Measuring Cognitive Functioning in Older Adults Using Artificial Intelligence Chatbot Narrative Interviews

This study examined whether older adults (age 55+) could effectively interact with an Intelligent Cognitive Assistant (ICA) chatbot during a narrative interview, how they perceived the system, and whether textual data from the interview could be used to assess cognitive ability and relate to cognitive functioning and health outcomes. Participants were recruited from Prolific, a crowdsourcing platform (N = 164), and completed four cognitive ability tasks, a narrative interview with an ICA, and a battery of survey measures. Our findings indicate that older adults can engage meaningfully with an ICA and provide enough textual cues to assess personality on par with previous studies on younger adult samples. These older adults also reported that they enjoyed the interaction and found it easy to use and concentrate throughout the interview. We were able to use the textual responses from the narrative interview to assess cognitive ability to a limited extent, primarily capturing aspects of verbal fluency; however, correlations of narrative-based cognitive ability scores with self-reported cognitive functioning and health outcomes were weak. Overall, these results suggest that ICA systems are not only feasible but also engaging and easy for older adults to use. This presents a promising first step towards the development of automated, scalable tools for cognitive evaluation. We encourage future research to design and validate an ICA system specifically for cognitive evaluation.