When Doctors Lose Trust in AI: What Should We Discuss at TechCrunch Disrupt?

At 10 AM, in the radiology department of a top-tier hospital in Shanghai, Director Zhang pushed the freshly run lung nodule AI screening results right in front of me. The screen was covered with dense red bounding boxes, each labeled with confidence scores like "0.87" and "0.92." He sighed: "Your model missed a 4mm mixed ground-glass nodule here last week, and today it flagged three false positives. I see two hundred patients a day; I don't have time to do secondary verification for you."