I don't see how LLMs are "very different" from hypothesized AI. 

Personally my p(doom) was already high and increased modestly but not fundamentally after recent advances.