TLDR: Building genuine trust in healthcare AI necessitates a shift from mere system design to a focus on real-world application and continuous feedback from clinicians and patients. While AI offers significant potential, its widespread adoption hinges on addressing concerns around data privacy, transparency, and practical utility, as highlighted by recent surveys and initiatives in regions like Singapore.
The integration of Artificial Intelligence (AI) into healthcare promises transformative advancements, from diagnostics to operational efficiencies. However, a critical challenge remains: fostering trust among the very clinicians and patients it aims to serve. Experts argue that this trust cannot simply be ‘designed’ into a system; it must be ‘felt’ through lived experience and continuous, responsive feedback loops.
According to Adriana Banozic-Tang, PhD Adjunct Faculty, and Heng Wang, Professor of Law, both from Singapore Management University, current approaches to building trust in healthcare AI often over-rely on theoretical system design, neglecting the practical realities of medical care. They emphasize that governance in this rapidly evolving field lags behind, leading to inconsistent adoption despite the technology’s potential.
Recent data underscores this gap. A 2025 survey by the American Medical Association (AMA) revealed that while a significant 68% of physicians recognized the value in AI tools, and 66% were already utilizing them, nearly half (47%) identified increased oversight from medical practitioners as the most crucial regulatory step for building trust in AI-generated recommendations. This suggests a strong desire for human accountability and understanding in AI deployment.
Patient confidence also remains a hurdle. A 2023 YouGov survey in Singapore found that only 14% of residents would engage with AI-enabled mental health counseling, indicating a substantial trust deficit. This sentiment is echoed in a YouGov survey sponsored by Lexacom, which found that 74% of general practice patients do not trust large companies, a factor that can impact their acceptance of AI solutions.
Historical events, such as the 2018 Singhealth cyberattack in Singapore, which compromised 1.5 million patient records, highlight the critical need for robust data security and privacy measures to underpin trust. In response, the Singapore Ministry of Health introduced its 2021 AI in Healthcare Guidelines, promoting safeguards like explainability, human oversight, and clear risk communication – steps aimed at embedding trust rather than merely enforcing compliance.
Practical examples demonstrate the potential of AI when trust is cultivated. Singapore General Hospital’s AI2D model (Augmented Intelligence in Infectious Diseases), which assists doctors in determining antibiotic necessity, has achieved 90% accuracy in early validations. Such tools, when proven effective and transparent, contribute to building confidence.
However, challenges persist. A 2025 npj Digital Medicine perspective noted that continuous-monitoring pipelines and retraining workflows are becoming significant operating costs, and adjustments for ‘data drift’ (shifting data patterns) strain institutions without dedicated engineering teams. Furthermore, the issue of ‘hallucinations’ – where AI generates clinical entities not present in the original data – remains a concern, necessitating tools like Lexacom’s Patient Shield, which redacts sensitive data before AI processing.
Also Read:
- Artificial Intelligence and Generative AI Driving Transformative Growth in Healthcare Sector
- Australian Regulator Prioritizes New AI Guidance for Medical Device Sector
Experts stress that trust is not static; it is built through ‘context, feedback, and use.’ Context involves aligning AI systems with existing clinical workflows and data sources. Feedback requires continuous input from users and performance monitoring to identify and mitigate risks. And ‘use’ emphasizes that trust is earned through repeated, safe, and effective interactions. As one expert stated, ‘Trust cannot be retrofitted.’ The call to action is clear: to truly shape the future of healthcare AI, governance must evolve from the ground up, ensuring trust is earned through lived experience, not merely designed on paper.
