared to traditional worksheets and static visuals, GenAI offers dynamic, personalized, and interactive content that enhances engagement and relevance.
  • Provides a flexible starting point for emotional expression, particularly for youth with internalizing symptoms or neurodivergent profiles.
• Experiments / evaluation:
  • Interviews with 20 clinicians using a technological probe to simulate GenAI-assisted ER scenarios.
  • Participants included a balanced mix of art therapists, psychotherapists, psychologists, and psychiatrists with diverse professional settings (hospitals, private practices, universities).
  • Probe interactions focused on emotional identification, strategy selection, and implementation phases.
• Limitations and future work:
  • Limited to U.S.-based clinicians; future studies should include global perspectives.
  • Technological probe used only GPT-4 and DALL·E; other GenAI models may behave differently.
  • Single-session interviews; future work should assess long-term integration of GenAI tools in clinical and home settings.
  • Did not include youth or caregivers; future studies should involve these groups to validate findings and co-develop interventions.

Summary

This study investigates the potential of Generative AI (GenAI) to support youth emotion regulation (ER) through personalized visuals and narratives, grounded in cognitive behavioral therapy (CBT). Interviews with 20 clinicians revealed that GenAI could enhance emotional awareness, expression, and coping practice by offering dynamic, engaging, and personalized content. However, risks such as triggering trauma and reinforcing maladaptive thinking necessitate safeguards, including adult supervision and symptom-specific adaptations. The findings contribute to the design of safe and effective GenAI tools for youth ER, emphasizing the importance of integrating clinical oversight and family involvement.