Publication Details
Abstract
This scoping review focuses on the perspectives of patients and surgeons on AI-assisted robotic surgery (RAS) in perioperative settings, an area that current literature covers at best indirectly. Most work on AI in surgery usually includes RAS in a broad discussion while failing to acknowledge it separately, and without examination of whether the patient would be willing to pay for the procedure. Searches followed PRISMA-ScR guidelines across Google Scholar, PubMed, JSTOR, and ScienceDirect, covering studies between 2022-2026. Only three sources met the inclusion criteria. The findings have been analyzed based on five different dimensions: knowledge, trust, previous experience, expectations, and willingness to pay. A common finding across the three sources was the fact that people trusted RAS more than they understood its technical ability. People consistently overestimated how autonomous the robot actually is and underestimated how much control the surgeons have. That misunderstanding caused anxiety about malfunction. The same was true for the surgeons whose knowledge of AI increased from 14.5% in 2021 to 44.6% in 2024, but most still couldn’t explain basic concepts like machine learning and computer vision. The most notable aspect was the trend of change: as exposure to AI increased, the preference for full autonomy decreased rather than going up. Across both groups, trust held up best when AI stayed in a decision-support role rather than operating independently. This review is the first to treat RAS separately rather than folding it into the broader context of perioperative AI. It also identifies how little the field knows about willingness to pay and prior experience - both nearly absent from current studies. Surgeon acceptance closely correlates to how much control they get to keep. But, acceptance by patients is quite ambiguous; it appears related to misunderstanding robotic autonomy rather than to the technology’s actual performance, but, since the pattern is based on three sources only, it can be viewed as a hypothesis to be tested. The review’s obvious limitations include a small source base, and English-only search, and no formal quality assessment of the studies included. Future work should study RAS on its own terms rather than borrowing conclusions from general surgical AI research. Most importantly, it should directly analyze patients who paid for the AI procedure and what they have actually experienced.