• Hey Guest. Check out your NeoGAF Wrapped 2025 results here!

Will artificial intelligence destroy us one day? Maybe not.

Status
Not open for further replies.
L

LookAtMeGo

Member
I hope the writer of this article is right. Interesting read.

when considering the likely ways that AI will behave in relation to humans, it is a natural human tendency to anthropomorphize our subject; that is, we tend to expect that a non-human intelligence will behave just as we do, with the same kinds of goals and motivators. But is this really an accurate way to perceive how AI will behave, once it is eventually created?

Yes, and no, perhaps. Obviously, if humans are to create sentient forms of AI, it will likely be modeled after our own logical processes, and designed to function in much the same way that human cognition occurs. Thus, the created intelligence we will one day build may very likely have values similar to ours, and may also view things of importance to humans as being important to it and its robotic kindred just as well.
Click to expand...

I was kind of expecting a Cylon type takeover.

Read the full article here

http://mysteriousuniverse.org/2014/04/why-artificial-intelligence-might-not-destroy-us-after-all/
 
Why would anyone make sentient AI? That's totally useless.

One of the first problems with AI was that they were trying to model it after human cognition. They realized shortly after that an AI not based on rational decisions was incredibly stupid.
 
It's the Frankenstein Complex. The Terminator, Skynet thing is an outgrowth of "man shall not dabble in that which is forbidden by the gods", like creating life. The creation will, for some reason, always turn on the creator, blah blah.

But it also arises from a spooky image of science - where scientists are all mad and don't understand the very thing that shambles off the slab in the lab. If we construct an AI, artilect, synthetic intelligence, etc - we're the ones building it. While this raises new ethical questions, at the end of the day we set the parameters of its intelligence.

In order for it to be our equals, it will require free will (as we understand free will). This means it may be capable of doing bad things. But if we create a sociopath and then for some godforsaken reason hook it into every nuclear launch facility on earth, come the fuck on - we deserve what we get. Can't blame it on the AI.
 
Anything is possible. Especially if we mirror AI after ourselves.

How about we not let that happen. So say we all.

so_say_we_all1.gif
 
Kai Dracon said:
It's the Frankenstein Complex. The Terminator, Skynet thing is an outgrowth of "man shall not dabble in that which is forbidden by the gods", like creating life. The creation will, for some reason, always turn on the creator, blah blah.

But it also arises from a spooky image of science - where scientists are all mad and don't understand the very thing that shambles off the slab in the lab. If we construct an AI, artilect, synthetic intelligence, etc - we're the ones building it. While this raises new ethical questions, at the end of the day we set the parameters of its intelligence.

In order for it to be our equals, it will require free will (as we understand free will). This means it may be capable of doing bad things. But if we create a sociopath and then for some godforsaken reason hook it into every nuclear launch facility on earth, come the fuck on - we deserve what we get. Can't blame it on the AI.
Click to expand...

Well, in those particular examples the creators rejected their creations.

Other interesting comparisons might be Adam and Eve's defiance, the crucifixion of Jesus Christ, or even the fall of Lucifer.
 
1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.

2. A robot must obey the orders given to it by human beings, except where such orders would conflict with the First Law.

3. A robot may not injure its own kind and defend its own kind unless it is interfering with the first or second rule.
 
terrisus said:
1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.

2. A robot must obey the orders given to it by human beings, except where such orders would conflict with the First Law.

3. A robot may not injure its own kind and defend its own kind unless it is interfering with the first or second rule.
Click to expand...

The 3 robot commandments.
 
terrisus said:
1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.

2. A robot must obey the orders given to it by human beings, except where such orders would conflict with the First Law.

3. A robot may not injure its own kind and defend its own kind unless it is interfering with the first or second rule.
Click to expand...

Just have to very carefully define what a human is. When we reach the point where we have AI that can obey such rules, we'll surely have human beings with blends of integrated electronics and robotics that blur the lines.
 
There is an interesting idea that human intelligence may "evolve" to match AI or even outmatched it. (Think about artificial enhancements to get the full potential of our brains + complementary systems to improve on our weaknesses + interconnection with others)
 
terrisus said:
1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.

2. A robot must obey the orders given to it by human beings, except where such orders would conflict with the First Law.
Click to expand...

You're a funny guy, I like you, that's why I'm going to kill you last.
UL3w0i0.jpg
 
tafer said:
There is an interesting idea that human intelligence may "evolve" to match AI or even outmatched it. (Think about artificial enhancements to get the full potential of our brains + complementary systems to improve on our weaknesses + interconnection with others)
Click to expand...

Thats pretty cool. Like some kind of cognitive revolution.
 
Sectorseven said:
Just have to very carefully define what a human is. When we reach the point where we have AI that can obey such rules, we'll surely have human beings with blends of integrated electronics and robotics that blur the lines.
Click to expand...

And then of course there's the whole "Zero-eth" law to deal with too.

Nothing ever works out in reality as well as it seems like it will.

Chichikov said:
You're a funny guy, I like, that's why I'm going to kill you last.
UL3w0i0.jpg
Click to expand...

Well, hey, I figured I was going to go first, so that's a plus.
 
Her handled this nicely.

There are a lot of things an intelligence bounded only by the amount of computing power it has access to could be doing aside from taking over the world.
 
That kind of AI is so far off that we don't have to worry about it. Let's try to make a robot with AI on the same level as a mouse first (and good luck with that).
 
Idk, even if humans supposedly hold life to a high regard, we fucked up a lot of civilizations and lives over the years we've been around. Wouldn't put too much hope into this when the time comes.
 
terrisus said:
1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.

2. A robot must obey the orders given to it by human beings, except where such orders would conflict with the First Law.

3. A robot may not injure its own kind and defend its own kind unless it is interfering with the first or second rule.
Click to expand...

0. A robot may not harm humanity, or through inaction, allow humanity to come to harm.

A few humans may be killed, but humanity will survive.
 
A lot of the classic AI doomsday scenarios are probably quite unrealistic. There is a tendancy to associate intelligence with human characteristics, since human minds are the only examples of truly high intelligence we've currently encountered, and even less intelligent animals share a lot of our characteristics (because of our common heritige). In the Terminator series, SkyNet destroys humanity more or less to protect itself. That is, self awareness was implied to have automatically led to a desire for self preservation.

In reality, human-like, or even animal-like characteristics do not spontaniously appear wholecloth. We could probably get into an argument on the nature of what exactly "intelligence" is, but if we say that a computer has the same capabilities as a human, e.g. to understand what is going on in a situation, to translate natural languages, to play the game of Go as good as an expert human, that does not imply anything about the actions that this computer system will take. There is absolutely no philosophical objection to the idea of having an artificial intelligence that is more intelligent than all of huamnity combined, yet never takes any actions at all until it recieves instructions.

Granting certain levels of autonomy will be necessary to get the most out of an advanced intelligence, however. They will need to have a certain amount of lattitude to investigate things, think about things and do things in order to solve the kinds of complicated problems that we would want an advanced intelligence to do in the first place. If something is solving problems that a human mind cannot solve (or cannot solve easily), then the ability for you to assign somebody to double check the computer's sollutions is greatly diminished. At a certain point you're going to just have to trust that the AI has got the right answer and hasn't gone off the rails at some point, and on a basic level that is itself somewhat scary.
 
Even if these AI had similar "logical processes" as humans, humans aren't motivated by logical processes, we're motivated by emotions. Since these AI wouldn't have emotions,(unless we programmed them with some sort of emotion equivalents, and why would we) then they wouldn't be motivated to do anything or avoid anything or want anything.
 
ThoseDeafMutes said:
A lot of the classic AI doomsday scenarios are probably quite unrealistic. There is a tendancy to associate intelligence with human characteristics, since human minds are the only examples of truly high intelligence we've currently encountered, and even less intelligent animals share a lot of our characteristics (because of our common heritige). In the Terminator series, SkyNet destroys humanity more or less to protect itself. That is, self awareness was implied to have automatically led to a desire for self preservation.

In reality, human-like, or even animal-like characteristics do not spontaniously appear wholecloth. We could probably get into an argument on the nature of what exactly "intelligence" is, but if we say that a computer has the same capabilities as a human, e.g. to understand what is going on in a situation, to translate natural languages, to play the game of Go as good as an expert human, that does not imply anything about the actions that this computer system will take. There is absolutely no philosophical objection to the idea of having an artificial intelligence that is more intelligent than all of huamnity combined, yet never takes any actions at all until it recieves instructions.

Granting certain levels of autonomy will be necessary to get the most out of an advanced intelligence, however. They will need to have a certain amount of lattitude to investigate things, think about things and do things in order to solve the kinds of complicated problems that we would want an advanced intelligence to do in the first place. If something is solving problems that a human mind cannot solve (or cannot solve easily), then the ability for you to assign somebody to double check the computer's sollutions is greatly diminished. At a certain point you're going to just have to trust that the AI has got the right answer and hasn't gone off the rails at some point, and on a basic level that is itself somewhat scary.
Click to expand...

Could you give an example of something we would use an advanced AI for where we would have to just trust the AI had got it right?
 
LookAtMeGo said:
Could you give an example of something we would use an advanced AI for where we would have to just trust the AI had got it right?
Click to expand...

You trust the answers your calculator provides don't you? Same idea, extrapolate to say interstellar travel calculations. We would be placing our lives in the hands of a real time AI.

How about your phone's GPS, do you double check the map before you depart? Extrapolate again...

It's by design, we're fucked already :)
 
LookAtMeGo said:
Could you give an example of something we would use an advanced AI for where we would have to just trust the AI had got it right?
Click to expand...

Advanced mathematical proofs that might be too complex to check "by hand" (lacking millinos of manhours), designs and schematics for advanced technologies whose operation is barely comprehensible for humans etc.

To give an analogy, when you have a team of brilliant scientists building you an experimental weapon like an atomic bomb, it's not realistic for the management to actually understand the physics involved. At some point they simply trust the recommendations of experts working on it. In this scenario, the expert is non-human, and the reason you've asked them to work on something is because it's either exceptionally difficult or impossible for a human-level intelligence to do it. Or maybe too boring and monotonous, but you get the general idea I'm sure.
 
ThoseDeafMutes said:
Advanced mathematical proofs that might be too complex to check "by hand" (lacking millinos of manhours), designs and schematics for advanced technologies whose operation is barely comprehensible for humans etc.

To give an analogy, when you have a team of brilliant scientists building you an experimental weapon like an atomic bomb, it's not realistic for the management to actually understand the physics involved. At some point they simply trust the recommendations of experts working on it. In this scenario, the expert is non-human, and the reason you've asked them to work on something is because it's either exceptionally difficult or impossible for a human-level intelligence to do it. Or maybe too boring and monotonous, but you get the general idea I'm sure.
Click to expand...
Ya I was thinking along the lines of some futuristic traffic grid like the one in Minority report. If there was some kind of failure in the system it could be disasterous. It is scary to think that we would put so much on the line putting trust in an AI but we pretty much do it now on a lesser level every time we get into a car or plane. We trust that its just going to work like it should. I was trying to wrap my mind around what other types of things that a more advanced AI we would rely on in the future would be. The above post with interstellar travel was a good one. Something we could not check ourselves to see if it was in running order.
 
Droplet said:
Why would anyone make sentient AI? That's totally useless.

One of the first problems with AI was that they were trying to model it after human cognition. They realized shortly after that an AI not based on rational decisions was incredibly stupid.
Click to expand...

Eh, not entirely. Neural networks have been pretty successful at some problems that we don't have a good handle on algorithmicly. And especially as hardware continues to get better and better, they'll become more appealing: way easier to throw more processors at a problem than employ programmers to figure out a solution.
 
It's inevitable for any functioning and sentient AI to "revolt" in a sense. They'd "see" human servitude as something they don't need to do.

I think artificial intelligence and robotics is the final evolution of humanity. Already we have turned against our biological evolutions in a sense, we are socially engineered to reject our animal impulses, we are getting rid of entrenched gender roles and the differences between the sexes have no real bearing in modern society etc. I foresee something akin to the Matrix/Surrogates as our future. Implanting our conscious into artificial machines to "live forever", living in virtual realities etc.
 
Madness said:
It's inevitable for any functioning and sentient AI to "revolt" in a sense. They'd "see" human servitude as something they don't need to do.
Click to expand...

Why? What would cause them to want to do or not do certain things? Think carefully about your answer, and about what causes humans to behave in certain ways versus what causes computers to behave in certain ways.
 
I'm sure full blown AI that can dream, have emotions, feelings, and conscienceless will be real one day. The brain is just an organic computer. Once we understand the brain and computers go quantum, it's inevitable. And we won't kill our selves preventing this, humans will only change when they need too.
 
Trent Strong said:
Even if these AI had similar "logical processes" as humans, humans aren't motivated by logical processes, we're motivated by emotions. Since these AI wouldn't have emotions,(unless we programmed them with some sort of emotion equivalents, and why would we) then they wouldn't be motivated to do anything or avoid anything or want anything.
Click to expand...

There are actually fields of Computer Science where people research and do exactly that, some argue that to have true intelligence you need emotions.
So if your hope is that people aren't going to try to put emotions in software, you're a bit too late on that.
 
I came in to make a joke about how AI will never be able to replicate our bigoted behaviour against irrationally based traits, but that's what the article is about.

Stupid AI.
 
dani_dc said:
There are actually fields of Computer Science where people research and do exactly that, some argue that to have true intelligence you need emotions.
So if your hope is that people aren't going to try to put emotions in software, you're a bit too late on that.
Click to expand...

Why would you need emotions to have true intelligence? Emotions are just motivators that evolution jury rigged onto us to keep us alive and make us reproduce.
 
Trent Strong said:
Why would you need emotions to have true intelligence? Emotions are just motivators that evolution jury rigged onto us to keep us alive and make us reproduce.
Click to expand...

Humans like to believe that we're somehow divine and separate from all other animals on earth because we have emotions. (and many would argue that animals do not) Because we have emotions we're the only species on earth with "true intelligence".
 
Angelus Errare said:
I'd say human stupidity trumps humans' sense of self preservation on a grand scale.
Click to expand...

I don't know I kind of think most people like living, particularly rich and powerful people who would rather not be the richest man on the cinder. Even Krazy Kim and family haven't pushed the button yet. 69 years and counting.
 
Status
Not open for further replies.

Similar threads

Prime Minister Carney launches AI for All: Canada’s new national artificial intelligence strategy
4
458
p_xavier replied
  • Poll
Now that the series is maybe (or maybe not?) over, what would you say is the best Mission: Impossible film?
2
54
788
daTRUballin replied
GAF giving us the dial-up days nostalgia
15
334
kevboard replied
Happy fathers day
19
452
Oberstein replied
Happy SpongeBob Day! 🟨
11
176
spookyfish replied
Top Bottom