On the Limits of Innate Planning in Large Language Models

Charles Schepanowski, Charles Ling

Large language models (LLMs) achieve impressive results on many benchmarks, yet their capacity for planning and stateful reasoning remains unclear. We study these abilities directl…