If you've been using the latest frontier large language models (LLMs), you've almost certainly noticed by now that the new models are getting worse at conversation. They're more robotic, speak in jargon, spit out verbose text, and do things you didn't ask for. How did this happen? Well, I'm not the