2026
MIRA: Designing for Disruption, a Framework for Routine Recovery & Autonomy-Preserving AI for Elder Care
Sapna Bajaj
This thesis explores how assistive technology can better support older adults who want to maintain independence while aging in place, particularly during periods of recovery when interruptions, fatigue, and cognitive friction can make familiar routines more difficult to resume. Research with older adults and caregivers revealed that many existing technologies emphasize reminders, surveillance, and task compliance, approaches that can increase anxiety, undermine dignity, and contribute to technology abandonment. In response, the project reframes independence not as the flawless completion of daily tasks, but as the ability to recover when routines are disrupted.
The resulting concept, MIRA, is a voice-first, ambient AI framework designed to provide
calm, contextual support during these moments of disruption. Its Routine Recovery
Loop detects hesitation, stabilizes cognitive load, offers a small and actionable
next step, and then returns control to the user. Through interaction patterns such
as action-oriented scaffolding, non-reflective pragmatism, and threshold lowering,
MIRA supports routines without taking them over. The system also introduces an information-asymmetry
model that provides caregivers with meaningful reassurance while intentionally protecting
the older adult鈥檚 private moments and mistakes, proposing a more respectful and autonomy-preserving
direction for future elder-care technology.
