Bridging the Cognitive Gap: Co-Designing and Evaluating a Voice-Enabled Community Chatbot for Older Adults

A research team led by Feng Chen has published a study detailing the co-design and evaluation of a voice-enabled community chatbot specifically for older adults. The work, conducted at a continuing care retirement community in the Pacific Northwest, aimed to bridge the digital divide created by traditional portals that often lead to digital avoidance. The team's innovative 'Glass Box' approach paired a voice-interface for a Large Language Model (LLM) with intentional AI literacy workshops, moving beyond the opaque 'Black Box' nature of typical generative AI systems. This method focused on multimodal accessibility and education to foster informed reliance rather than blind trust.

Through mixed-methods workshops involving 25 participants, the researchers achieved statistically significant results. Participants' technical understanding of the AI system improved (p=0.004), and their perception of its transparency increased (p=0.001). The study found that voice input successfully reduced cognitive load for many users, facilitating more natural interaction. However, a critical finding emerged: for the oldest cohort (aged 80 and above), usability scores showed a significant negative correlation (r=-0.50), indicating that voice-only interfaces are not a universal solution. This key insight pushes the field toward developing truly age-inclusive AI that must evolve beyond both touch-based and voice-based interfaces toward intuitive, zero-touch navigation systems.