Since you are checking your thoughts, here are a couple of posts that argue the contrary to your 1.2:
https://www.lesswrong.com/posts/MmmPyJicaaJRk4Eg2/the-limit-of-language-models
https://www.lesswrong.com/posts/mEbYbekPXoWvrHraW/human-level-ai-can-plausibly-take-over-the-world
Since you are checking your thoughts, here are a couple of posts that argue the contrary to your 1.2:
https://www.lesswrong.com/posts/MmmPyJicaaJRk4Eg2/the-limit-of-language-models
https://www.lesswrong.com/posts/mEbYbekPXoWvrHraW/human-level-ai-can-plausibly-take-over-the-world