Generative AI models with “reasoning” may not actually excel at solving certain types of problems when compared with conventional LLMs, according to a paper from researchers at Apple.

The researchers examined what they call "large reasoning models" (LRMs), which attempt to simulate a logical reasoning ...