Entity · paper

Compositional Reasoning Depth Predicts Clinical AI Failure

paperactivecompositional-reasoning-depth-predicts-clinical-ai-failure-de9495aa·1 events·first seen Jun 16, 2026

Aliases: Compositional Reasoning Depth Predicts Clinical AI Failure

Co-occurring entities

Claude Sonnet MedAlign GPT-4o OpenAI GPT-5.5 Anthropic

More like this (12)

Evaluating Multi-Turn Multimodal Diagnostic Reasoning on Challenging Real-World Clinical Cases From Black-Box to Clinical Insight: A Multi-Stage Explainable Framework for Speech-Based Cognitive Impairment Detection Clinician-Level Agreement Without Clinical Caution: LLM Evaluator Limits in Medical AI Benchmarking RL Post-Training Builds Compositional Reasoning Strategies Reflection AI Deceptive Grounding: Entity Attribution Failure in Clinical Retrieval-Augmented Generation Meta-Learning for Compositionality UTS at ELOQUENT 2026 Voight-Kampff: structural shifts in AI writing bypass state-of-the-art detectors Concrete Problems in AI Safety Long-context Reasoning Benchmarks Cognitive Episodes in LLM Reasoning Traces Enable Interpretable Human Item Difficulty Prediction DAIS: Dependency-Aware Intermediate QA Supervision for Complex Reasoning

Recent events (1)

6arXiv · cs.CL·Jun 16, 2026·source ↗

Hop-count taxonomy predicts LLM failure on clinical EHR question answering across architectures

Researchers introduce a 'hop-count' taxonomy — the number of distinct inferential steps required to answer a clinical EHR question — as a principled predictor of LLM failure, finding monotone accuracy decline with reasoning depth across Claude Sonnet, GPT-4o, and GPT-5. The pattern holds across two providers and two OpenAI generations, with odds ratios per hop of 0.58–0.80, and is not explained by EHR context truncation. Extended thinking (chain-of-thought) did not significantly flatten the accuracy-depth curve, though token usage scaled with hop count. The findings ground transformer compositionality limits in a clinically consequential domain and suggest hop count as a deployment risk-stratification tool.

Evaluation and Benchmarking AI Safety Research Compositional Reasoning Depth Predicts Clinical AI Failure Claude Sonnet MedAlign +4 more