Human-AI Reading Simulator

Play the role of a radiologist reviewing 20 simulated cases alongside an AI probability, then see whether you, the AI, or the combined team performed best.

  • human-AI interaction
  • automation bias
  • joint evaluation
  • clinical workflow
AI accuracy
Human accuracy
Cases AI influenced

AI vs. human assessments

threshold0246810121416182000.20.40.60.81CaseProbability
AIHuman

Assessment history

"Influenced" means your assessment landed within 0.15 of the AI probability.
CaseAI prob.Your prob.TruthAI influenced?
What this simulates
Reading-room performance is a property of the human-AI system, not the algorithm alone. Automation bias, anchoring and alert fatigue can erase — or reverse — a model's raw accuracy advantage.