Large Language Models and Mathematical Reasoning Failures

Johan Boye, Birger Moell

This paper investigates the mathematical reasoning capabilities of large language models (LLMs) using 50 newly constructed high-school-level word problems. Unlike prior studies tha…