Bridging, Open, Discerning, Humble, Inquiring
The courage to ask before answering.
An engineering framework for curiosity driven and humble AI in clinical decision support.
This repository hosts the project website: criticaldata.github.io/bodhi (source: github.com/criticaldata/bodhi)
pip install bodhi-llm| Resource | Link |
|---|---|
| Website | criticaldata.github.io/bodhi |
| Python package | bodhi-llm on PyPI |
| Evaluation scripts | humbleai-healthbench |
| Curiosity driven QMoE | curious-qmoe |
| ModaLens code | criticaldata/MODALENS |
| Deterministic math solver code | BODHI-Medcalc-NeurIPS |
- ModaLens: Measuring Image Sensitivity in Report-Conditioned Medical VLMs (arXiv, 2026) · code
- Towards a Deterministic Math Solver for Clinical Language Models (NeurIPS Global South CV Workshop, accepted; arXiv, 2026) · code
- Beyond Overconfidence: Embedding Curiosity and Humility for Ethical Medical AI (PLOS Digital Health, 2026)
- An Engineering Framework for Curiosity Driven and Humble AI in Clinical Decision Support (BMJ Health & Care Informatics 33(1): e101877, 2026; preprint) · code
- Humility and Curiosity in Human-AI Systems for Health Care (The Lancet, 2025)
- Uncertainty Makes It Stable: Curiosity Driven Quantized Mixture of Experts (arXiv, 2025) · code
- Teaching Machines to Doubt (Nature Medicine, 2025)
- The BODHI Framework: Engineering and Testing Epistemic Virtues for Collaborative Intelligence in Clinical Decision Support. NeuroMonster 2026, AI Spotlights, June 2026 (schedule).
- Sebastián Andrés Cajas Ordoñez (sebasmos@mit.edu)
- Maximin Lange (mlange2@mit.edu)
- Leo Anthony Celi (Google Scholar)
© 2026 BODHI Research Group. Developed by MIT Critical Data and partner institutions worldwide.