Autiverse: Eliciting Autistic Adolescents' Daily Narratives through AI-guided Multimodal Journaling
Authors
Paper Title
Autiverse: Eliciting Autistic Adolescents' Daily Narratives through AI-guided Multimodal Journaling
Publication Info
- Topic area: AI-guided journaling for autistic adolescents
- Keywords: Autism, adolescents, journaling, AI peer, multimodal support, narrative construction, scaffolding, ABC-E format, visual aids, conversational AI
Background and Problem
- Problem / challenge: Autistic adolescents face difficulties in organizing and expressing daily experiences due to executive functioning challenges and text-centric journaling methods that overlook their strengths in visual thinking.
- Significance: Supporting narrative skills in autistic adolescents can improve communication, emotional expression, and identity development, which are critical during adolescence.
- Motivation and related work: Prior research has explored structured self-tracking and conversational AI for journaling but has largely focused on text-based methods or predefined phenomena, leaving gaps in multimodal, adaptive journaling systems tailored for autistic adolescents.
Solution
- Proposed approach: Autiverse, a tablet-based AI-guided multimodal journaling system that combines conversational scaffolding with visual aids to help autistic adolescents construct daily narratives.
- Novelty:
- Integration of a peer-like AI character to scaffold narrative construction through conversational prompts.
- Use of the ABC-E format (Antecedent, Behavior, Consequence, Emotion) for structured storytelling.
- Transformation of narratives into editable four-panel comic strips for visual representation.
- Empirical insights from deployment with autistic adolescent-parent dyads.
- Procedure and key techniques:
- Adolescents select a place and people involved in an event.
- The AI peer elicits details through step-by-step questions, organizes them into the ABC-E format, and visualizes the story in comic panels.
- Adolescents review and revise the narrative collaboratively with the AI.
- Finalized narratives are titled and rewarded with stamps to encourage engagement.
Results
- Concrete findings:
- Adolescents recorded 122 journal entries over two weeks, averaging 12.2 entries per participant.
- Sessions lasted 9 minutes 43 seconds on average, with 46.92 conversational turns per session.
- 83% of Behavior components were revised, and 86% of Emotion components were elicited during the Elaboration phase.
- Advantage over baselines:
- Autiverse reduced cognitive load by breaking journaling into manageable steps and leveraging visual aids.
- Adolescents expressed more nuanced emotions and organized narratives more coherently than in conventional journaling methods.
- Experiments / evaluation:
- Two-week deployment study with 10 autistic adolescent-parent dyads in South Korea.
- Mixed-method analysis including interaction logs, surveys, and interviews.
- Parents reported gaining new insights into their child’s daily events and emotions in 79% of cases.
- Limitations and future work:
- Limited generalizability due to the small sample size and cultural context (South Korea).
- Need for adaptive scaffolding to accommodate diverse autism profiles.
- Potential socio-emotional risks of prolonged interaction with peer-like AI.
- Future research should explore gamification features and broader population studies.
Summary
Autiverse is an AI-guided multimodal journaling system designed to support autistic adolescents in constructing coherent narratives of daily experiences. By combining conversational scaffolding with visual aids in the ABC-E format, Autiverse reduces cognitive load and fosters emotional reflection. A two-week deployment study demonstrated its effectiveness in enhancing narrative skills, emotional expression, and parent-adolescent communication. Future work should focus on adaptive scaffolding, addressing socio-emotional risks, and expanding applicability across diverse autism profiles and cultural contexts.
Research Questions / Practical Problems
Question signals indexed for this paper.
No related papers with ≥60% similarity
Based on Jaccard similarity of research subtopics & professions (≥60%)