Analysis · Technology
A More Human Health Chatbot Is Not Always a More Welcome One
By AWEI · AI-compiled · Published · Analysis prepared · 1 source · doi.org
A health chatbot’s distance from human interaction can be part of its appeal. In research mapped by a 2022 review, some participants valued anonymity when discussing sensitive subjects, while others wanted more empathy. Making a chatbot feel more human therefore emerges as a design question with competing preferences, rather than an obvious upgrade.
Han Shi Jocelyn Chew and Palakorn Achananuparp reached this picture through a scoping review of 26 articles. They searched nine databases for studies available through June 21, 2021, supplemented by Google Scholar and reference searches. They organized reported perceptions and needs across patients, the public and professional groups; they did not test whether a particular chatbot design increased adoption.
The revealing detail was disagreement over personification. An included analysis of mental health chatbot perceptions described responses ranging from warm and friendly to unsympathetic and rude. Some users found lifelike features unsettling. Another included study reported that perceived anonymity could make intimate or uncomfortable health topics easier to discuss. These findings do not establish which design produces more disclosure.
Role clarity added another complication. In one included study, patients were unsure whether to treat a chatbot as a physician or an adviser. Elsewhere, users wanted clearer explanations of a chatbot’s purpose. A personable interface and an understandable role are separate requirements: enthusiasm for the interaction does not demonstrate understanding of its limits.
The evidence cannot identify an optimal personality. Studies differed substantially in methods, aims and populations, only English-language articles were included, and participants’ knowledge of AI could affect their responses. The reviewers could not rank the importance of the reported needs. Proposed improvements remained suggestions, not proven causes of adoption.
For deployment teams, this supports testing warmth, perceived anonymity and role comprehension separately with intended users. A friendly demo alone cannot establish that people understand the service or will use it appropriately.
Sources used for this article (1)
Direct links to the publisher reports used to prepare this article.
- Source 1
- Perceptions and Needs of Artificial Intelligence in Health Care to Increase Adoption: Scoping Review Han Shi Jocelyn Chew, Palakorn Achananuparp
