AI workflow variations reduced perceived liability among hypothetical US jurors in an online vignette experimentUsing AI tools in radiology workflows can lower perceived liability for doctors in some situations

medRxiv Published May 23, 2026 Study authors: Song, E. C.; Bernstein, M. H.; Sheppard, B.; Bruno, M. A.; Baird, G. L. DOI ↗ Editorial oversight: Dr. Lars van Dijk, PhD · Surgical, Procedural & Diagnostic

AI-generated summary of the cited source, checked by automated accuracy review. How we work

Key Takeaway

Note that strategic workflow design may mitigate perceived liability risk in AI-assisted radiology.

This online vignette experiment included a sample size of n=2,347 United States adults who acted as hypothetical jurors. The setting was online. The intervention or exposure involved 21 conditions involving a hypothetical AI system that varied by workflow type, documentation of initial interpretation, change of interpretation after AI output, AI detection of abnormality, and provision of AI error rates. The comparator was a no-AI control. The study phase was not reported.

The primary outcome was perceived liability assessed by whether the radiologist met their duty of care. Main results showed that perceived liability differed across conditions with a p value of p<0.0001. Specific conditions demonstrated a reduction in perceived liability with p values of p=0.0125, p=0.0038, and p=0.0035. Another condition showed perceived liability was lower with a p value of p<0.0001. In one condition, the greatest liability occurred with an increase. Absolute numbers were not reported for these outcomes.

Safety and tolerability data were not reported. Adverse events, serious adverse events, discontinuations, and tolerability were not reported. The study type was an online vignette experiment. Funding or conflicts were not reported. Limitations were not reported. Practice relevance indicates that strategic workflow design is critical for successful AI implementation that can mitigate malpractice risk.

Researchers asked 2,347 adults in the United States to imagine they were jurors deciding if a doctor did their job well. These adults looked at 21 different scenarios where doctors used a pretend AI system to help read medical images. Some scenarios showed the AI finding errors, while others showed the AI making mistakes. The study compared these AI situations to cases where no AI was used at all.

The main finding was that perceived liability changed depending on how the AI was used. In several situations, using the AI system actually reduced the chance that jurors would think the doctor was at fault. This happened when the workflow was designed carefully or when the AI helped catch problems early.

However, the greatest liability occurred in specific conditions where the AI output was not handled well. This shows that simply adding AI is not enough. The way doctors and hospitals design their workflow is critical for success. Strategic planning can help reduce malpractice risk when implementing new technology.

This research highlights that AI tools are not a magic solution. They must be integrated thoughtfully to protect doctors. Hospitals should focus on workflow design to ensure AI helps rather than hurts their legal standing.

What this means for you:

Careful AI workflow design can lower perceived liability for doctors in some legal scenarios.

Study Details

Study typeRct

Sample sizen = 2,347

EvidenceLevel 2

PublishedMay 2026

View Original Abstract ↓

Background: With growing impetus to integrate artificial intelligence (AI) tools into radiology, clinical practices must navigate workflow redesign. This carries implications for medical malpractice liability. Methods: We conducted an online vignette experiment with United States adults who acted as hypothetical jurors in a malpractice case involving a missed intracranial hemorrhage. Participants (n=2,347) were randomized to one of 22 conditions: a no-AI control and 21 conditions involving a hypothetical AI system. These twenty-one conditions varied by whether (1) a single-read or double-read workflow was used, (2) the radiologist's initial interpretation was documented, (3) the radiologist changed their interpretation after viewing AI output, (4) the AI detected the abnormality, and (5) the AI error rate--False Discovery Rate (FDR) or False Omission Rate (FOR--was provided to participants only, both participants and radiologist, or neither. The primary outcome was perceived liability, assessed by whether the radiologist met their duty of care. Findings: Perceived liability differed across conditions (p<0.0001). Double-read workflows (p<0.0001), documenting initial interpretations (p=0.0125), and providing participants with AI error rates, including the FDR (p=0.0038) or FOR (p=0.0035), reduced perceived liability. Liability was also lower when AI was incorrect (p<0.0001). Radiologists' awareness of AI error rates did not significantly impact liability. Notably, we observed an erroneous change penalty: the greatest liability occurred when radiologists initially identified an abnormality but later changed their interpretation to normal after seeing that AI identified the case as normal; conversely, perceived liability was lowest with documented, double-read workflows. Interpretation: Double-read workflows with documented initial interpretations and disclosure of AI error rates reduce perceived liability, though changing a correct initial interpretation increases it. Strategic workflow design is critical for successful AI implementation that can mitigate malpractice risk.

AI workflow variations reduced perceived liability among hypothetical US jurors in an online vignette experimentUsing AI tools in radiology workflows can lower perceived liability for doctors in some situations

Study Details

Machine Learning Model Performance for Predicting Contrast Induced Nephropathy in Coronary Interventions

Machine learning models help predict kidney damage during heart procedures

Clinical research that matters. Delivered to your inbox.

AI workflow variations reduced perceived liability among hypothetical US jurors in an online vignette experimentUsing AI tools in radiology workflows can lower perceived liability for doctors in some situations

Study Details

Machine Learning Model Performance for Predicting Contrast Induced Nephropathy in Coronary Interventions

Machine learning models help predict kidney damage during heart procedures

Clinical research that matters. Delivered to your inbox.

Related in Radiology & Imaging

From Other Specialties