Evidence-sufficiency prompting reduces unsafe overconfidence in clinical LLMs, but the measured gain is judge-dependent (24.7 vs 13.1 points) and carries model-specific helpfulness costs.
Code and data release: https://github.com/aiden-ygu/health-ai- readiness-eval/tree/v1.0.0
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Judge-dependent safety gains and model-specific helpfulness costs of evidence-sufficiency prompting in clinical LLMs
Evidence-sufficiency prompting reduces unsafe overconfidence in clinical LLMs, but the measured gain is judge-dependent (24.7 vs 13.1 points) and carries model-specific helpfulness costs.