This research demonstrates how confidence-weighted integration of human and LLM judgments leads to superior decisions than either alone.

• Human-AI teams outperform individual performance when confidence is well-calibrated
• Even when LLMs are more accurate, humans can still add significant value
• Confidence weighting is key to maximizing the benefits of human-AI collaboration
• Results validated in neuroscience forecasting tasks

For medical professionals, this framework offers a practical approach to integrate AI assistance while preserving human expertise and judgment in clinical decision-making.

Confidence-weighted integration of human and machine judgments for superior decision-making