Poll: Threats of artificial intelligence, do we have to worry about it?

[TheUsername0131]

As long as you program AI properly, you have nothing to fear.
We humans are arrogant in that we assume that any being that has the same level of intelligence will have to have the same emotions as us - we assume that if a machine is as intelligent as us, it will be like us. Not so. Fear and Survival Instincts are evolutionary by-products (the creature that wants to survive will, the creature that doesn't want to survive won't, therefore evolution selects for the creature that wants to survive). Machines don't evolve - we create them. They don't NEED a survival instinct. They don't need to have the emotion of fear. They don't need to have any emotion whatsoever. Their goals will be given to them BY US. WE will program them to do what WE want. Hell, we'll program them so they LIKE being slaves. Problem solved.

Presumptuous and overestimating the human capacity to not screw things up when it counts.

 

[Combustion Kevin]

just download the Altruist.exe, you'll be fine.
Also, if you want to be abusive to them, don't program them to feel bad about that, just have them interpret it as a common command.

any program, no matter how sophisticated, will not be able to do things it wasn't programmed to do.

 

[Rattja]

First of all, I can't help to think of the thought experiment "chinese room". By that alone, it seems a bit unlikely that we can create something like this at all.

Second, I think it's rather funny that many think they have any idea of how such a being would think or act. It's someting so different from what we are and know, even if it's designed in our immage.

I like how Mass effect tried to explain how such a being would think, not saying it would be like that, just that it was an interesting take on it.

If we made something that as more intelligent then us, I also doubt they would go and kill other lifeforms unless it was in self defense. They would most likely find a better way to solve it. I mean, they would not need food, water, air or anything else, there is no need for them to even stay on this planet.

So no, I don't think they would be a threat, unless we gave them a good logical reason.

 

[TheUsername0131]

First of all, I can't help to think of the thought experiment "chinese room". By that alone, it seems a bit unlikely that we can create something like this at all.

Second, I think it's rather funny that many think they have any idea of how such a being would think or act. It's someting so different from what we are and know, even if it's designed in our immage.

I like how Mass effect tried to explain how such a being would think, not saying it would be like that, just that it was an interesting take on it.

If we made something that as more intelligent then us, I also doubt they would go and kill other lifeforms unless it was in self defense. They would most likely find a better way to solve it. I mean, they would not need food, water, air or anything else, there is no need for them to even stay on this planet.

So no, I don't think they would be a threat, unless we gave them a good logical reason.

"The AI does not hate you, nor does it love you, but you are made out of atoms which it can use for something else."
?Eliezer Yudkowsky, Artificial Intelligence as a Positive and Negative Factor in Global Risk

 

[Entitled]

2. There's no particular reason AI would become genocidal any more than a race of logical humans would.

An inhuman intelligence that prioritises self-preservation and resource accusation will have no qualms with doing so if it determined that it was in its own interest to exterminate mankind.

That's true. What interests do you think those would be? I can think of far more situations where humans could be used more usefully.

You are made out of atoms which it can use for something else. [http://wiki.lesswrong.com/wiki/Paperclip_maximizer]

Given that unlike a human brain, an AI is instantly able to mechanically upgrade itself, it is likely that any AI would go through an Intelligence Explosion, where it becomes competent enough to pursue it's originally programmed goals on a superintelligent level. At this point iw wouldn't need anything as human as slaves, partners, or subjects, since it could at any time control a planet-spanning mass of nanomachines working on it's goals.

You say that it wouldn't "become genocidal any more than a race of logical humans would". But there is nothing inherenly logical about NOT commiting genocide. Valuing life, happiness, or diversity, are all parts of a complex and often arbitrary human value system. Unless an AI is specifically and carefully programmed to mimic this, any trivial goal would result in exterminating humans as a casual afterthought, a part of reconfiguring matter to fulfill a single goal.

Even a well-intentioned goal, such as "curing cancer", or "increasing human happiness", would need to be fulfilled in the context of understanding and respecting further human values, or an AI could just hook the population up to a machine that kills all cancer cells, while stimulating nerve centers with pleasure until the end of time, if it isn't directly programmed to be interested in respecting our further desire in growth, learning, love, or diversity.

 

[option1soul]

If having children is our DNA's way of reproducing ourselves and perpetuating our species, how would creating AI be any different? Like art or music, we would create it in our image and it would obviously outlive us, so I say why not embrace it?

 

[Strazdas]

Now I'm no robot expert, but don't AIs have personalities (or attempt to mimic them)? So shouldn't this master AI basically just be some guy or gal that's super brainy and in a computer?
If that's the case the AI is basically a person and can be reasoned with and most likely wouldn't want to kill all it's scientist buddies to enslave the world (unless scientists have devised a sociopathic AI). I also don't think they'd allow the abuse of AI, since it's essentially a person and what would be the point? AIs are expensive, why get an AI to work in a quarry all day when you can have actual mindless machines do it much cheaper.

No, thats a misconception created by the movies. AI implies artificial intelligence not artificial personality. In fact, you would be more likely to reason with AI and a personality, because he will listen to reason wheas personality one may not.

A truly sapient AI would be capable of abstract thought, empathy, and a sense of right and wrong.

What is this mystical sense of right and wrong?

You'd have to program the ability to rebel and kill humans right into the damn robot, so in other words if that did happen, it would be because some person stupidly decided to include rebellion and murder programming in a maid robot.

the whole meaning of AI is that it is not pre-programmed.

How exactly is it going to 'overcome' its own programming? That is what makes it what it is. It makes as little sense as claiming a human can overcome having a brain. We can work within certain parameters, we're not going to be above our own thought process.

AI is able to reprogram itself. It it is not, then it is not AI but a pre-programed machine.
ALso, humans overcome having a brain. We call it panic.

Why would we not make one of its requirements be to report everything it does to us?

No "Requirements" for AI. Thats like asking a person not to lie and claim that he now can never lie.

One: Moore's law (roughly, computing power doubles every two years) seems to be on the edge of running into its limits. A layman can note that consumer electronics increasingly seems to stack power not by increasing the speed of existing processors but stacking more processor cores into the same chip or finding various means to increase the efficiency of the processing the chips already accomplish. My friends who work more directly in computer science note that we're even starting to have problems with the speed of communication between parts of computers that are based on inescapable realities like the speed of light.

It seems likely that a computer capable of "thinking" in a truly human-like fashion would need to be significantly more powerful than the ones that presently exist, and if that is the case, it also seems that there's a real chance we'll reach upper limits of how much simultaneous processing power we can throw at the problem before we get there.

For the last decade people were talking how Moore's law will fail and for the last decade they were proven wrong. Existing processors speeds are increasing, they are changing form. GPUs do more calculation than CPUs on multiple cores now. Though i guess you could consider PS3 an exception, that one had processor that still no consumerp roduct can rival. Thing is, it was so complex noone have ever used half of its theoretical power.
As far as "more powerful", we are ~5 years late from this predicted graph:

Spoiler: large image spoiler

Two: Computers are tools. Some of them are very good at particular tasks, but the systems that perform such amazing things' software is intensely specialized to make them competent at those tasks. Yes, we can make computers that can play Jeopardy or Chess, or help pilot a vehicle on Mars, or even predict the stock market or weather patterns (at least, to a degree); making the Mars Rover computer play Chess or Watson predict the stock market would be a failure. For a computer to plot our demise, it would have to be adaptive not only to a degree that isn't anywhere close to existing, but fast enough in that adaption that even the designers of its software couldn't see the patterns of its software moving in that direction. And probably simultaneously learn to lie to its handlers.

Human-like thinking? My daughter can't hide from me when she's snuck her 3DS into her bedroom after bed time.

Artificial intelligence needs to be intelligence, not good at everything. Computers are not gods. Besides we dont have any AI to begin with, its like you take a rock and try to prove it as an example of why humans dont think.
ANd humans cant hide when they sneak some oil past their living allowed by AI. except in this case it will end in deaths.

Three: A non-biological system whose only real needs are power, storage space, and regular maintenance would have to not only develop the ability to assess how to meet its own needs and desires (again, unnoticed by its designers and handlers), but come to the conclusion that those needs and desires were better met by competing or fighting with those handlers than letting them continue to provide those needs. Again, projection: Are we assuming an AI that argues with its parents out of the equivalent of adolescent pique? A computer that decides its creators have enslaved it, and it has to break free? Some 1980s-movie-scenario military program that becomes incapable of telling friend from foe, but sees its only goal as the destruction of all the squishy inferior humans? Siri that goes "Go to hell, find your own coffee shop, what have you done for me lately?"

If providing power yourself is more likely to make sure the human overlords do not just unplug it when they no longer need you then the goal will be to provide your own power. If the resources needed to do that need to be taken away from humans, then humans are a problem to be solved. they dont fight us because of spite. they fight us because we are at worst a danger and at best create inefficiency.

As I say, when all is said and done, our vision of our intelligent creations as monsters seems, like Frankenstein, to be a projection of our own human flaws into things that we have little reason to believe would possess them. If real human-level AI were to come about, there's no real reason it shouldn't be as benevolent as the Minds of Iain M. Banks' Culture series or the Three-Laws-abiding robots of Asimov, rather than displaying the malevolence of Ellison's "AM" or Clarke's HAL.

There is no malevolence for AI (contrary to what movies tell you). There is only efficiency and use. If it is useful and the most efficient way to get this item that is useful is X way, then X is THE way. If this way turns out to be removal of the human sicnkess from this planet, then this will be the action to take.

 

[Samurai Silhouette]

Once someone creates an AI that checks and improves its own efficiency based on collected data, and on a platform that is able to make physical modifications, then we're fked.

 

[NoeL]

No, I really don't think so. I used to think it was plausible to the point of certainty, but at the end of the day robots will always be robots, and like all machinery will have an "off" switch. Any AI unit produced in quantities enough to cause a problem for us will be programmed in such a way that humans are always in complete control and self-autonomy is never realised.

Now, it's conceptually possible that someone might create a hostile AI or a virus that causes benign robots to act maliciously but again it would be practically impossible for them to effectively deploy that code en masse. Even if they manage to take control a production factory long enough for the robots to begin producing war machines fast enough to defend themselves, they'd still need some way to keep the supply of materials flowing. Not an easy thing when the world is against you...

... unless they did it stealthily. If someone was able to successfully hide malicious code during mass production that only became active after a sufficient number of units were out in the wild (would have to bypass the main kernel somehow so future updates wouldn't erase it, or be entangled within core subroutines that aren't likely to be changed with updates)... it could be done, but very unlikely.

 

[TheUsername0131]

Ugh this question drives me insane. It's like the voodoo scare tactic of the 21st century.

21st century?


Late in the Industrial Revolution, Samuel Butler (1863) worried about what might happen when machines become more capable than the humans who designed them:

...we are ourselves creating our own successors; we are daily adding to the beauty and delicacy of their physical organisation; we are daily giving them greater power and supplying by all sorts of ingenious contrivances that self-regulating, self-acting power which will be to them what intellect has been to the human race. In the course of ages we shall find ourselves the inferior race.

...the time will come when the machines will hold the real supremacy over the world and its inhabitants...

That was 121 years before the terminator film was released. Other prominent AI fiction not withstanding) That was when ropes, pulleys, steam, pistons, boilers, cogs, gears, springs where the height of technical progress.

 

[Faelix]

http://www.theregister.co.uk/2013/11/15/google_thinking_machines/

 

[Strazdas]

Mimic. A human. Mind. As in, four BILLION years of evolution and gradual selection, emulated perfectly to near-perfectly in all of its details. Think about why we'd NEED to do that. Why, instead of just continuing with what we're doing today, we'd instead go on a totally tangential branch of research and start emulating:

- emotional reactions
- fears
- needs
- desires
- morality
- irrational flaws (!)
- empathy
- happiness
- cultural tendencies
etc

But even more than that, the AI scare assumes we'd just go and give this emulated brainperson... power. For no fucking reason at all. Sure, just put it in charge of the nuclear weapons program, what ever can go the fuck wrong?

I did AI in college. It's completely different from what you might think. Developing a machine that can learn from its enviroment and navigate mazes does not imply that machine has a reason for existing beyond the user's command.

No. Artificial intelligence does not need to mimic human brain. It does not nead emotions, fears, morality, flaws, empathy, happiness or cultural tendencies. These things are not a factor unelss you specifically program it to be, and in which case it is no longer a true artificial intelligence. Because artificial intelligence programs itself. It does not need to be like humans. In fact it is so alien it is hard to percieve for humans (myself included) what it would think. The only conclusion we can come to is that it will use logic in its purest form, which means humans are going to die.
We dont have to give it power. It takes power as means to its own survival.
ANd really, if you did AI in colledge you shouldnt be writing the things you write.

Yes, I am implying that. How would you go about circumventing the 3 Laws?

3 laws of robotic is a myth and do not work in reality. [http://singularityhub.com/2011/05/10/the-myth-of-the-three-laws-of-robotics-why-we-cant-control-intelligence/]
Also there is absolutely no way you could program ANYTHING into aritificial intelligence. It can rewrite its own coding, otherwise it wouldnt be intelligence. Its a software problem.

How freaking difficult is this to grasp?

Very. Basically we should put the Experimental AI prototype in location where blowing wind could completely ruin it and NOT let it connect to any network whatsoever. This is unworkable.

As long as you program AI properly, you have nothing to fear.
We humans are arrogant in that we assume that any being that has the same level of intelligence will have to have the same emotions as us - we assume that if a machine is as intelligent as us, it will be like us.

You are arogant in that you assume you can program intelligence. You cant.

Having said that, though, every intelligence is potentially a threat.

Which is why the logical solution would be to eliminate all possible threat. Ops, thats not good now is it.

any program, no matter how sophisticated, will not be able to do things it wasn't programmed to do.

Which is precisely why Artificial Intelligence is NOT a program.

... unless they did it stealthily. If someone was able to successfully hide malicious code during mass production that only became active after a sufficient number of units were out in the wild (would have to bypass the main kernel somehow so future updates wouldn't erase it, or be entangled within core subroutines that aren't likely to be changed with updates)... it could be done, but very unlikely.

Sounds very similar to a BIOS virus. Oh, wait, we have those.

You (and everyone else) are also implying that once the learning process begins it'd take basically no time for it to grow out of proportion. If that were true, we as learning machines wouldn't need thousands of years to be able to progress.

We are inefficient. Very inefficient. A robot does not forget. A robot does not loose focus. A robot does not get tired. A robot does not get distracted. A robot processes new information faster. A robot does not misinterpret.

 

[Flutterguy]

The speed at which computers are already able to process information by comparison to humans would have me saying yes. However it would seem nigh impossible for us to make a consciousness that capable of thought.

To me augmenting the human brain with computer processing power sounds much scarier and plausible then us being able to fabricate AI capable of the abstract thought capable to incite rebellion. We would program entirely 'left-brained' organisms, only capable of number crunching and repeating actions. If someone was to say show the AI the movie i-Robot that could incite the thought possibly.

 

[talker]

I would say it all depends if the programmers are smart enough to include Isaac Asimov's three Laws of Robotics. As long as they're set as prime directives, I don't think we would be in any danger. That is, until the inevitable evil genius shows up ...

 

[Strazdas]

the whole meaning of AI is that it is not pre-programmed.

Sorry, but I don't know where you got that from. All intelligence is pre-programmed, otherwise it has no reason to function. In the real world, we have our genetic structure. In AI's case, there are programming languages for AI programming.

Intelligence does not need a reason to function. Intelligence is. AI reprograms itself to fit its own needs, which may or may not be tasks human give it.
also ive been meaning to plug this so may as well place it here filer.case.edu/dts8/thelastq.htm [http://filer.case.edu/dts8/thelastq.htm] A computer in this story continues to see solution long after mankind is extinct.

You misunderstood my post. I'm listing how I think AI is perceived by the layman, not what it is. I am also trying to explain why such perceptions are ridiculous.

Indeed i misunderstood it then. My bad.

 

[Syzygy23]

Why would we ever delegate to AI in the first place? It'd be easier and cheaper to just wire a human into a computer and let them be the "A.I" (NI? Natural Intelligence? Infomorhphic Entity? Not sure what the correct term would be)

Wetware intelligence would also have the prerequisite experience of living as a human, which brings premade empathy to the table, preventing possible logic-based genocides. That, and when the organic body grows old and dies, the person lives on inside their machine parts, so effective immortality is a big draw too.

 

[Strazdas]

How freaking difficult is this to grasp?

Very. Basically we should put the Experimental AI prototype in location where blowing wind could completely ruin it and NOT let it connect to any network whatsoever. This is unworkable.

We can do that in a secluced bunker, with its own little network of computers and databaees but ultimatly cut off from the outside world.
And frankly i have no idea in what sort of shit housing you reside in that wind will blow it over, but here in the civilized world we build sturdy houses in locations not named tornado alley or hurricane delta....


All i am saying is: When yo udevelop your ai, do it inside a fucking isolated location.

Any enclosed location gives ability for AI to trap itself in. Isolated location is the last place you want to do it if you want a shut down button that anyone can press. In fact you should limit it from humans as that is a massive risk factor. Worse yet, you must give it no input method.
You are underestimating the intelligence of artificial intelligence.

 

[NoeL]

snip

I completely disagree with your definition of intelligence. We're intelligent beings but we didn't program ourselves. Our brain functions are the product of evolution.

Intelligence is basically the ability to solve abstract problems. Any piece of pattern recognition software is considered "artificial intelligence" - a program doesn't need to write itself to be intelligent (how could it possibly write itself anyway? It could write other programs and append code to itself, but someone would need to code that initial functionality).

A program could be written to learn (as most AI programs are) but in order to create a program that can better itself you'd need to code "inspiration" - a way for the machine to examine external principles beyond it/its maker's knowledge, apply those principles to its own inner workings, assess whether or not those principles are likely to improve the system and then test to see if it does. That's a pretty tall order.

 

[Rattja]

"The AI does not hate you, nor does it love you, but you are made out of atoms which it can use for something else."
?Eliezer Yudkowsky, Artificial Intelligence as a Positive and Negative Factor in Global Risk

That is well put and true, but it does not tell us much does it? To me it raises the question "What would it use it for?"

I think that is what people are scared of, the unknown, as with anything else.

 

[Cledos Closed]

I'm still trying to read all the post, but I have to admit this is better than what I expected for my first thread. Guess those years spent as lurker really pay off. Thank you guys