LLM are tunable, so we'll get some kind of kindergartened version of AI which will be still useful, yet too polite.