From data analysis to robotic surgery: nurses’ ranking of autonomous AI in clinical care

Background

Artificial intelligence (AI) is entering clinical pathways, yet acceptance varies by clinical task. From a public health perspective, understanding acceptable AI applications is important for safe implementation. This study compared and ranked nurses’ acceptance of different AI applications in clinical practice.

Methods

A cross-sectional online survey was conducted among nurses in Poland (N = 800). Quota sampling obtained a large, demographically structured sample. Respondents assessed eight AI applications in diagnostics and treatment on a 1-10 scale, assuming no or minimal involvement of medical staff. Ordinal items were analyzed using the Friedman test, followed by Bonferroni-corrected pairwise comparisons. The study was funded by the National Science Centre, Poland (Grant No UMO-2021/43/B/NZ7/02140).

Results

A nonparametric repeated-measures test showed significant within-respondent differences in acceptance rankings across the eight AI applications (χ²(7)=103.55, p < 0.001). Acceptance (mean±SD) ranged from 5.12±2.61 for remote consultation without medical staff involvement to 6.21±2.16 for genetic/clinical data analysis to guide therapy. Other relatively highly ranked applications included medical image analysis (5.97±2.23) and adherence monitoring (5.93±2.26). The two lowest-ranked applications were remote consultation without medical staff involvement (5.12±2.61) and autonomous robotic surgery (5.37±2.49). Bonferroni-corrected comparisons supported this gradient, especially the contrast between genetic/clinical data analysis and the lowest-ranked applications.

Conclusions

Nurses’ acceptance of AI was not uniform but followed a clear hierarchy, with higher acceptance of data-intensive analytical applications and lower acceptance of autonomous patient-facing or procedural functions. Implementation strategies should differentiate between AI use cases and address concerns related to autonomy, professional oversight, and patient safety.

Key messages

• Nurses distinguish between AI use cases, accepting analytical applications more than autonomous patient-facing or procedural functions.

• Implementation strategies should address autonomy, oversight, and patient safety rather than treating clinical AI as one category.