LLM solutions claimed to be significantly better than task authors’ solutions

A user shared a list of tasks for which they say LLMs—mostly GPT-6 Pro—found solutions significantly better and non-trivially different from the authors’ solutions. They said the list was still being updated and that they would mark solutions they found particularly interesting.