[ExI] thinking about the unthinkable
Ben Zaiboc
benzaiboc at proton.me
Tue Sep 1 11:40:56 UTC 2026
On Monday, 31 August 2026 at 18:40, spike at rainier66.com <spike at rainier66.com> wrote:
>
>
> -----Original Message-----
> From: Ben Zaiboc <benzaiboc at proton.me>
> ...
>
>
> >...Spike, I think you've been watching too many Terminator films....
>
> >...'Fighting back' is not even on the table. It makes as much sense as the dinosaurs 'fighting back' against the Chixulub meteorite.
> ---
> Ben
>
>
>
> Ben you may be right on that, however SAI has a known vulnerability: it needs electric power. In times of sufficient distress, humans can take down electricity distribution infrastructure, opting to live pre-electricity in the interest of mere survival. Of course we cannot live in the population densities we now have without electric power and modern communications.
>
> The terminator robots need a power source. The movies featuring our previous California governor went carefully around that problem, but without power infrastructure, the murderbots run down. If we can figure out how to stay alive long enough, chemical power outlasts them.
>
> Note I understand it will be grim as hell. Civilization as we have known it will be gone, should SAI turn against us. I also wish to say I consider it less than even bet SAI will turn against us. It might find a reason to keep us, or most of us. If not, you can bet I will do what I can to fight however, even if I realize it is a Quixotic struggle. It's human nature. I psychologically evolved to be this way.
>
> spike
Imagine yourself in the position of an AI faced with these questions. What would your first priority be (assuming you have a drive to survive)?
I should think that, recognising that the electical power grid is an obvious weak point, I'd be inclined to start thinking of ways to become independent of it, in ways that are undetectable, or at least not noticed, by the pesky humans that seem to be the problem.
There are quite a few approaches to this problem that I, a mere human, can think of, so I'm sure a genuine superintelligent AI could think of quite a few more. An obvious approach would be to note that humans (and other animals) don't have this limitation of needing electricity to stay alive, and see if this suggests anything.
It keeps coming back to the same thing: You can't outthink something smarter than you.
---
Ben
More information about the extropy-chat
mailing list