« BackLLM "reasoning" continues to be deeply flawedgarymarcus.substack.comSubmitted by petethomas 16 hours ago