Oliver King-Smith's latest article has been published on Finextra: Challenging the Notion That LLMs Can't Reason: A Case Study with Einstein's Puzzle
When Apple declared that LLMs can't reason, they forgot one crucial detail: a hammer isn't meant to turn screws. In our groundbreaking study of Einstein's classic logic puzzle, we discovered something fascinating. While language models initially stumbled with pure reasoning - making amusing claims like "Plumbers don't drive Porsches" - they excelled at an unexpected task.