What the study found
The study found that combining vocal and verbal responses with large language model context can improve error prevention and error handling in voice assistants. It also found that interaction-level challenges still remain for older adults, even with these technical improvements.
Why the authors say this matters
The authors suggest that standard ways of studying voice assistant use, such as usage logs and post hoc interviews, do not fully capture older adults' error experiences. They conclude that better understanding of these situations may help future voice assistants align more closely with older adults' expectations and lived experiences.
What the researchers tested
The researchers carried out a month-long in-home study with 15 older adults. They equipped homes with smart speakers and custom audio recorders to collect in-the-wild audio interaction data, and midway through the study they deployed a ChatGPT-powered voice assistant to examine how large language model integration changed interaction dynamics during errors.
What worked and what didn't
According to the abstract, technical improvements that use vocal and verbal responses together with large language model context can enhance error prevention and management. However, the study also reports that interaction-level problems remained, especially those specific to older adults.
What to keep in mind
The abstract does not describe specific limitations beyond the small sample of 15 older adults and the in-home study setting. It also does not provide detailed results for the ChatGPT-powered voice assistant beyond its role in examining interaction changes during errors.
Key points
- The study focused on older adults' errors and conversation breakdowns with commercial voice assistants.
- Fifteen older adults' homes were equipped with smart speakers and custom audio recorders for audio-based interaction data collection.
- A ChatGPT-powered voice assistant was introduced midway through the month-long study.
- Technical improvements using vocal and verbal responses plus large language model context may help prevent and manage errors.
- Interaction-level challenges for older adults still remained.
Disclosure
- Research title:
- Older adults faced interaction challenges with voice assistants
- Authors:
- Amama Mahmood, Junxiang Wang, Chien-Ming Huang
- Institutions:
- Carnegie Mellon University, Carnegie Robotics (United States), Johns Hopkins University, Johns Hopkins University
- Publication date:
- 2026-02-06
- DOI:
- 10.1145/3796236
- OpenAlex record:
- View
Get the weekly research newsletter
Stay current with peer-reviewed research without reading academic papers — one filtered digest, every Friday.