[wagwang]

I would expect people who doom about AI existential risks to not train cutting edge models and give them agentic freedom.