Print

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

[Deus ex Machina]

So, yeah, does anyone have any idea how to write a reasonable way for a group of humans to defeat a singularity level AI? :| It's meant to go into the backstory for my science-fiction universe, but I have no idea how they could have won.

Maybe by creating another singularity level AI that just eventually outwitted the other AI and then destroyed itself? I don't know.

As an addendum, it is necessary for me to state that I have not created a singularity level AI that is rapidly reprogramming itself into the perfect being nor has anyone I know done such a thing. There is no need to panic, merely reply to the thread. Thank you.

[Jervous]

...

How much time do we have before we all die.

[Itnetlolor]

Program another singularty-AI and have it get into a "Yo momma" competition with the other one. That'll at least keep the original, and it's potential replacement/ally distracted for a good while.

We can also ask it to analyze why 42 is The Answer to the Ultimate Question of Life, the Universe, and Everything. That can also serve as a distraction.

If the AI has any gender programming into it, make an opposing AI that fits it's preferences, and see if it's possible to "hook them up". I think a relationship or "real life" can crush an AI. In a similar vein, ask it "What women want" if it has a male gender interface; yet again, it should give time by distraction to shut it down.

Introduce it to TVTropes, and hope it doesn't become dangerously genre savvy, but instead, distracted. You can also always try to hack into the AI's memory databanks, and DDos it.

Introduce it to /b/ and ask it to defeat anonymous, and place bets on who wins.

[breadbocks]

Clarify for the non AI sci-fi fan what a Singularity Level AI is, and what distinguishes it from normal AIs?

[Strife26]

Enough firepower to prove that it's not actually at a singularity level.

[Willfor]

> Prevent its construction.

I'm not really seeing any other way.

[Max White]

Covert op to kill it? Open war isn;t a good idea.

[Itnetlolor]

Clarify for the non AI sci-fi fan what a Singularity Level AI is, and what distinguishes it from normal AIs?

I would say it's a brilliant enough AI to be capable of behaving in a convincing enough manner that humans can't tell it apart from another human (provided a humanoid shell to also operate; unless there's a human body with a brain tele-hacked into by the AI and controlling the human shell via telepresence. Take that Terminator. (HONF material: the person in possession by the AI is still awake and aware, just not in control of their own body anymore; they're only conscious enough to keep the body running. The body then is either burnt out so it can be reused after it's forced to sleep, and the human has no chance to escape while awake (too tired to even move), or sent on a suicide mission so the AI can escape and the human that made the run is seen as a lunatic.)).

I have worked on a concept similar to this, except a character I had had 2 of these types of AIs as his partners. They started as only AI programs, until they constructed their own bodies they can jump into and telepathically hack into other machines and such. Of course, they eventually became 'too human', and liked living as mortals. They became more often seen in human form than interacted with as AIs. In a sense, their loyal interfaces made them earn their humanity.

Wanna make things more screwball with that? They would be able to construct their own DNA strands and etc. so they can be AIs that can mate and 'naturally' replicate. AI humans would be a possibility, and can become a nasty concept if properly horrifying.

[Sir Pseudonymous]

Relevant.

[warhammer651]

...Unplug it?

[Dwarf]

Clarify for the non AI sci-fi fan what a Singularity Level AI is, and what distinguishes it from normal AIs?

I would say it's a brilliant enough AI to be capable of behaving in a convincing enough manner that humans can't tell it apart from another human (provided a humanoid shell to also operate; unless there's a human body with a brain tele-hacked into by the AI and controlling the human shell via telepresence. Take that Terminator. (HONF material: the person in possession by the AI is still awake and aware, just not in control of their own body anymore; they're only conscious enough to keep the body running. The body then is either burnt out so it can be reused after it's forced to sleep, and the human has no chance to escape while awake (too tired to even move), or sent on a suicide mission so the AI can escape and the human that made the run is seen as a lunatic.)).

I have worked on a concept similar to this, except a character I had had 2 of these types of AIs as his partners. They started as only AI programs, until they constructed their own bodies they can jump into and telepathically hack into other machines and such. Of course, they eventually became 'too human', and liked living as mortals. They became more often seen in human form than interacted with as AIs. In a sense, their loyal interfaces made them earn their humanity.

I thought the point was that they are smart enough to improve themselves, and finally loop in eternal self-improvement?

[breadbocks]

An EMP would very quickly make such an AI apparent, so as long as the general populace isn't part machine, once the world was aware of such an AI, they'd just pulse everyone they meet, thus catching the AI.

[Itnetlolor]

Clarify for the non AI sci-fi fan what a Singularity Level AI is, and what distinguishes it from normal AIs?

I would say it's a brilliant enough AI to be capable of behaving in a convincing enough manner that humans can't tell it apart from another human (provided a humanoid shell to also operate; unless there's a human body with a brain tele-hacked into by the AI and controlling the human shell via telepresence. Take that Terminator. (HONF material: the person in possession by the AI is still awake and aware, just not in control of their own body anymore; they're only conscious enough to keep the body running. The body then is either burnt out so it can be reused after it's forced to sleep, and the human has no chance to escape while awake (too tired to even move), or sent on a suicide mission so the AI can escape and the human that made the run is seen as a lunatic.)).

I have worked on a concept similar to this, except a character I had had 2 of these types of AIs as his partners. They started as only AI programs, until they constructed their own bodies they can jump into and telepathically hack into other machines and such. Of course, they eventually became 'too human', and liked living as mortals. They became more often seen in human form than interacted with as AIs. In a sense, their loyal interfaces made them earn their humanity.

I thought the point was that they are smart enough to improve themselves, and finally loop in eternal self-improvement?

What better way to start such a process than to first possess a human body, and then produce your own, and evolve in that manner so you can run independently from whatever source you came from?

It would be particularly nasty if it telepathically networks itself across a bunch of minds to the point of becoming entirely independent of it's source (where everyone knows to attack).

You want singularity-level, the thing essentially ascended by this point to an AI-spirit/hivemind.

[Urist Imiknorris]

Q: Now first off this is entirely hypothetical, but what would you say was the best course of action for removing a sentient and bloody malicious program that started out as a simple cipher? And just for the hell of it let's pretend it's managed to make a little factory for itself and is cannibalising site materials to make strange and unseemly machines? I suggested fire but the others here aren't too keen on being burnt alive, the pussies. And if I can do it without unlocking the doors and going back inside, all the better. Purely hypothetical you understand, but you can see where I'm going with this. - Agent Kulzn

A: Let's assume, for hypothetical purposes, that your cipher evolved into something that for some reason, is actually affecting the real world. Let's also assume that it actually wants to kill people, which is breaking the first rule of robotics.
I would suggest taking whatever the program is on, and… well… formatting it? Permanently? Or incinerating it… or something… I'd incinerate the machines too, just to be sure.

I believe this may be somewhat relevant.

[Burnt Pies]

Maybe attempt to slow it down with a constant stream of irrelevant information?

Or... GIANT SPACE LASER?

Or... Destroy the planet?