Everyone seems to use LLMs the wrong way. Instead of using them to do useful work (for example: static code analysis with automatic fixing of found bugs, writing high-quality documentation), they create useless dumb chatbots. Waiting for someone to integrate a chatbot into a calculator.