Finding 4247Emerging EvidenceValidation V0
Commercial LLMs consistently outperform open-source models on FinLBench tasks, but all models struggle with hallucinationfabricating information, especially when confronted with trap questions, exposing a critical and surprising weakness in current AI systems.
78%Confidence
1Evidence objects
v1Version
DraftStatus
Evidence trail
Supporting78% linkage confidence
Commercial LLMs consistently outperform open-source models on FinLBench tasks, but all models struggle with hallucinationfabricating information, especially when confronted with trap questions, exposing a critical and surprising weakness in current AI systems.
key_findings bullet 2 · key_findings
Inspect source: FinLBench: A Benchmark for Evaluating Large Language Models on Long-Text Financial Documents →This Finding was extracted from the configured corpus. It is versioned, traceable, and may evolve through editorial review or new corpus evidence.