Memory Printer: Exploring Everyday Reminiscing by Combining Slow Design with Generative AI-based Image Creation

Slow, deliberate interaction with AI can increase user agency: The study found that breaking image generation into incremental steps with physical control (using a wooden scraper to gradually reveal images) gave participants a heightened sense of control and engagement compared to typical fast, automated AI tools, suggesting that interaction design significantly shapes how people experience AI systems.

Tangible outputs strengthen emotional connection to AI-generated content: Participants experienced "vivid memory evocation" when AI-generated images were physically printed rather than displayed digitally, indicating that embodying AI outputs in physical form can deepen emotional resonance and make the technology feel more meaningful in personal contexts.

Generative AI poses specific risks in emotionally sensitive applications: The research identified critical concerns including false memory formation (where AI-generated reconstructions could be mistaken for actual memories), algorithmic bias, and data privacy issues—highlighting that deploying AI for personal memory work requires careful safeguards beyond what current web-based tools provide.

Making AI processes "legible" to users matters for trust: By designing the system to make the generative process experientially visible (through layered, step-by-step revelation), participants could better understand how the AI was creating images, suggesting that transparency about how AI works is essential for responsible deployment in intimate, personal contexts.

Design choices fundamentally shape human-AI relationships: The paper demonstrates that how we design interactions with AI—not just the AI itself—determines whether users feel empowered or passive, trustful or skeptical, underscoring that responsibility in AI requires attention to interaction design alongside technical considerations.