Remembering with Reminiscope: Codesigning with Generative AI for Reminiscence Among Older Adults

Paper Title

Remembering with Reminiscope: Codesigning with Generative AI for Reminiscence Among Older Adults

Publication Info

• Topic area: Leveraging generative AI and interactive systems for reminiscence among older adults.
• Keywords: Generative AI, reminiscence, older adults, co-design, textile collaging, interactive systems, memory sharing, subjective wellbeing, group reminiscence, HCI.

Background and Problem

• Problem / challenge: Existing reminiscence technologies rely heavily on personal memory cues (e.g., photos, objects) and treat older adults as passive users, limiting opportunities for dynamic and contextualized memory recall. Older adults with cognitive or verbal challenges face additional barriers, and current systems rarely explore co-design approaches with generative AI.
• Significance: Reminiscence enhances older adults' psychological and social wellbeing, fostering connections and meaning-making. Addressing the challenges of memory recall and sharing can improve quality of life, especially for those in care homes or with limited social interactions.
• Motivation and related work: Prior work has developed technologies for individual and social reminiscence, such as photo-based systems, music playlists, and social robots. However, these approaches often lack personalization, dynamic content generation, and active user involvement in the design process. This study builds on these gaps by exploring generative AI as a co-design material for reminiscence.

Solution

• Proposed approach: The study introduces Reminiscope, an interactive artifact combining textile collaging and generative AI to create memory-related videos, facilitating both individual and group reminiscence.
• Novelty:
  1. Engaging older adults in co-design practices that integrate textile collaging with generative AI for memory reconstruction.
  2. Developing Reminiscope, a prototype that extends individual memory co-creation into a group reminiscence context.
  3. Identifying design considerations for AI-powered technologies that support reminiscence and social interaction among older adults.
• Procedure and key techniques:
  1. Study 1: Individual co-design sessions with 16 older adults, combining textile collaging and AI-generated 5-second videos based on participants' memories and verbal descriptions.
  2. Study 2: Group workshops with 15 participants (14 from Study 1, 1 new), introducing Reminiscope as a tangible, interactive system for memory sharing and social engagement.
  3. Data collection through think-aloud protocols, interviews, group discussions, and thematic analysis of qualitative data.

Results

• Concrete findings:
  • Textile collaging facilitated memory recall, with participants reporting clearer and more detailed recollections.
  • Participants rated AI contributions to memory videos as significant, with mixed perceptions of AI as a collaborator or tool.
  • AI-generated inaccuracies often triggered deeper reminiscence and enriched narratives.
  • Reminiscope fostered curiosity, memory sharing, and social bonding among participants.
• Advantage over baselines:
  • Enabled memory recall without reliance on physical memory cues.
  • Combined tangible, hands-on activities with generative AI, lowering barriers to engagement.
  • Facilitated meaningful social interactions even among unfamiliar participants.
• Experiments / evaluation:
  • Study 1: Individual co-design sessions with 16 participants, generating textile collages and AI-powered videos.
  • Study 2: Four workshops with 15 participants, using Reminiscope for group reminiscence and memory sharing.
  • Metrics: Qualitative analysis of participant feedback, memory recall depth, and social interaction dynamics.
• Limitations and future work:
  • Limited generalizability to older adults outside care homes or with higher AI literacy.
  • Quality of AI-generated content constrained by current model capabilities.
  • Ethical concerns regarding AI hallucination and memory distortion.
  • Future work includes exploring long-term deployment of Reminiscope and addressing accessibility for diverse user conditions.

Summary

This study explores the integration of generative AI and textile collaging to support reminiscence among older adults, culminating in the design of Reminiscope, an interactive artifact for group memory sharing. Findings from two studies reveal that co-designing with AI fosters memory recall, reduces technology-related anxiety, and promotes social bonding. The work highlights the potential of AI-powered tangible systems to enhance subjective wellbeing and provides design implications for future reminiscence technologies.