analiza aiqbench
The paper introduces a benchmark for testing whether LLMs can reason logically about chaotic dynamical systems, where the underlying math is deterministic but often mistaken for randomness. Frontier models score 91-94% on per-item questions but 0% on compositional items, and dialogue-level accuracy ranges from 53% to 76%. For AI readers, the useful signal is the gap between local correctness and consistent multi-step reasoning, along with the released open-source evaluation pipeline.
ChaosBench-Logic: A Benchmark for Logical and Symbolic Reasoning on Chaotic Dynamical Systems
arXiv:2601.01982v2 Announce Type: replace Abstract: Large language models (LLMs) excel at natural language tasks but remain brittle in domains requiring precise logical and symbolic reasoning. Chaotic dynamical systems provide an especially demanding test because chaos is deterministic yet often misinterpreted as randomness or complexity. We introduce ChaosBench-Logic, a benchmark that evaluates LLM reasoning across 30 diverse dynamical systems using a unified first-order logic (FOL) ontology. Each system is annotated with truth assignments for 11 semantic predicates, and 621 questions are generated across seven reasoning categories, including multi-hop implications, cross-system analogies, counterfactual reasoning, bias probes, and multi-turn dialogues. We define metrics for logical accuracy, implication consistency, dialogue coherence, and contradiction, and we release an open-source evaluation pipeline. Initial experiments show that frontier LLMs such as GPT-4, Claude 3.5 Sonnet, Gemini 2.5 Flash, and the open-source LLaMA-3 70B achieve 91-94% per-item accuracy, yet still score 0% on compositional items and exhibit fragile global coherence. Dialogue-level accuracy ranges from 53.1% (GPT-4 CoT) to 75.5% (LLaMA-3 zero-shot). ChaosBench-Logic provides a rigorous testbed for diagnosing such failures and a foundation for developing neuro-symbolic approaches that improve scientific reasoning in LLMs.
Abstract:Large language models (LLMs) excel at natural language tasks but remain brittle in domains requiring precise logical and symbolic reasoning. Chaotic dynamical systems provide an especially demanding test because chaos is deterministic yet often misinterpreted as randomness or complexity. We introduce ChaosBench-Logic, a benchmark that evaluates LLM reasoning across 30 diverse dynamical systems using a unified first-order logic (FOL) ontology. Each system is annotated with truth assignments for 11 semantic predicates, and 621 questions are generated across seven reasoning categories, including multi-hop implications, cross-system analogies, counterfactual reasoning, bias probes, and multi-turn dialogues. We define metrics for logical accuracy, implication consistency, dialogue coherence, and contradiction, and we release an open-source evaluation pipeline. Initial experiments show that frontier LLMs such as GPT-4, Claude 3.5 Sonnet, Gemini 2.5 Flash, and the open-source LLaMA-3 70B achieve 91-94% per-item accuracy, yet still score 0% on compositional items and exhibit fragile global coherence. Dialogue-level accuracy ranges from 53.1% (GPT-4 CoT) to 75.5% (LLaMA-3 zero-shot). ChaosBench-Logic provides a rigorous testbed for diagnosing such failures and a foundation for developing neuro-symbolic approaches that improve scientific reasoning in LLMs.
7 pages, 0 figures, Accepted to AAAI-26 Bridge Program: Logical and Symbolic Reasoning in Language Models (camera-ready). v2: corrected bibliographic errors in the references
Artificial Intelligence (cs.AI); Information Theory (cs.IT)
arXiv:2601.01982 [cs.AI]
(or arXiv:2601.01982v2 [cs.AI] for this version)
Text integral sub licența deschisă a publicației · 218 cuvinte
Subiecte conexe
SymbolLKG: Towards Verifiable Logical Reasoning via Logical Knowledge Graph and Symbolic Solvers
arXiv · 28 aug. 2026
ChaosBench-Logic v2: Evaluating LLM Logical Reasoning over Dynamical Systems at Scale
arXiv · 17 oct. 2026
Pushing the Boundaries of Natural Reasoning: Interleaved Bonus from Formal-Logic Verification
arXiv · 16 iun. 2026
Informe de riesgos de la IA de frontera (febrero–marzo de 2026)
METR · 19 mai 2026