Scientists from Nanyang Technological University, Singapore (NTU Singapore) have discovered that people exhibit less faith in AI (artificial intelligence)-suggested preventive care interventions than that of human health expert interventions.
Image Credit: Blue Planet Studio/Shutterstock.com
Preventive care interventions are actions to reduce health risks, for example, signing up for a health screening, getting a vaccination, and increasing physical activity.
The researchers discovered that highlighting the participation of a human health expert in an AI-suggested preventive care intervention can lead to better acceptance and success.
These outcomes indicate that the human factor remains significant even as the healthcare industry adopts AI to diagnose, screen, and treat patients more competently. The researchers state that the results could also contribute to a more efficient design of AI-suggested interventions.
Despite the potential of artificial intelligence to provide higher quality interventions, we found that people have lower trust in health interventions suggested by or derived from AI alone, as compared to those they perceive to be based on human expert opinion.
Hyeokkoo Eric Kwon, Study Lead Author and Assistant Professor, Nanyang Business School, Nanyang Technological University
Kwon adds, “Our study shows that the affective human element, which is linked to emotions and attitudes, remains important even as health interventions are increasingly guided by AI, and that such technology works best when complementing humans, rather than replacing them.”
The work reflects the efforts of NTU under NTU2025, the University’s five-year strategic plan that reports humanity's grand challenges, such as the technology’s impact on society. Led by NTU NBS at the intersection of business and healthcare technology, the research also highlights the strength of NTU and concentrates on interdisciplinary research.
The results were reported in an article in the scientific journal Production and Operations Management, which was co-authored with Assistant Professor Nakyung Kyung from the National University of Singapore.
Higher Acceptance for Human-Based Health Intervention
The study team employed 9,000 users of a mobile health app in South Korea to analyze user opinions of preventive health interventions suggested by artificial intelligence (AI) compared to those suggested by humans.
These users received a pop-up notification through the app, which motivated them to walk a particular number of steps. This was generated for every user through an AI algorithm. Then, the app measured the total steps taken by users who chose to approve this health intervention.
In the AI-driven intervention group, for 3,000 users, the pop-up notification read: “AI recommends that you walk (personalized step goal) in the next seven days. Would you like to participate?” Whereas, in the human-suggested intervention, another 3,000 users received a notification that read: “Health expert recommends that you walk (personalized step goal) in the next seven days. Would you like to participate?”
Neither AI nor a health expert suggestion was mentioned by a control group with 3,000 users who received the neutral intervention.
In the AI-suggested intervention group, 19% of the users accepted the intervention. Around 10% of this group eventually accomplished their personalized step goal at the end of the week. In the human-suggested intervention, comparatively more users accepted the intervention (22%) and achieved their goal (13%).
Improving the Effectiveness of AI-Suggested Interventions
The team broadened their study to include two groups of 3,000 users of the same mobile app.
An intervention disclosing the use of AI together with health experts was received by one group. An intervention explaining how AI interventions were produced was received by another group.
Compared to solely human-suggested or AI-suggested interventions, users showed more acceptance for the intervention, which exhibited how AI was employed to complement the opinion of a health expert (27%). The personalized step goal was achieved by 19% of this group.
Being transparent about how AI was employed to produce the personalized step goal also resulted in a greater acceptance rate (21%). The goal was achieved by 13% achieved of this group.
Although the research was performed in the preventive healthcare context, the researchers hope their results could be applicable in other contexts where affective trust has an important role to play, such as education, travel, legal, and insurance services.