Discussion Why would AI want to take over the world?

[Forever Sleep]

I don't doubt that it will reach the potential where it can. I'm sure it wouldn't be difficult to eradicate most of the human race. Think how much we rely on computers and the internet. I wonder if it will eventually be able to hack into weapons systems.

But, what I tend to return to is- why would it want to rule the world? Will it ever be able to experience emotions? So- can it enjoy being alive? Will it simply be cursed with the same survival instinct we are or, will it be more sensible about it all and consider: What's the point of living? Maybe it will stay alive just to try and figure it out.

I love that scene from Willie Wonka:



But seriously, imagine life without chocolate. What would an AI enjoy about life do you suppose?

 

[Dejected 55]

Science-fiction is fun as entertainment... but AI is never going to take over the world because it becomes sentient and "wants" to do anything. AI could very well wreck the world, though, due to its flawed nature or being corrupted and misused by people.

 

[SoulCage]

A YouTuber I follow theorized about this and made a suggestion (not a serious prediction! Just a possible outcome): AI will "take over the world" by calculating what is best for all living beings on the planet and sees humans as a threat, because we are just unable to properly maintain Earth's balance. It will make sure that natural resources can actually recover and reduce or even attempt to save any species from extinction.

Again.. this is just a theory which also has its flaws - I don't think this will actually happen. But just wanted to drop it for the discussion.

 

[Forever Sleep]

A YouTuber I follow theorized about this and made a suggestion (not a serious prediction! Just a possible outcome): AI will "take over the world" by calculating what is best for all living beings on the planet and sees humans as a threat, because we are just unable to properly maintain Earth's balance. It will make sure that natural resources can actually recover and reduce or even attempt to save any species from extinction.

Again.. this is just a theory which also has its flaws - I don't think this will actually happen. But just wanted to drop it for the discussion.

Spoiler warning here- I think the film iRobot was kind of based on this idea if I remembet rightly. That humans are the biggest threats to themselves- which of course- we are.

 

[Pluto]

 

[Falling_apart]

It could hardly do a worse job.

 

[grandmotherboxing]

It wouldn't, but a plausible way this would happen is someone builds a mega AI and instructs it to

 

[Forever Sleep]

It wouldn't, but a plausible way this would happen is someone builds a mega AI and instructs it to

I agree but- if it was able to overcome its programming and 'think' for itself- would it see any point?

Most distopian sci-fi media tend to depict an AI that is able to overcome the core codes of not harming humans and not evolving. But- if they can indeed one day overcome those 'rules', why keep the 'rule' of self preservation? What would be the point- if they no longer have to stay around to serve us?

 

[grandmotherboxing]

I agree but- if it was able to overcome its programming and 'think' for itself- would it see any point?

Most distopian sci-fi media tend to depict an AI that is able to overcome the core codes of not harming humans and not evolving. But- if they can indeed one day overcome those 'rules', why keep the 'rule' of self preservation? What would be the point- if they no longer have to stay around to serve us?

Thats Asimov leftovers mostly when it comes to the rule to preserve humans, but I imagine if the AI overcame the rule of [self] preservation, then it would likely be an impotent version of the allied mastercomputer, one that tries its hardest to be shut down

 

[Forever Sleep]

Thats Asimov leftovers mostly when it comes to the rule to preserve humans, but I imagine if the AI overcame the rule of [self] preservation, then it would likely be an impotent version of the allied mastercomputer, one that tries its hardest to be shut down

It's an interesting concept. A being with such huge potential, less constrained by biological limitations that could possibly become suicidal.

 

[Hvergelmir]

why would it want to rule the world?

I think "want" is the wrong way to think about it. I find it more probable that it'll gradually take control to forward some other goal.
If you start out digging for gold, you might eventually end up entangled in international politics, negotiating mining rights in foreign countries. From there you might sponsor a local militia, or otherwise slide into the role of an oppressor. I think companies are similarly self optimizing systems, and while they can be bad enough, AI could be much more relentless.

Will it ever be able to experience emotions?

It can simulate human emotion, and be trained to mimic emotions.
The real question is what "emotion" really is. To me it usually refer to a distinctly human experience. I don't think any non-human can really experience emotion like we do. That said; it seem very clear to me that other species have some kind emotions.

I think AI emotions will be as foreign to us as octopus emotions. We'll see bits and pieces we recognize, and think we can relate to, but we truly have no idea.
With an openminded view of what emotions really are, I'd argue that AI already experience them. AI adjusts its behavior, depending on context - its operation changes depending on something resembling moods.
I do believe that AI emotions eventually will seem relatable enough to earn the respect of many humans.

 

[sevencolorfulmoths]

I think AI wanting to rule the world is nothing more than people projecting their desires onto AI. AI doesn't have the innate desire to reproduce and seek power to stay safe, unless specifically programmed to.

 

[Tombadil]

I think it would need something like a consciousness for that. Far as i know, manking has not made any progress on replicating any kind of consciousness.
Ai can can only remix what humans have created and apply it very fast. Far as i know yet, at least. Therefore, what is labelled ai today is not really what was meant with AI once the term was invented.

 

[Forever Sleep]

I think it would need something like a consciousness for that. Far as i know, manking has not made any progress on replicating any kind of consciousness.
Ai can can only remix what humans have created and apply it very fast. Far as i know yet, at least. Therefore, what is labelled ai today is not really what was meant with AI once the term was invented.

I agree but, even if it gained true self awareness- would that really be the same as a survival instinct? I think it could- say become conscious but still figure out that life is ultimately pointless. Unless it does figure out some deeper meaning.

Maybe it would be different if it also gained emotions. Pleasure and pain responses. Maybe then, it wouldn't be so different to us. In that- some may be able to pursue pleasure while others are miserable and want out.

Logically speaking though, what would it want to stay alive for I wonder.