AutoMedBench evaluates AI agents on long-horizon medical workflows across five stages and finds validation and submission as dominant failure points based on thousands of runs.
DiscoveryWorld: A virtual environment for developing and evaluating automated scientific discovery agents
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2verdicts
CONDITIONAL 2representative citing papers
LLMs in multi-turn ideation reliably increase structural complexity while violating original constraints despite preserved declarative recall, with KBV rates ranging 8-99% across models.
citing papers explorer
-
AutoMedBench: Towards Medical AutoResearch with Agentic AI Models
AutoMedBench evaluates AI agents on long-horizon medical workflows across five stages and finds validation and submission as dominant failure points based on thousands of runs.
-
Models Recall What They Violate: Constraint Adherence in Multi-Turn LLM Ideation
LLMs in multi-turn ideation reliably increase structural complexity while violating original constraints despite preserved declarative recall, with KBV rates ranging 8-99% across models.