Objective:

To examine the risks of automation bias in AI-augmented diagnostics, particularly focusing on how it can mislead pathologists, and propose strategies for effective risk management.

Key Findings:

• 7% of cases showed automation bias where pathologists abandoned correct evaluations due to erroneous AI recommendations, highlighting significant risks to patient safety.
• Time pressure did not increase the frequency of automation bias but worsened the severity of errors when they occurred.
• Repeated reliance on AI may erode manual skills and independent judgment of clinicians, raising concerns about long-term competency.

Interpretation:

The integration of AI in pathology must be approached with caution, as automation bias poses significant risks to diagnostic accuracy and patient safety, necessitating robust risk management strategies.

Limitations:

• The study findings may not fully represent real-world complexities of clinical practice.
• Further research is needed to explore long-term effects of AI on clinician skills.
• Potential biases in the study sample may affect the generalizability of the findings.

Conclusion:

Awareness and structured interventions are essential to mitigate automation bias and ensure safe and effective use of AI in pathology.