Artificial Intelligence in Ophthalmology: Trust, Bias, and Responsibility from the Perspective of Medical Students and Ophthalmologists

Yacoub A. Yousef^1*Areen Shdeifat²Leen Yousef³Mona Mohammad¹Tamara AlNawaiseh²Leena Abdullah²Mutasem Elfalah²Iyad Sultan¹Ibrahim AlNawaiseh¹

• ¹King Hussein Cancer Center, Amman, Jordan
• ²The University of Jordan School of Medicine, Amman, Jordan
• ³IGCSE, Islamic Educational College, Amman, Jordan

Abstract Background: Artificial intelligence (AI) is increasingly integrated into ophthalmology, offering advances in diagnostic accuracy and surgical decision support. However, perceptions, trust, and ethical concerns regarding AI among medical students and ophthalmologists remain insufficiently explored. Methods: A cross-sectional survey was conducted among 525 participants, including 353 medical students and 172 ophthalmologists. The questionnaire assessed perceptions of diagnostic reliability, AI-assisted surgical outcomes, responsibility attribution, ethical concerns, and trust in AI compared with clinician judgment. Results: Most participants in both groups perceived human clinical expertise as more reliable for diagnosis than AI-driven systems (medical students 80%; ophthalmologists 72%; p = 0.054). In contrast, more than half of respondents believed AI-assisted surgery could achieve superior outcomes compared with manual techniques (medical students 55%; ophthalmologists 56%). Primary responsibility for AI-related clinical outcomes was most commonly attributed to physicians rather than AI developers (medical students 62%; ophthalmologists 66%; p = 0.666), and bias was identified as the leading ethical concern (70% of medical students and 75% of ophthalmologists). Approximately 70% of participants viewed AI as a complementary tool rather than a replacement for ophthalmologists, although nearly half anticipated AI might replace some optometric functions. In human–AI disagreement scenarios, trust was context-dependent: 77– 79% deferred to AI when it contraindicated surgery recommended by clinicians, whereas 91% favored clinician judgment when AI recommended surgery against clinical advice. Early-career ophthalmologists demonstrated greater support for AI-assisted surgery compared with senior colleagues (p = 0.013). Conclusion: Both medical students and ophthalmologists recognize AI's potential in ophthalmology, particularly for surgical applications, while continuing to prioritize human expertise for diagnosis. AI is largely viewed as a complementary tool, with ethical concerns surrounding bias and responsibility remaining prominent. Trust in AI varies by clinical context, and acceptance of AI-assisted surgery is greater among early-career ophthalmologists.