Human-like features can build trust, but when they conflict with patient expectations for transparency in healthcare settings, they can undermine the very trust chatbots are trying to build.
Doris Dippold
However, the study also identified clear points of friction. Many patients find follow-up messages sent within 24 hours to be intrusive, and say commanding language like “make an appointment” is more offensive than helpful. For patients who were dealing with mental health issues, neurodivergent conditions, or seeking more caregiving responsibilities, the pressure to respond immediately felt unfair.
Ethical concerns were the most consistently negative finding across studies. Patients expressed concerns about data security, identity theft, and blurring the line between humans and AI. The chatbot’s statement that users can “chat with me as if I were a real person” backfired on many people, who found it more suspicious than reassuring.
Dr. Doris Dippold, lead author of the study and Associate Professor of Cross-Cultural Communication at the University of Surrey, said: “Our analysis shows that anthropomorphism is not universally positive. Human-like characteristics can build trust, but when it conflicts with patients’ expectations for transparency in healthcare settings, it undermines the very trust chatbots are trying to build.”
This research suggests that healthcare chatbot design should follow the following key principles:
- Help people achieve their goals
- Give you control over your decisions
- respond appropriately to their needs
- treat them with respect
- Ensuring fairness
- Be transparent about how the technology works.
The findings are particularly important given that cervical cancer screening uptake across the UK is expected to fall by 5.3% from 2023 to 2024, and ethnic minority groups are consistently under-represented in screening programmes. The GP surgery where Asa underwent the trial serves a highly diverse and socio-economically disadvantaged community in Islington and makes fair and accessible communication a priority.
This study highlights the importance of building trust between patients and medical chatbots through communication that feels respectful, transparent, and collaborative.
Dr Dippold added: “Feeling seen, valued and emotionally supported is not a luxury feature of medical AI; it is a condition of access. If patients disengage because they find the chatbot intrusive or untrustworthy, health services will lose patients altogether.”
sauce: University of Surrey
