Fascinating blog about improving performance of LLMs on tasks which they are terrible at, compared to humans.See the picture below, easy for us to put the blocks in/around the blue blocks, not so for an LLM.https://jeremyberman.substack.com/p/ho