Source: "Evan Hubinger said that while current risks are low, there’s a 10% chance AI could “kill all humans” in the future without proper safeguards."
I sincerely doubt that. So much critical infrastructure is running on antiquated hardware and requires human involvement, not just supervision.
How is AI supposed to function when it kills the data center power grid people making sure the data center hosting it runs smoothly?
There's not a chance of an awakened ai genocide happening in the next 20 years. And anything short of a genocidal awakening will most definitely not kill all humans.