I'm reposting this because (unlike most of my blog posts these days) it's actually on-topic for mathstodon.xyz. It's a detailed report on what happened when I tried to get Anthropic's LLM "Claude" to solve a simple graph theory problem.

I've been impressed lately with Claude's performance on other tasks, but unfortunately not this time. https://blog.plover.com/tech/gpt/graph-theory.html