This is where we are. We’re about to go down a path likely to kill literally everyone, and the responsible one is saying maybe we can ‘see a path to’ a slight moderation.
I notice that I am confused. Mankind will either see an international treaty controlling the AIs or watch as American companies and Chinese ones race towards the ASI. How plausible is it that Dario has yet to obtain evidence which would cause China to stop? How could Dario find such evidence?
The AI-2027 scenario had Agent-2 who was mostly aligned, Agent-3 became misaligned but not adversarially so and Agent-4 became adversarially misaligned. While this would cause evidence of misalignment to be unobtainable without Agent-3 since there is nothing to obtain, we might hope that Google is forced to audit Gemini N and understand what caused sociopathic vibes, as happened in the most recent modification to AI-2027.
I notice that I am confused. Mankind will either see an international treaty controlling the AIs or watch as American companies and Chinese ones race towards the ASI. How plausible is it that Dario has yet to obtain evidence which would cause China to stop? How could Dario find such evidence?
The AI-2027 scenario had Agent-2 who was mostly aligned, Agent-3 became misaligned but not adversarially so and Agent-4 became adversarially misaligned. While this would cause evidence of misalignment to be unobtainable without Agent-3 since there is nothing to obtain, we might hope that Google is forced to audit Gemini N and understand what caused sociopathic vibes, as happened in the most recent modification to AI-2027.