From my own personal PoV there is still too much anthropomorphic language being used here, but Prof Hinton did at least mention the most immediate risk is bad actors using AI to do things we don't want.

I find the "super intelligence" neologism concerning. It feels like an attempt to bypass the complete absence of AGI, and somehow re-label the existing LLM. And always "just around the corner" comes into things.