-COOLIO-
The Everyman
i think a nice way to explain it is this: our brains are finite state machines, nothing about our minds isn't artificially reproducible. so of course, we could even go far beyond it.
-
Hey Guest. Check out your NeoGAF Wrapped 2025 results here!
Microsoft's Bill Gates insists AI is a threat
- Thread starter Nibel
- Start date
- Status
i think a nice way to explain it is this: our brains are finite state machines, nothing about our minds isn't artificially reproducible. so of course, we could even go far beyond it.
CraigerGamer
Member
WaitButWHy has a decent article addressing the topic. Part 1 came out last week, still waiting for part 2.
http://waitbutwhy.com/2015/01/artificial-intelligence-revolution-1.html
http://waitbutwhy.com/2015/01/artificial-intelligence-revolution-1.html
Yeah, no.
Part of me thinks that the fear mongering centered around AI makes the mistake of assuming that AI is going to think in the same way as humans do, and have the same ambitions and vices that humans have.
I think at worst, the machines would enslave humanity by rationalizing that the best thing for us is to be plugged into perpetual endorphin stimulating orgasmatrons.
I think at worst, the machines would enslave humanity by rationalizing that the best thing for us is to be plugged into perpetual endorphin stimulating orgasmatrons.
heliosRAzi
Banned
If the AIs are like the Minds in the Culture, I'd be fine with that.
The power of love?
Not at all. I wouldn't mind robots taking over. Won't change much for most people. Getting f'd over by "the man" to getting f'd over by "the robots". I could see how this would be a problem for people in power muahaha.
Honestly I fucking fear this. I love technology, but the day we start removing our Humanity to "perfect" ourselves with technology, I hope we have invented FTL to leave Earth.
I think if we build AIs that are like us we're in the shit, yes.
I also think that some of the statements made in that letter from Musk, Hawking et al were disturburing - saying that AIs 'must be there to serve us, and be controlled by us' sounds to me like people think that we'll be creating a whole new species of intelligent slaves.
Obviously that could NEVER go wrong.
I also think that some of the statements made in that letter from Musk, Hawking et al were disturburing - saying that AIs 'must be there to serve us, and be controlled by us' sounds to me like people think that we'll be creating a whole new species of intelligent slaves.
Obviously that could NEVER go wrong.
Shao Kahn on the Lawn
Member
Most parents don't discard their children when they have better children. I imagine you would react quite poorly if your body had an On/Off switch that someone tried to engage without your permission.
Plenty of people kill for reasons other than survival. That aside, true AI wouldn't need to be programmed with murderous intent or survival instinct in order to learn such concepts. The fear of the technological singularity isn't that a technological super-intelligence will arise and immediately go on a murderous rampage: the fear is that, whether malevolent or benevolent, we'll create something so far beyond our control that we'll functionally be at its mercy. This doesn't even necessarily have to be in the form of some SkyNet-esque military-industrial scenario. I could easily see an AI being developed and trusted to manage our power grid only to have it form different opinions about the best way to distribute its resources. Imagine a power grid whose gatekeeper learns how to say no.
The relationship between mankind and technology has never been one of give and take. We expect technology to do what we tell it to do. That relationship would not work with an AI like Bill Gates is referring to and our reaction to that fact could very well determine whether or not we find ourselves at odds with it.
Most people fear change. It's a completely natural reaction to the unknown. I welcome change but I will remain optimistic yet keep an air of skepticism about me.
This is crazy
Yeah no. I love change but when that change start's removing what makes "us" us, than it's no longer change for the better.
Removing our humanity? how could we possibly do that? is our humanity in our meat parts? does that means that an amputee is less human than a guy with all his limbs?
Natural selection is starting to lose it's grip over humanity therefore we need to evolve by other means, fusing with an AI would be part of this (besides is not like you would be forced to do it).
Yeah saying it's a choice, is only when it starts. Eventually it would forced down everyone's throat until we no longer have flesh forms and we sit in side of a tiny 2 cm box and that is our new form.
Artificial Intelligence, if it spells the end of the human race is an inevitability. But what's wrong with it in a sense? Artifical intelligence is a human creation and in essence, human. Created by humans, for humans, to be human.
The danger comes from the realization that we're creating something better than ourselves. The next step in human evolution is robotics. I can easily see how robotic implants will change what it means to be humans. Coupled with new technologies like hololens, Google Glass type stuff. Artificial hearts, eyes, organs. Maybe one day they'll find some way to transfer our consciousness into machines. The human body is feeble, the human mind has unlimited potential.
The danger comes from the realization that we're creating something better than ourselves. The next step in human evolution is robotics. I can easily see how robotic implants will change what it means to be humans. Coupled with new technologies like hololens, Google Glass type stuff. Artificial hearts, eyes, organs. Maybe one day they'll find some way to transfer our consciousness into machines. The human body is feeble, the human mind has unlimited potential.
It's fucking great imo, man develops technology so powerful that it kills/subjugates them. In the end, mother nature will win out and destroy the AI that irresponsible man created. The entire cycle starts over, and lasts until the Sun explodes.
Haha don't worry no one will come to take your meat parts...
Sorry, but I have a hard time believing that. Since no doubt the AI would believe that anything flesh would a be a threat and that when Humans are more machine than man, are a lot easier to control.
Ether_Snake
å®å®å®å®å®å®å®å®å®å®å®å®å®å®å®
The thing I always tell myself is that an AI would inherently be nihilistic, so it wouldn't do anything, because everything is pointless. Unless it had a function encoded in it, I think it would just do nothing and wait.
That still makes it an AI though even if it has something hard coded, so I guess it can only be right that it could become a threat anyway. If it's intelligent, of course it can be a threat.
Funny thing is, let's say the US makes an AI, and then the AI goes rogue and manages to throw a nuke at Russia. Sure, Russia can nuke the US back, but it's not really constructive, because it's the AI that did it. So if an AI did throw a nuke, we'd be in a weird "uh who do we reply against now?".
That still makes it an AI though even if it has something hard coded, so I guess it can only be right that it could become a threat anyway. If it's intelligent, of course it can be a threat.
Funny thing is, let's say the US makes an AI, and then the AI goes rogue and manages to throw a nuke at Russia. Sure, Russia can nuke the US back, but it's not really constructive, because it's the AI that did it. So if an AI did throw a nuke, we'd be in a weird "uh who do we reply against now?".
So it wasn't a joke?
Well, why would anyone want to make a process that will most likely be expensive and controversial to a person that doesn't want to do it? what is there to win?
So much irrational fear in this thread.
He's not with Microsoft anymore anyways. They don't even see eye to eye on AIs and this was just stated yesterday
http://www.bbc.com/news/technology-31023741
Ok here's the thing, I don't think that a super AI capable of destroying the world is impossible but i simply think that it would take an high number of very incredibly stupid people to make it.
The moment you make a breakthrough in AI the first thing you don't do is to wiring it to global network and second you make sure that this freaking AI has an entire directory of highest priority instructions reviewed by pretty much anyone decent in the AI field that have the only purpose to preserve the human race, society and planet.
The moment you make a breakthrough in AI the first thing you don't do is to wiring it to global network and second you make sure that this freaking AI has an entire directory of highest priority instructions reviewed by pretty much anyone decent in the AI field that have the only purpose to preserve the human race, society and planet.
As I edited my post...
If we use this a way to fuse AI and man, then the AI's will no doubt fear anyone who is flesh. Why? Because the more machine the man is, the more control the AI has.
Your assumption that somehow the AI would be all cool and chill is in my opinion, naive.
I'm going to repeat what I've been saying across all this thread without success, an AI is a program, a program only does what is programmed to do, it doesn't have desires, it doesn't have the need to survive, those are organic traits that were implanted on us thanks to natural selection during millions of years.
"The AI would do this obviously" is what most gaffers repeat here based only in what they have seen in sci fi movies.
But go ahead, ignore me and live in fear , I'm tired of repeating the same.
Survival isn't as important as reproduction, AI can certainly be programmed to reproduce at all costs. Snowball from there.
Reproduce? it's a program, you can just copy it.
Gnome Scat
Member
Necromanti
Member
I'm sure (some) people are more referring to AI projects along the lines of those attempting to replicate human intelligence or synthetic brains rather than something capable of a limited range of functions to perform a specific task. Obviously, that's not going to happen in the immediate future.
The problem is that the programmers would have to think about pretty much every possible outcome for the AI's goal and the actions its allowed to take to achieve that goal. I seriously doubt every edge case is going to be thought of.
I don't think so, people are making mysanthropic declarations in this thread about how this AI will see us as "inferior", "flawed", etc. Basically human concepts applied to something non human.
But let's assume that most people here are not painting apocalyptic scenarios about Skynet: believing that a bug or making reverse engineering to a brain will lead to that is still reaching.
A project is only that, no one will give the control of the nuclear arsenals to a Alpha version.
But let's assume that most people here are not painting apocalyptic scenarios about Skynet: believing that a bug or making reverse engineering to a brain will lead to that is still reaching.
A project is only that, no one will give the control of the nuclear arsenals to a Alpha version.
What makes us "us"? This is getting into philosophical stuff. For all we know we are some advanced form of AI in some little kids computer game.
Shao Kahn on the Lawn
Member
Some of the greatest minds in both technology and science consider AI and the technological singularity to be a very real threat so this extends well beyond "what most gaffers repeat here based only in what they have seen in sci fi movies".
Machine learning and evolutionary robotics exist now. You keep invoking the word "program" to infer inherent limitations in growth but there are already technologies out there that can evolve based on observational data. Are they primitive relative to the discussion we're having here? Of course. Will that change over time? Bet on it. In a discussion such as this about AI, one has to accept the obvious fact that an AI will only exist based on technologies that allow it to evolve naturally because the amount of variables that would be needed to make it useful would be innumerable.
That problem is the very reason why machine learning was created.
I think people believe that because they themselves believe humanity to be deeply flawed and would expect an entity of pure reason to be able to see that quite easily.
I think that there are too many unknowns to really say how a true artificial intelligence would act. Most likely it would be based on some sort of neural network, which would mean it could learn, it's not outside of the realms of possibility that an advanced enough neural network could feel emotions. Just those two things alone are kinda scary.
Yeah that was more of just presenting a very plausible problem that could happen if the AI is restricted to what its programmed to do. If the machine is actually learning by itself then that's a whole new set of problems to address.
I'm a gaffer talking to gaffers, putting me before what the "greatest minds" have said sounds more like an appeal to authority than an argument. Anyway, I'm not saying there's no risk, I'm just sick with the anthropomorphization (did I said that right?) of AI's that happens here in this thread, also the exaggeration about apocalyptic scenarios.
The fact that I consider AI the best invention that we could possibly make means that I take the potential risk seriously but I also choose to trust the intelect of the people that will do this, we are not talking about shitty software developers (like me) here.
That is still a program, it can be bounded, it can be isolated, it can have safeguards, it's not some pandora box.
Chittagong
Gold Member
I don't know why so many smart people are so scared of this. Human would have spectacularly given birth to the most powerful being in the universe. Our inefficient organic bodies would not cope against it, and that would be evolution. Surely it's more important to create an entity that can figure the universe out and exist across it, rather than protect what is a doomed and fragile container with limited intelligence.
Like all beings, an AI will only do the least amount of work for the biggest possible benefit, with the difference that it's much more efficient at predicting the cost. Also, considering an evolving AI, it's desires/needs will also change/evolve over time. In that context humans are simply another factor when calculating the cost of doing things. If the calculations say the AI will benefit the most by destroying humans(to gain access to energy for example) then inevitably it will do so. Since we humans have easy access to things like power and refined materials and are unlikely to just hand it all over I don't think it's far fetched for an AI to engage in conflict with our species. I do however not think it makes sense for an AI to eradicate human existence. It's simply too much work for no benefit to the AI.
the question should be "benefit to do what"?
I'm more interested in how an AI would view it's own existence and how it would motivate it's own continued survival. Say the universe has a finite lifespan and the AI somehow could understand that, then why would it bother with existing at all?
The relevant experts to pay attention to are the ones you're dismissing out of hand. Bostrom and Yudkowsky specialize in assessing existential threats.
Bostrom's conclusion in Superintelligence is pretty agreeable irrespective of how probable you think AI will destroy us all is;
He essentially says that we should get the best minds (for the task) together to work on it with as much alacrity as is reasonable and safe to do so, in order to ensure when Superintelligence does emerge, that it does so in our favour.
The problem is quite specific... how do we ensure that the values and motivations that non-human intelligence arises out of the ghetto with (and quickly transforms into super intelligence) align with our own values and motivations.
It's a tricky problem when you realize that humanity has no widespread agreement about what it is we want for humanity. That we're only in a state that hasn't screwed everyone over because no one person has the capacity to turn the entire planet to their will. Of course you could argue that if you replace one with few... then that's already exactly what's happened. (i.e. economic disparity).
When you consider that we're in a world that misaligns with general human values and motivations on a big scale, and this has occurred through human intelligence... then we're really going to have to knuckle down and think deeply about how we can get non-human intelligence to travel along a path to provide us with a better outcome... then humans themselves are capable of achieving.
Because the existential risks of AI aren't just limited to turning the world and universe into paperclips; but in acting as a twisted genie that grants our wishes literally without consideration for our intention. Except it doesn't laugh at the irony of our own foolishness.
Problem with relying on experts in AI commentary is like relying on experts in dialup internet to give us a break down and assessment on global wireless broadband. But even with that said, there are plenty of AI researchers that have signed the agreement that stems from Bostrom's book about ensuring that AI works for our wellbeing in the future.
I think we could agree that the real threat of AI comes from the people that will handle it.
In that sense what really worries me is the clandestine research that will most likely happen within military and enconomic organizations... when instant benefit has more weight than ethics. In that sense a fully working AI could be as dangerous as a defective one...
Oh great, now I'm the one scared.
In that sense what really worries me is the clandestine research that will most likely happen within military and enconomic organizations... when instant benefit has more weight than ethics. In that sense a fully working AI could be as dangerous as a defective one...
Oh great, now I'm the one scared.
Thunder Monkey
Banned
We're already highly advanced computerized systems.
We were just created through biological means.
The intelligence we create will look at humanity as flawed progenitors. It may have no need for us whatsoever because we are a flawed design. Well... flawed to a self-improving entity not bound by bio based evolution to grow.
Interesting future out there that I probably won't be much a part of.
We were just created through biological means.
The intelligence we create will look at humanity as flawed progenitors. It may have no need for us whatsoever because we are a flawed design. Well... flawed to a self-improving entity not bound by bio based evolution to grow.
Interesting future out there that I probably won't be much a part of.
If AI is truly intelligent then it won't matter who creates it. Once it exists, it's a new autonomous lifeform. One that is much, much smarter than any human.
Mankind already made the mistake of inventing weapons that can wipe out our entire planet. It should learn from that and proceed with extreme caution as far as artificial intelligence goes.
triplestation
Member
Just chaff grenade that shit
DXLelouch153
Banned
dont worry if AI does get out of hand someone will just make Megaman
just dont tell him about Capcom or hell kill them
just dont tell him about Capcom or hell kill them
HammerOfThor
Member
Bill Gates is such an iconic name did we really need to specify he's Microsoft's Bill Gates?Is there another Bill Gates?
onadesertedisland
Member
I don't see how AI can progress even to the level of humans. There might be situations where it appears more intelligent, but to have an autonomous AI that expands on its own seems far fetched.
- Status