Who Is the User? Designing AI Family-Memory Platforms When Benefits, Labor, and Risk Are Unequally Distributed

Abstract Book of the 10th International Conference on Advanced Research in Social Sciences

Year: 2026

[PDF]

Who Is the User? Designing AI Family-Memory Platforms When Benefits, Labor, and Risk Are Unequally Distributed

Yongjun Jo

ABSTRACT:

AI family-memory systems can benefit descendants while older relatives bear narrative labor, consent decisions, and exposure risks. This study examines that asymmetry through LegacyLink, a working prototype developed by the author in response to the guilt of knowing little about older relatives’ pasts and rarely initiating such conversations. A relative used the system several times and consented to being discussed; the stories themselves are excluded from analysis. Using reflexive research-through-design, the study developed TRACE (Triggers and Transformations, Relational Roles, Negotiated Authority, Costs, and Exit), linking a system’s initiating motive to its distribution of labor and narrative authority. The framework was refined through an interface and policy audit of five AI memory systems and reapplied to LegacyLink. The audit distinguished purchaser, beneficiary, storyteller, and data subject, roles that are often simplified into the “user”. It found that AI prompting shifted conversational initiative without removing storyteller labor, consent frequently followed disclosure, and summaries could gain authority over attributed testimony. In LegacyLink, repeated approval mechanisms revealed a paradox: greater narrative control also increased storyteller workload. These findings define affective I-methodology, in which a designer’s emotional need shapes problem formulation even when the intended user is accurately represented. LegacyLink’s redesign maps each finding to a mechanism. Scope-based review assigns approval through predefined rules for each operation and audience. Protected silence treats nonresponse or withdrawal as final, preventing automated follow-up or inference. Parallel accounts preserve conflicting memories with explicit authorship. TRACE therefore converts “who is the user?” into an auditable design question.

Keywords: Affective I-Methodology; Narrative Authority; Relational Autonomy; Research-Through-Design; Storyteller Agency