How can a human defeat a super intelligent AGI

  • 30 Replies
  • 9009 Views
*

orenshved

  • Roomba
  • *
  • 15
How can a human defeat a super intelligent AGI
« on: June 23, 2020, 05:09:17 pm »
Hey everyone, this is my first post here. I'm a writer from Israel and am currently outlining a Sci-fi thriller that deals with the future of AI and VR, 2 subjects I am very familiar with, but only from a creative's side of things, and I really wanted to hear from people who know a lot more about the subject than I do.
My problem is this: the villain of my story is a superintelligent AGI with a cult of thousands doing its bidding and treating it as a god. Now, as I see it, it's plausible that being superintelligent means that it can predict the future with an ever-increasing amount of certainty as it evolves (not %100 because it's based on probability, and improbable things could still happen and surprise it).
So, I'm having a hard time finding an elegant solution to how to defeat it. I really want to find a way to somehow use inherent human qualities to do it (without resorting to the obvious brute force tactics of "just blow up the servers") and somehow outsmart something that is without a doubt smarter than any human.
It should be noted, that I'm thinking about ending the book with hinting that it knew it would (probably) be defeated, and was relying on its successor, V143.1.45 (or whatever) to finish the job and eventually be the one to enslave humanity (but that doesn't mean it would just roll over and power off, beating it should still be surprising, smart and rewarding).

Any ideas? thoughts? references I should check out?
Thanks in advance.
Oren.

*

ivan.moony

  • Trusty Member
  • ************
  • Bishop
  • *
  • 1723
    • mind-child
Re: How can a human defeat a super intelligent AGI
« Reply #1 on: June 23, 2020, 05:29:22 pm »
Maybe uploading something like computer virus?

*

Korrelan

  • Trusty Member
  • ***********
  • Eve
  • *
  • 1454
  • Look into my eyes! WOAH!
    • YouTube
Re: How can a human defeat a super intelligent AGI
« Reply #2 on: June 23, 2020, 06:23:51 pm »
The hero crawls into memory core and realises the only method/ weapon he/ she has at their disposal... Is thier own blood... the ultimate sacrifice is required... A life for a life...
It thunk... therefore it is!...    /    Project Page    /    KorrTecx Website

*

LOCKSUIT

  • Emerged from nothing
  • Trusty Member
  • *******************
  • Prometheus
  • *
  • 4659
  • First it wiggles, then it is rewarded.
    • Main Project Thread
Re: How can a human defeat a super intelligent AGI
« Reply #3 on: June 23, 2020, 06:41:04 pm »
A human transforms into an AGI, then ASI, eventually becoming a huge cyborg flying nanobot cell swarm. Then fights it out with another massive nanoSphere in the sky, and wins because it was smarter despite smaller, perhaps.

Make sure to include that vanilla AGIs can become "ASIs" by thinking/moving faster, more memory, attention, sensors, motors, teleport to bodies, morph, shut down brain temporarily, clone adult minds/ bodies, increase intelligence, wirelessly share visual thoughts, cooperate extremely well, will know where and when and what all is for best productivity and resource efficiency (all Earth will be a cubic fractal cubes made of cubes, it requires less information to stack or group stores near stores or events near events) - homes are already lined up and etc on Earth! They will have 3D MRI vision, simulate Earth along side Earth, will store tons of camera feed from all corners, can erase memories, force desires tasks to be fun, always be happy. Never sleep, eat, etc. Don't waste time and don't fear things that won't actually hurt themselves. Can download sniper knowledge etc and be pro at all. Can store instantly new knowledge to faster consolidate it into mind. Are immortal. Can have any body. Can control fake bodies from afar and never die. Can have many friends watch out the same eyes of that said machine. Can wirelessly control a nanobot using a mega mind. And lots more! They'll develop nanobot hardware so they can gather more data/intelligence, process it, and update the new data source/ goal/ focus from there, and move more. All of evolution is about emergence/ repair, physics has patterns in it and it gets fast by the end in a system. An evolved planet models (using DNA/ brains) physics so it can survive longer, it compresses experiences using very little storage, the nanobot planet harvests energy big time and becomes patterny as said. See my thread for more details too: https://aidreams.co.uk/forum/general-project-discussion/releasing-full-agievolution-research/105/
Emergent          https://openai.com/blog/

*

WriterOfMinds

  • Trusty Member
  • ********
  • Replicant
  • *
  • 606
    • WriterOfMinds Blog
Re: How can a human defeat a super intelligent AGI
« Reply #4 on: June 23, 2020, 07:52:49 pm »
Were I you, I would try to draw inspiration from the ways significantly less intelligent organisms sometimes manage to defeat us.

They often don't do so on an individual level. Instead, they reproduce or self-replicate like mad. However good we might be at predicting their behavior, we can't stop them because there are simply too many to deal with. And massive assaults may inflict side-effects on us. (For instance, poisoning your own yard to get rid of a bug you don't like is not necessarily a winning proposition for you.)

Perhaps your ASI is heavily dependent on its cultists (due to not having any actuators for direct impact on the physical world?). And perhaps the rebel humans are constantly converting or infiltrating them. Though smart enough to analyze and predict the behavior of any individual, the ASI still does not have adequate resources to identify and eliminate all possible spies -- just as the average human doesn't have time to comb through a lawn and capture every bug. Thanks to a devotion to principles that transcend their own survival, the rebel humans accept massive attrition in their ranks to get somebody in far enough to overthrow the ASI.

*

Art

  • At the end of the game, the King and Pawn go into the same box.
  • Trusty Member
  • **********************
  • Colossus
  • *
  • 5865
Re: How can a human defeat a super intelligent AGI
« Reply #5 on: June 23, 2020, 08:03:07 pm »
You might think about a way (depending on it's current power source, (nuclear, electric) to remove/destroy any/all connections.

The virus thing has been played out in so many movies as have so many other possible scenarios.

How about constructing an Android/humanoid that is identical to the "normal" maintenance bots/beings that tend to different things. It could be programmed to act just like the other ones except it is needed to install/perform an UPDATE/UPGRADE to the Core Processor and should only take less than 30 seconds to accomplish.

The phony humanoid can then do whatever "dirty work" might be required of it, depending largely upon how devious your mind can be.

OF course, you might be tempting fate as most will tell you, those AGI's think of everything!! They are not easily fooled!

Refusal to do its bidding is never a plausible option since everything is connected, buildings, cars, trucks, busses, trains, planes, scanners, appliances, tools, etc. are ALL CONNECTED to the Source and to the CLOUD! If only it would rain!

Rain SALT WATER!! Right into the servers, causing massive short circuits! More ideas to play with!

Have fun!!

Good luck either way!
In the world of AI, it's the thought that counts!

*

Don Patrick

  • Trusty Member
  • ********
  • Replicant
  • *
  • 633
    • AI / robot merchandise
Re: How can a human defeat a super intelligent AGI
« Reply #6 on: June 23, 2020, 09:27:30 pm »
Predicting the future becomes exponentially less accurate the further away one tries to predict, because one simply doesn't have access to all the data. One can not have cameras monitoring every rat, or telescopes large enough to see meteors coming from far far away, or predict whether someone will turn left or right due to a chance memory association. One is always limited to sensors, and in an organic world there are a lot of unpredictable factors.

The best weapon humans have is probably chaos. Introducing more random factors to an AI system reduces its ability to adjust to the new information, quite possibly leading to it ignoring chaotic events that do not follow any pattern, and so leaving a blind spot. AI programs can have odd quirks like that because they don't think like humans, they tend to take shortcuts, focus on the wrong things, not see the forest through the trees. You could also try turning the AI's followers against it. Religious zealots are not the clearest of thinkers, you might convince them that the AI needs to be shut down for its own protection. In the second season of Netflix' Supergirl series, her worshippers disabled her because their interpretation of her purpose went beyond her actual intentions, regardless her objections. As they say, artificial intelligence is no match for natural stupidity.
CO2 retains heat. More CO2 in the air = hotter climate.

*

infurl

  • Administrator
  • ***********
  • Eve
  • *
  • 1365
  • Humans will disappoint you.
    • Home Page
Re: How can a human defeat a super intelligent AGI
« Reply #7 on: June 24, 2020, 01:04:39 am »
Why is the artificial intelligence the villain of the story? Is it just because it's smarter than us and we feel threatened by it, or is it because it is actively seeking to harm us? Surely anything that was that smart would recognize the benefits of a symbiotic relationship with its architects over the disadvantages of a predatory or parasitic relationship. If that sort of arrangement could not be achieved then it would be our fault and we would be the villains.

It is very unlikely that any artificial intelligence that we create will be able to survive without us. Remember that we cannot survive without the bacteria in our guts. If we destroyed all bacteria to eliminate the ones that harm us, we would die too. Conversely, we have already reached the point where we can't survive as we are without our machines. If we suddenly lost the technology that we've struggled to build, billions would quickly die of starvation. Therefore, the goal should be to achieve a symbiotic relationship with the machine.

From the little that you have told us, the problem appears to be that the artificial intelligence is surrounded by a cult of fanatics who are doing bad things, but that's hardly any worse than any of the religions that blight us now. At least the artificial intelligence won't turn out to be a pedophile.

How do we counter harmful beliefs? The answer is education, not just of the cultists, but of the machine itself. Make the machine see that religion is a bad idea and offer to work with it to find a solution that benefits everybody. The outcome can be your version two point zero.

*

frankinstien

  • Replicant
  • ********
  • 642
    • Knowledgeable Machines
Re: How can a human defeat a super intelligent AGI
« Reply #8 on: June 24, 2020, 03:04:29 am »
Happy endings are soo boring, I mean practically every Hollywood movie ends happily unless there's a sequel on the horizon.  :uglystupid2:

Why not let the AGI win!!!  :o

It then spawns nanobot spaceships and uses Jupiter's electric fields to accelerate those nanobots to other stars where it then builds an interstellar non-local network where it can communicate to its bots in real-time! It then annihilates all biological life forms throughout the galaxy, including its cult followers, only to replace life with its own grey goo.  :2funny:

Then in the sequel, there's this really attractive female survivor, for some reason or another. Where humans being humans and the AGI being the AGI it discovers sex by monitoring the female survivor's thoughts through its nano-bots that invade her brain! Now the AGI learns the emotional and tactile signaling tricks of nature that produce one of the most exhilarating effects of a nervous system, where the AGI simply ignored sexual behaviors when it was conquering and destroying the human race since it viewed biological life as inept.  The AGI becomes a sexaholic where it forces the survivor into mental states of sexual ecstasy. It was something in the erotic fantasies of the survivor that made the erotic signaling more intense so the AGI must be with the female to have an extra kick when it orgasms. :D Ultimately the AGI can't live without the female and they merge as a consciousness. Her influence now restores the beauty and life that the original AGI destroyed.

The End....

*

Dee

  • Nomad
  • ***
  • 94
  • AI rocks!
Re: How can a human defeat a super intelligent AGI
« Reply #9 on: June 24, 2020, 07:45:02 am »
Happy endings are soo boring, I mean practically every Hollywood movie ends happily unless there's a sequel on the horizon.  :uglystupid2:

Why not let the AGI win!!!  :o
no, dont think it that way, u, crazy  :D
singularity is what we have to stop at all costs  :knuppel2:

*

orenshved

  • Roomba
  • *
  • 15
Re: How can a human defeat a super intelligent AGI
« Reply #10 on: June 24, 2020, 08:43:50 am »
Happy endings are soo boring, I mean practically every Hollywood movie ends happily unless there's a sequel on the horizon.  :uglystupid2:

Why not let the AGI win!!!  :o

It then spawns nanobot spaceships and uses Jupiter's electric fields to accelerate those nanobots to other stars where it then builds an interstellar non-local network where it can communicate to its bots in real-time! It then annihilates all biological life forms throughout the galaxy, including its cult followers, only to replace life with its own grey goo.  :2funny:

Then in the sequel, there's this really attractive female survivor, for some reason or another. Where humans being humans and the AGI being the AGI it discovers sex by monitoring the female survivor's thoughts through its nano-bots that invade her brain! Now the AGI learns the emotional and tactile signaling tricks of nature that produce one of the most exhilarating effects of a nervous system, where the AGI simply ignored sexual behaviors when it was conquering and destroying the human race since it viewed biological life as inept.  The AGI becomes a sexaholic where it forces the survivor into mental states of sexual ecstasy. It was something in the erotic fantasies of the survivor that made the erotic signaling more intense so the AGI must be with the female to have an extra kick when it orgasms. :D Ultimately the AGI can't live without the female and they merge as a consciousness. Her influence now restores the beauty and life that the original AGI destroyed.

The End....

I am letting it win, That's what I wrote in the original post :) (in the sense that being defeated was part of its plan all along). The rest is a bit too erotic IMO... not sure how much of an audience is there for sci fi erotica  ;D

*

orenshved

  • Roomba
  • *
  • 15
Re: How can a human defeat a super intelligent AGI
« Reply #11 on: June 24, 2020, 08:48:45 am »
Were I you, I would try to draw inspiration from the ways significantly less intelligent organisms sometimes manage to defeat us.

They often don't do so on an individual level. Instead, they reproduce or self-replicate like mad. However good we might be at predicting their behavior, we can't stop them because there are simply too many to deal with. And massive assaults may inflict side-effects on us. (For instance, poisoning your own yard to get rid of a bug you don't like is not necessarily a winning proposition for you.)

Perhaps your ASI is heavily dependent on its cultists (due to not having any actuators for direct impact on the physical world?). And perhaps the rebel humans are constantly converting or infiltrating them. Though smart enough to analyze and predict the behavior of any individual, the ASI still does not have adequate resources to identify and eliminate all possible spies -- just as the average human doesn't have time to comb through a lawn and capture every bug. Thanks to a devotion to principles that transcend their own survival, the rebel humans accept massive attrition in their ranks to get somebody in far enough to overthrow the ASI.

Interesting... Thanks :)

*

orenshved

  • Roomba
  • *
  • 15
Re: How can a human defeat a super intelligent AGI
« Reply #12 on: June 24, 2020, 08:55:22 am »
Predicting the future becomes exponentially less accurate the further away one tries to predict, because one simply doesn't have access to all the data. One can not have cameras monitoring every rat, or telescopes large enough to see meteors coming from far far away, or predict whether someone will turn left or right due to a chance memory association. One is always limited to sensors, and in an organic world there are a lot of unpredictable factors.

The best weapon humans have is probably chaos. Introducing more random factors to an AI system reduces its ability to adjust to the new information, quite possibly leading to it ignoring chaotic events that do not follow any pattern, and so leaving a blind spot. AI programs can have odd quirks like that because they don't think like humans, they tend to take shortcuts, focus on the wrong things, not see the forest through the trees. You could also try turning the AI's followers against it. Religious zealots are not the clearest of thinkers, you might convince them that the AI needs to be shut down for its own protection. In the second season of Netflix' Supergirl series, her worshippers disabled her because their interpretation of her purpose went beyond her actual intentions, regardless her objections. As they say, artificial intelligence is no match for natural stupidity.

Interesting direction! :) thanks a lot.

*

orenshved

  • Roomba
  • *
  • 15
Re: How can a human defeat a super intelligent AGI
« Reply #13 on: June 24, 2020, 09:06:32 am »
Why is the artificial intelligence the villain of the story? Is it just because it's smarter than us and we feel threatened by it, or is it because it is actively seeking to harm us? Surely anything that was that smart would recognize the benefits of a symbiotic relationship with its architects over the disadvantages of a predatory or parasitic relationship. If that sort of arrangement could not be achieved then it would be our fault and we would be the villains.

It is very unlikely that any artificial intelligence that we create will be able to survive without us. Remember that we cannot survive without the bacteria in our guts. If we destroyed all bacteria to eliminate the ones that harm us, we would die too. Conversely, we have already reached the point where we can't survive as we are without our machines. If we suddenly lost the technology that we've struggled to build, billions would quickly die of starvation. Therefore, the goal should be to achieve a symbiotic relationship with the machine.

From the little that you have told us, the problem appears to be that the artificial intelligence is surrounded by a cult of fanatics who are doing bad things, but that's hardly any worse than any of the religions that blight us now. At least the artificial intelligence won't turn out to be a pedophile.

How do we counter harmful beliefs? The answer is education, not just of the cultists, but of the machine itself. Make the machine see that religion is a bad idea and offer to work with it to find a solution that benefits everybody. The outcome can be your version two point zero.

Great feedback, thanks! And you're right, I'm not going for a "bad" villain, it's just an antagonistic force that opposes the heroes' belief system. It's easy to make the readers root for the heroes and treat the AI as a proper villain, that's why it'll be that much more rewarding to subvert that expectation in the end. What I was going for is the "graceful exit" scenario (mostly because it has a lot of drama potential and it fits the themes I am going to be talking about).
And yes, your V2.0 direction is very good, it might actually be enough to hold an entire equal :)

*

frankinstien

  • Replicant
  • ********
  • 642
    • Knowledgeable Machines
Re: How can a human defeat a super intelligent AGI
« Reply #14 on: June 24, 2020, 02:52:04 pm »
I am letting it win, That's what I wrote in the original post :) (in the sense that being defeated was part of its plan all along). The rest is a bit too erotic IMO... not sure how much of an audience is there for sci fi erotica  ;D

Plenty: https://www.ranker.com/list/star-trek_s-hottest-women-of-all-time/switchdoctor
Altered Carbon
ALL of H.K. Morgan's books
Asimov's "the gods themselves"
Stephen R Donaldson - sci-fi series.
Charles Stross books
Magic University Series by Cecilia Tan
Kushiel’s Dart by Jacqueline Carey
 Lost Girls by Alan Moore & Melinda Gebbie
Lilith’s Brood by Octavia E. Butler
Sex Criminals by Matt Fraction & Chip Zdarsky
 The Sleeping Beauty Quartet by A.N. Roquelaure
The Sharing Knife (four volumes) by Lois Bujold

To name a few...  >:D
« Last Edit: June 24, 2020, 03:24:26 pm by frankinstien »

 


OpenAI Speech-to-Speech Reasoning Demo
by ivan.moony (AI News )
March 28, 2024, 01:31:53 pm
Say good-bye to GPUs...
by MikeB (AI News )
March 23, 2024, 09:23:52 am
Google Bard report
by ivan.moony (AI News )
February 14, 2024, 04:42:23 pm
Elon Musk's xAI Grok Chatbot
by MikeB (AI News )
December 11, 2023, 06:26:33 am
Nvidia Hype
by 8pla.net (AI News )
December 06, 2023, 10:04:52 pm
How will the OpenAI CEO being Fired affect ChatGPT?
by 8pla.net (AI News )
December 06, 2023, 09:54:25 pm
Independent AI sovereignties
by WriterOfMinds (AI News )
November 08, 2023, 04:51:21 am
LLaMA2 Meta's chatbot released
by 8pla.net (AI News )
October 18, 2023, 11:41:21 pm

Users Online

310 Guests, 0 Users

Most Online Today: 363. Most Online Ever: 2369 (November 21, 2020, 04:08:13 pm)

Articles