You said:

> I haven't seen one convincing model of AI existential risk. Can anyone offer one?

I apologize for answering. This is the only episode of this podcast I have ever watched, and I have watched it four times now. The person being interviewed knows what they're talking about. You would learn from them.