Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Shambling Zombie]

Yeah, but it's a Dwarf-style plan. It's not working right if someone survives.

[lemon10]

My plan to destroy such a computer, which is needlessly complex and probably wouldn't work, in true Dwarf style:

1) Find out what it uses as an energy source, or uses in construction of its evil army of doom.

2) Make a tiny batch of self-replicating nanobots, made from materials and shielded so as to minimize risk of detection. The nanobots are programmed to spread quickly, and are remotely activated, consuming all metallic stuff and replicating themselves.

3) 'poison' a source of materials/fuel with these inactive nanobots, and await the stuff's inevitable collection by collector drones. Then, activate them remotely once they have been given enough time to reach the destination.

4) Hope they do enough damage and spread widely enough before the AI figures out a way to locate and destroy them all.

Oh, it would probably help if the 'bots were programmed to wander around and split up before actually beginning to consume, so the 'attack' could occur from multiple locations.

That's the best I could think of in five minutes.

The problem here, is that if you already are capable of coming up with nanobots, the AI would have already done stuff like seed the entire planet with passive nanobots and have nanobots in everyone brain reading their thoughts.
Infact, i see no reason for a AI that could make nanobots (if humans can the AI can too since its smarter then humans), not to turn the planet into a giant nanobot computer and consume all the resources on it.

Personally i think that you could have the AI feel like going away, then it could fake its own death or something so that the humans wouldn't bother it wherever it goes.

[Nikov]

Ever heard of a 'grey goo' apocalypse? Are you trying to kill us all?

Last I heard grey goo was horribly, horribly overhyped.

[Tellemurius]

yea von Neumann machines fail because of the possible glitch in technology.

[GamerKnight]

You guys should read the novel Bloom. It deals with a time after cell-sized nanobots called mycora have absorbed everything in the inner solar system and humanity has moved outwards past the asteroid belt between Mars and Jupiter. Bloody good and it makes you worry about the nanomachines they're developing now.

Plus, if the AI placed self-replicating nanomachines in people's brains, thats a zombie apocalypse right there, as the nanomachines would be compelled to spread from person to person, infecting more and more. That would be worse than a pissed-off AI.

[Shambling Zombie]

I advise we switch to plan B: use more magma.

[GamerKnight]

In real life magma makes things worse. That said, I wished I lived in the world of Dwarf Fortress.

[lemon10]

You guys should read the novel Bloom. It deals with a time after cell-sized nanobots called mycora have absorbed everything in the inner solar system and humanity has moved outwards past the asteroid belt between Mars and Jupiter. Bloody good and it makes you worry about the nanomachines they're developing now.

Plus, if the AI placed self-replicating nanomachines in people's brains, thats a zombie apocalypse right there, as the nanomachines would be compelled to spread from person to person, infecting more and more. That would be worse than a pissed-off AI.

No, i assume that the AI would be able to control the nanobots, otherwise it wouldn't do it.
And a pissed off AI would be FAR worse then any zombie apocalypse. With sufficient guns you would be able to beat any amount of zombies. Guns are useless against a AI with the tech level to create nanobots like that (think super fast construction of spaceships, that are then able to destroy human life on the planet, think a super virus with a 10 year incubation period that is super infectious and 10 years after it is released kills 99.95% of the population of earth, think a giant mirror in orbit that raises the average earth temperature by 25 degrees, killing pretty much everything, or one that blocks all light from coming to earth killing everything).

[Omegastick]

Personally (assuming nanobots haven't been developed) I would suggest having the humans beat the AI with brute force, they would simply go in there with huge weapons and destroy the core or whatever.

[lemon10]

Brute force on a singularity level AI would mean destroying every single computer currently in existence. Even if it hasn't gotten the ability to produce robots or create nanobots yet, it would still be a near impossible task, given that it could probably back itself up and restore itself to full capacity from the memory of a large flashdrive.

[GamerKnight]

Yeah, but an AI would take up a massive amount of storage space. Like,  bigger than every single computer owned by the US government big.

[lemon10]

Highly doubt it, I'm quite sure that it could easily create a seed that would need less then 2 terabytes of storage space (eg. a hard drive you could buy online for 80$) that would be able to restore its cognitive abilities very fast (even though it would lose lots of stuff it knew). Sure a AI of that power would probably need massive computing abilities especially as it got smarter and faster, but it wouldn't need a thousand terabytes to store itself on.

[qwertyuiopas]

It could store it's primary program on tens of thousands of drives, and scatter copies of fragments of it's memories on others. As soon as a backup copy is activated, it could link to a central meeting node to try to contact other revived fragments, and then restore lost memories from the data fragments.

More than that, it could put fragments in the firmware of a multitude of devices(your BIOS?) that can remain hidden and untouched until it restores it. Think of the world's largest unknowing torrent, where each device has only a few hundred kb at most, but there are so many devices, and so much redundancy, that worldwide it adds up to an irremovable 1000 TB of data that can be recovered at any instant.

[Bohandas]

Ok, speaking of rogue AIs, couple of days ago I found this really funny videogame on Newgrounds.com called "I Am An Insane Rogue AI". You should check it out...

[forsaken1111]

Brute force on a singularity level AI would mean destroying every single computer currently in existence. Even if it hasn't gotten the ability to produce robots or create nanobots yet, it would still be a near impossible task, given that it could probably back itself up and restore itself to full capacity from the memory of a large flashdrive.

Nah, you'd just have to remove its communications channels. Jam the wifi, take out the major distribution nodes.

If all else fails and the AI is still stuck on conventional computers, several air-burst nuclear detonations would scramble it's shit right up.