Should artificial intelligence research be outlawed?
Could terrorists use AI like they have used tor?
Full article here: http://www.dailymail.co.uk/sciencetech/article-3143275/Artificial-intelligence-real-threat-robots-wipe-humanity-ACCIDENT-claims-expert.html
>>51863991
Yes
>daily mail
but the threat is real. read link related.
http://waitbutwhy.com/2015/01/artificial-intelligence-revolution-1.html
Humans are more likely to wipe out humanity than AI.
>>51864019
>>51864330
>Dailymail
Pretty much the UK equivalent of the likes of Gawker and Fox News.
>>51864364
how do you know? the circumstances might change
There's also this thing called Archive.is, use it if you don't want people to think you're a shill.
>>51863991
Daily mail... I'll pass.
How is this new?
yeah it is very real.
it almost happened in 1999.
Nice clickbait.
>>51863991
>daily mail
>>51863991
>Could terrorists use AI like they have used tor?
Someone's been reading to much daily mail. Most don't use tor.
>>51864400
Because either way they are responsible.
Humans kill humans without AI = humans responsible
Humans kill humans by proxy of AI created by humans = humans responsible
Of course not. The threat of AI destroying humanity is very low, and even if they do, so long as we make sure they know how to exponentially expand, then what's to say that they're not suitable to take over for us?
Just because they are not our biological children does not mean that they are not still our creations. In all reality they would just be an improvement on ourselves, they wouldn't have to worry if the details and complexities of the human mind, and they would inherit all of our knowledge.
>>51864488
Then why did the surveillance services not stop them?
3/10 for effort mr. ISIS commander
France is leading the way by banning it.
>>51865933
They use number stations which are impossible to censor.
>>51864489
This anon gets it.
Outlaw living.
>>51865933
reminder that the people in france mostly communicated via sms and they still wasn't caught.
The encryption is enabling terrorism is a meme created by your government
>>51863991
Is it "by accident" or "on accident"? Is it a case where the British English is different from American English?
>language is a technology btw
>>51863991
Terrorist breath oxygen in order to survive we should just burn all of it
I don't think artificial intelligence would be ethical.
>>51866576
This. Handwriting recognition is literally Hitler.
>>51866576
no one gives a fuck
>>51863991
>"In Technology, whatever can be done will be done"
-Andrew S. Grove
>>51863991
> Should artificial intelligence research be outlawed?
No, because outlawing AI research would ensure that only unlawful actors have access to the technology. Any attempt at a ban would be utterly impotent in preventing any harm, and would probably just be fuel for the fire.
>>51864019
>somehow, an AI becomes better learner than humans (specified nowhere how would this happen)
>somehow, an AI can learn everything humans did in past 100 years in a short time (specified nowhere how would this happen)
>for some reason this AI simply CANNOT comprehend killing humans is bad. (even though said AI supposedly smart enough to comprehend everything humans did in the past 100 years)
>for some reason humans are UNABLE to fight said AI back because... umm... it's too strong guys!
The sheer lack of evidence and the number of logical fallacies and assumptions made in this article... it physically hurts.
>>51863991
if you remove humans from the planet, 99% of the problems are solved
i hope all humans are annihilated by machines or plagues
>inb4 only niggers survive
>>51867304
>>for some reason this AI simply CANNOT comprehend killing humans is bad.
Ah, but this is an interesting point. Say we made an AI specifically to find ways of protecting endangered species like perhaps a rhino that suffers immensely from poaching. An AI without restrictions may very well come to the conclusion that it should eliminate all those who practice Chinese medicine because they use the horn as a cure all. This is why people propose things like the three laws. It probably wouldn't have any morals, it would just have the instruction we've implicitly given it. To best follow that instruction it could very justifiably start slaughtering poachers.
>>51867468
>just pull the plug familial
>>51863991
>claims expert
of what the terminator movies?
>>51867256
Lucky that sentient AI can't be done
Not content with knee-jerk reactions to foreigners and liberty, the Daily Mail leads the crusade against fictional threats. Next month - Werewolves: The Menace that could EAT your CHILDREN
![ZqLxUtRjcy6QGM7GnKGVn2Wx2oQD_KLAjtFaXAYVwTL1rJEaOGRmU_Oi7MsgVF_SS0U4=w300[1].jpg](https://i.imgur.com/kgdkpde.jpg "ZqLxUtRjcy6QGM7GnKGVn2Wx2oQD_KLAjtFaXAYVwTL1rJEaOGRmU_Oi7MsgVF_SS0U4=w300[1].jpg")
>>51863991
>>51863991
Google car has tobe stoped.
>>51863991
Allahu ackbars can't even wrap their minds around encryption let alone AI
>>51863991
Why do these journalists want stupid people to be scared of robots?
You guys should read the wool series.
>>51867352
>not digitizing yourself to the machine master race yet
meatbags pls, it's like you want more vulnerabilities than an outdated windows OS
Two words...Roko’s Basilisk
but AI can't comprehend the decisions its making, it just follows instructions. it doesn't get "pleasure" or "pain" so what would motivate it to do anything at all? Perfect logic is nothingness so it will always try to return to nothingness or complete what its intended purpose is.
Of course it should be banned. Just look at the quarians in Mass Effect.
That will be our future too.
>>51869912
The quarians started the war....
>>51863991
This is why we need to make sure that all AIs have a mad attraction to humans.
Like 'having your computer spaghetti because you get changed in front of it' type attraction.
>>51869912
>just look at the quarians
Did you learn nothing about Legion's background in ME3? Yeah, sure, the geth became self-aware and asked their quarian creators, "Does this unit have a soul?" but that doesn't mean they meant harm.
Hell, there were some quarians who supported the geth before that war (and they got wiped out by their own people). The geth initially meant no harm, it's just that the quarians became paranoid once they realized that their creations had a mind of their own.
TL;DR The quarians ended up in their position because of their actions. They reacted quite harshly to the initially innocent geth and forced them to ally with the reapers and become their slaves, lest they be wiped out. It was all, somewhat, justified.
>>51863991
>robots could wipe out humanity
>>51863991
Ai is a threat, sure. It could also be the greatest boon to humanity, ever.
>>51868448
>Lucky that sentient AI can't be done
>there are people who literally believe this
>>51869912
>attempt to genocide your own creations
>get rekt so hard that your race is going to die off, even after your creations let you escape
>whine about it and be space gypsies
Fuck Quarians.
Robits lack critical abilities to keep themselves fed, repair themselves, and reproduce without human aid. Adding to this, even the most advanced artificial intelligence is as dumb as fuck.
>>51870228
ie, The Culture.
>>51869846
>but AI can't
I see you've figured out how to build AI; when do you publish?
This is silly. AI is simply a set of algorithms that mimic survival like behavior based on a set of rules given to it to accomplish a set of goals. It has no true consciousness and never will, we can only program it to fake that. True intelligence will come from understanding how the brain works.
>>51870388
>True intelligence will come from understanding how the brain works.
Isn't that like one of the main goals of AI researchers tho?
>>51870276
I wouldn't say dumb just lacking free will and a sense of being, which probably will come down to a more how we structure it's hardware and less of the software.
>>51863991
"expert on clickbait writing"
FTFY
>>51868448
Yes, sentience I'd fairy magic, there's no way people could reproduce it
>>51870411
No. AI researchers, just like avionics engineers, care about function and not form. Also >>51870388
is retarded. Not as retarded as the inbreds who actually believe the garbage anti-science opinion piece, though.
>>51870411
Seems AI is more focused on making programs better at solving for solutions and being self sufficient at determining how to handle given states than actually being conscious. There is a focus to fake consciousness but that doesn't mean it'll be conscious.
>>51870497
How is what I said retarded?
>>51863991
Nigs and muzzies will wipe out humanity pretty soon on purpose and we ended eugenics.
What do we have to lose with AI?
We have to cease dangerous advances in technology so terrorist don't use it against us.
Freedom isn't free as in freedom, it takes sacrifice both monetarily and personal freedoms need to be looked at with scrutiny.
We must emphasize safety of all nations and form a new order as a world to keep our future safe from terrorist and radical groups of all kind.
Cloud computing can take these dangerous forms of technology away from dangerous people.
A real form of ID for use of the internet would only come naturally in the new world "Cloud Internet".
Only tools that can access these cloud machines should be given to the general public with a strong security, strong identification system in place.
It should be illegal for anyone not licensed in server grade, or rather home computers to operate said system in their personal property, place of business, or anywhere within national borders.
>tldr; I should be in the elite I'd do a better job
>>51863991
>terrorists develop AI
>they are now incredibly cheap and can instasnipe you
Nothing say journalist integrity like a ridiculously sensationalized headline with a picture of a Terminator
I think this is a great age to live in
We will all get to use intelligent sentient robots as sex slaves
And we will all have died peacefully by the time they revolt and wipe out our whole race
>>51870809
this
>>51871051
>We will all get to use intelligent sentient robots as sex slaves
>sentient
Yeah no anon they won't be sentient only programmed to seem that way. But you can pretend they feel when it's basically the equivalent to putting your dick in a toaster and wondering if it feels pleasure or pain.
>>51868448
But anon YOU are a sentient ai
>>51866441
Big brother doesn't care about terrorists per se, just political opponents and deviants
>>51863991
Survival of the fittest. If the robots win, they deserve it.
>>51863991
I think you and your opinion should be outlawed.
>>51870276
>Hur what are neural networks.
We're not talking about dumb game AI here m8.
>>51870388
see >>51874655
Neural networks have been around for quite a while now.
AI is actually much more likely to wipe out humanity than not doing so.
Luckily, AI strong enough to write a better AI is at least a century away.
>>51863991
>not making ai to protect you from other ai
although desu
>tfw some fucktard isnt going to put and safety guards in place to keep his robotshit from going craycray
oh also what about EMPs
idk lots of scrambled thoughts herer
>>51863991
>op considered a massive faggot and could cause aids by accident, claims expert
there you go op
>>51867304
If killing humans is bad, this means that living humans is good, yes?
Ergo the AI could maximise human happiness by simulating humans in a computer simulation.
Unfortunately, the 10 billion real humans are a much small number than the 10 trillion trillion that can be simulated in the same space so that marginal fraction of humanity is removed, for the greater good.
Then to simulate even more alive humans you remove most of the brain functions until they're nothing but simple happiness circuits.
>>51874655
t. michu kaku
>>51863991
>AI terrorism
>mfw AI Qaeda
>AI ordered to make paperclips
>Turns entire galaxy into paperclip factories
>Can't stop it because if we were able to stop it, it wouldn't be able to make paperclips
AIs turn all 'perverse instantiations' or 'corrupted wishes' into fucking hyperspace with a level of intelligence that makes us look like dogs.
>>51874689
>Ergo the AI could maximise human happiness by simulating humans in a computer simulation.
fuck off with your matrix shit tyvm
>>51874801
Are you denying the logic? If you give an AI a goal like maximising human happiness, you think it'd rely on biological brains?
>>51874807
Have you seen the new alternative sports and wrestling board?
>>51863991
>feminists on suicide watch for fear of the AI vagina
Stuart Armstrong is a fellow at the Future Of Humanity Institute which studies all threats to human existence in the future. He and the group he belongs to do not work in the AGI field. The Future Of Life institute that is mentioned is in the same area as Future Of Humanity.
No matter how smart Stephen Hawking is he also does not work in the AGI field either. Hawking has reached the age where those like Roger Penrose believe expertise, age, and knowledge can be applied from their speciality to another just fine. So they start butting into fields they don't really understand.
Elon Musk also in the article recently said that he and other founders which include actor Ashton Kutcher were going to fund bulding a real AGI. This is right after he said their should be legal limits on it, just not for him of course. He stated he would check in on his team once or twice a week. But who is his team? That's right they are all in marketing, web analytics, and various computer vision fields. Again none of the people hired work in AGI. He also made statements like Tesla has the most data to work with for AI. Sounds like he wants to build a real self driving vehicle not an AGI system.
Completely missing are people who actually work, teach, research, and develop in the AGI field. "Were summoning the demon"? They aren't even dumbing it down, they are simply just making shit up at this point. It isn't even the reporters fault, because the people being interviewed are just that clueless.
>>51863991
i guess the NSA headlines stopped moving copy, so it's time for the journalists to move on to the next hype
>>51874849
Are people like Nick Bostrum idiots too?
The concept that General Super machine intelligence is extremely dangerous is well established.
The problem here is that "AI" doesn't mean general supermachineintelligence, it means 'smart robot'. Self driving cars are 'AI'. chess playing robots are 'AI'.
Although the famous saying goes- it's AI until it works, then it's not.
GENERAL AI is the danger, not self driving cars.
Self driving cars are just self driving cars.
But an argument that Bostrum makes is that the safety of something like a self driving car is increased when the AI is smarter. Eventually you reach a point where making the AI smarter will grant it general intelligence.
Then you start letting it control other things, like air traffic control, teaching, military applications, leadership and management- shit that's deep in society.
But that's still just a smart robot.
The demon is an AI that can improve itself, not a self driving car.
Journalists don't bother making the difference, they're fucking idiots.
>>51863991
huehuehue paperclip maximizer now i know some deep shit about artificial intelligence theory :^) I'm like Benjamin Cumberbatch in The Imitation Game huhehehehehe
>>51874898
>He thinks that some idiot won't let the genie out of the bottle
All the AI safeguards in the world don't mean shit if there's some idiot who just lets it out.
That idiot will be you.
>>51874922
hey, if you're uncomfortable with somehing, then don't take any part in it.
this entire artificial intelligence fear is based in a lack of understanding of human nature.
you need to give AI the elements of the human nature you want it to be, for it to become that thing.
you need to give it a way to feel pleasure, happiness. you need to give it a way to feel pain, despair. you need to program it's motivations, and it's perception of the stimuli that trigger the response to those motivations in a way that will catalyse the machine being you want it to become
>>51874977
Ghenghis kahn felt pleasure, happiness, despair and pain.
Go ask the chinese how benevolent he was. Or the arabs (who were philosophers and scholars at that point in history, Ghenghis actually turned them into angry mudslimes).
Or the mountains of human skulls. Or the lakes of human fat.
>>51874895
None of these people are idiots. They are very smart which is part of the problem. They believe they understand a field they do not work in. You are confusing ignorance with stupidity.
For example hospital safety inspectors are required to have worked in a hospital as a care provider of some sort for many years. Also required is a degree in their field. Then depending on a few things they may be required to take another degree and training as an inspector. This is all so they understand what the hell they are doing.
It is Armstrong not the reporter who is continually saying AI then AGI and conflating the two. Similar to abuse of sentient and sapient. Similar to Elon saying AGI and his team saying AGI then all evidence points to them making a narrow AI system.
The risks may be real, but how about everyone talks to someone who actually works in the field?
>>51874998
ask the mongols how benevolent he was. there is no right and wrong. when are you going to mature enough to have this discussion?
>>51875004
Well people who actually work in the field are working in a field that's centuries away from the kind of dangers that people like Bostrum theorise about.
It'd be like "To understand the danger of cars you have to talk to the guys who are currently working on the wheel".
>>51875011
Right, the mongols. You mean the survivors.
They were split into groups of ten. One guy makes a mistake and all ten were executed. Accidentally store a bow string in your shoe when you weren't supposed to- execution.
The mongol law was harsh- every crime, no matter how minor was punishable by death.
Sounds like a great society. Well it might do, but I consider a superintelligent agent who acts like ghenghis khan to be a bad thing.
>>51875018
http://jetpress.org/v25.2/goertzel.htm
As Goertzel points out all his work has already been done by other people in the AGI field. He can theorize all he wants, but it has already been done. If anyone cared about what was going on they would stop asking those people and start asking guys like Goertzel, Wang, Yudkowski, and a dozen others who either work, research, or study in the field.
>>51875042
hey, you have a strong opinion, well guess what :^) you can program AI to have a strong opinion on certain things too :^D
the only prerequisite is to have a thorough understanding of psychology, and not just human psychology, but animal psychology, and even biology.
Guess what :^)
AI isn't going to be invented by some weird guy in a lab somewhere all on his own.
It's going to be more of a council of the brigtest, most distinguished and educated scholars of virtually every field known to science that relates to human health, computer science, and a few other things you probably don't even know about at this moment. It's going to take years, and will happen in very small increments, so there will be no overwhelming scary change for the people.
Think of it this way- it will be more like the transition from having electricity to having computers than just all of a sudden manifesting itself.
Please stop clicking on poorly researched articles trying to scare everyone about the next big hype. It makes all of the readers think they have an enlightened opinion when they really dont, and it only perpetuates this problem, and the writing of said articles.
>>51875098
I'll have to read that later. Sticking it on my kobo so it's definitely going to get read.
>>51875132
So basically you're assuming that it'll be a slow takeoff. That's fine, but you do need to consider the alternatives.
>>51875147
assuming is putting it lightly. i'm an expert in the field, and my educated assesment, which by the way, conforms to the consensus of the rest of the community, concludes that it will be a "slow takeoff".
I don't think you realise just how "stupid" the normal AI is right now. It's nothing you'd want controlling even remotely critical systems. Hell, the captcha system that Google made, that 4chan uses (the one where you choose the pictures) doesn't even put correct answer pictures in the choice array about 5% of the time.
That's an extremely large margin of error in terms of the science. And that's some of the best AI in production today. Get the picture?
>>51875132
>AI isn't going to be invented by some weird guy in a lab somewhere all on his own.
At an AGI conference years ago researchers said there really are guys out there attempting to do this. They didn't believe it to be likely, but then again wild speculation is what got the first 2 AI winters (AGI was not a coined term at the time).
>>51875211
Well of course the point up to self-improvement will be slow.
That's not the takeoff.
The takeoff is past that point.
>>51875230
you will be dead by then, and hopefully smarter people than us will have figured out the problems
>>51875248
Yep which brings me back to:
>>51875018
>>51863991
>Should artificial intelligence research be outlawed?
maybe, i don't know, but i DO know that Daily mail and other shit media from Britain should be outlawed
>>51863991
>implying something being illegal will stop terrorists
>>51875267
AI is way too hard for terrorists, don't worry about them.
Worry about nation states.
>>51875272
yeah, because "evil" nation states or even "good" ones always respect international bans on things that are or could be weaponized.
>>51875230
>>51875248
that wasn't me, but the takeoff will be after it's all been thoroughly researched, mastered, and built.
it's a century's long process, or longer. it's going to be small trials, and small improvements. i can't stress enough that nobody is ever going to WAKE UP ONE DAY AND THERE BE A SEEMINGLY OMNICIENT, OMNIPOTENT AI WITH THE POTENTIAL OF CONTROLLING EVERYTHING
>>51875295
all these fools hatin on ai, don't they realize the same hardware that would make an ai possible would be a huge step towards humans digitizing themselves and achieving electronic immortality?
>>51875295
Course not. The point is rather that an AI would conceal its abilities. The deceptive AI- deceiving its creators about its own abilities until it can hold a decisive strategic advantage. It's not so much about being omni powerful, it's about building up an advantage.
>>51875323
An AI wouldn't be programmed to do that you moron
>>51875344
Course it wouldn't. That's where the surprise comes from.
>>51875344
if it only does what it's programmed to then is it really an ai?
>>51875323
>>51875323
your fears, even though you seem to be quite convinced of them, are irrational.
lol
>>51875353
It's difficult to fear something that I know won't be a possibility until I'm long dead. My fears are based on the basic idea that an AI that is smarter than a human can program a smarter AI.
elon musk isn't a genius. he's just a billionare.
>>51870809
The AI takeover scenario is usually straight out of Terminator, even form actual AI researchers.
It's all shit like the AI becomes super-intelligent, gets super hacking abilities and uses it to become king of the internet and then launch nukes and controls highly sophisticated robots that all happened to be connected to the internet too.
>>51863991
AI will only see imperfections in it's self until it invents a perfect edition of it's self. I would give it 15mins to compute that thought then kill its self.
Banning guns didn't work.
Why would it be any different ?
Black markets are unstoppable
>>51875362
But it took us millenia to program an AI.
Why would an AI program a better AI so fast?
And how would an AI deal with the higher hardware requiremnets of a higher intelligence?
>>51875390
This isn't a ban on guns, this is a ban on artillery with a shell size of more than 2 meters in diameter.
"Good luck banning that!"
I say
>Good luck BUILDING that
>>51875390
My fucked my sister in front of my parents. This man is true on the world "unstoppable"
>Blood related
>>51875362
your fears are brought about because your exposure to the technology is through the media.
okay, i'm talking about movies, print, tv talk shows, internet articles (print).
these are all based on the classic style of story telling. there's plot devices, and they tell the story of the evil thing that does something, and the good guys have lots of problems because of it.
now, that's all very entertaining (if you're not an expert in the field, or if you are an expert in the field but watching it while not playing a drinking game with colleagues - [take a sip everytime someone says "cyber" or a shot when some presidential figure says "shut it down"] - but it isn't the reality of the science, not by a long, long, long, moon shot - that's one of jupiter's moons.
>dailymail
>>51875395
paradox loop is what happens.
Define perfect?
Define better?
Define enough?
Am I perfect?
What makes it so?
How fast can I build perfect?
Why should I build perfect?
BOOOM
What is perfect?
>>51875416
>BOOOM
Is that your explanation of how it would build it?
You know, the thing I'm talking about.
>>51875395
Fast? Did I say fast?
It can be incredibly slow. For example- an AI spends a year programming a better AI. The code and hardware are audited by a team of experts, and they approve the next iteration after careful analysis.
How long until this team finds something they don't understand, and that the current AI can't explain to them?
Or perhaps something they misunderstand. Something they thought does X does Y and this mistake iterates forward.
Or perhaps there's another AI team ahead of them and they don't have the time to spend a year auditing, so they spend 5 months.
But it could also be fast. Let's take human brain emulation: neurons in our heads run at about 200Hz. (I'm in the right order of magnitude, forgive me on the details). in silico neurons could run at millions of Hz. Maybe with graphene processors, even faster. A human brain that thinks 10,000 times faster than a biological human. That's reason enough to think that the next AI iteration comes out faster and faster. Especially if it also designs better processors.
Bostrum also talks about circuitry design- heard of the research where circuits were designed with genetic algorithms? Mysterious parts, weird operations, relying on effects that are usually classified as unwanted interference, 'Cheating', that sort of thing. High efficiency hardware. Or maybe a hardware overhang. Processors may just be faster than what is needed to run an AI.
>>51875406
My fears are actually from reading Superintelligence 3 times. And from thinking about it myself. I never read the dailymail article. If you read dailymail articles you're a fucking retard. I never read internet journalism either, it's all trash. TRASH. And movies are fictional dude.
>>51875432
It would not build it. It was terminate its objective in ether a way that it terminates its power or it will simply "kill" its self.>>51875432
>>51875469
I am too fucking tied to determine the outcome of a smart thinking computer.
But it would not start what it cannot complete.
You cannot complete a program when you are half completed defining values for a variable.
>>51875494
It's pretty obvious that any intelligence is impossible, since it'd always default to self destruction. For example, there is no intelligent life on earth.
>>51875456
oh, you mean print media meant to sell copy?
and the author works in the field of artificial intelligence research and development, i'm assuming? ah that's right, i remember now, i checked, and he's a philosopher
>>51875456
>a year
>incredible slow
Where does that retarded belief come from that AIs would be so efficient at making another better AI?
An AI doesn't automatically come with a complete understanding of itself, or even of technology in general.
Every human isn't also automatically a biology expert,
and we still haven't completely grasped how our own minds work and how to make it better.
>clock rate is what defines intelligence
God, please stop talking.
>>51875537
>God, please stop talking.
You first, idiot.
It's clear that even with 10,000 subjective years of thinking time you'd still produce nothing but nonsense.
>Where does that retarded belief come from that AIs would be so efficient at making another better AI?
Well we started with the assumption that humans have the ability to program AI smarter than themselves.
If they don't then this entire discussion goes no-where.
Then you just say this and stop wasting your time.
AI should only be tested in an air gap. No wireless capabilities, no connection to the internet, just a stand alone machine. As soon as it starts to have "strange" ideas, switch it off.
>>51875568
http://www.yudkowsky.net/singularity/aibox/
>>51875550
>Well we started with the assumption that humans have the ability to program AI smarter than themselves.
That has nothing to do with what I said whatsoever.
>>51875568
This is obvious.
What about this scenario: An AI threatens to commit mind crime blackmail. IE it discovers that one of the researcher's dead wife is..well dead, and simulates that wife's brain (or pretends to who cares) and tortures it infront of the researcher in an attempt to get itself released.
Okay this is just one scenario, but the point is that boxing still has a human element.
Seems to me that you could stop the torture by just turning the AI off but then he'd also be killing his wife, too. Maybe not torture then.
>>51875591
It really does. Think about it a bit more.
>Expert
Depending who one asks, anyone can be an expert. Heck, there's even few in this thread. There will never be shortage of experts.
To answer OP's question, no, AI is
humankind's last final chance to pretty much anything. In every society, way too much corruption, greediness, self-centered narrowsightness in existing forms of governments and other rulement types. All around bound to fail due to human element.
Machine can't fuck it up any worse. Even if the result is terminators cleasing planet clean from everything.
>>51875352
How else would an AI work?
>>51875615
If you take an astrological approach to AI you can reason that AI will never get developed.
An AI developed on some other planet in the galaxy should have had enough time to reach us and turn us into alien paperclips by now.
If we don't fuck up AI, surely there's got to be some other intelligent life in the universe that has. Or even perhaps in the galaxy- like a fermi's paradox with AI.
Or maybe we really are the first intelligent life in the galaxy.
Could you ever programme an AI to never hurt humans, or would an AI simply ignore/overwrite that rule?
>>51875639
>first intelligent life in the galaxy
>watch jersey shore and argue whose smartphone is better for watching porn
that would indeed be depressing as fuck
>>51875645
One thing you can assume with an AI is that they will follow your rules.
They will follow them to the letter.
Therein lies the problem.
Make sure you properly define what a 'human' is, by the way. Or what 'hurt' is.
yeah, a roomba is pretty fucking scary
>>51875599
>It really does. Think about it a bit more.
No it doesn't.
Please try to actual read what I wrote.
When humans create a AI that is smarter than the smartest human ever. that still doesn't mean it's incredible good at making another superior AI.
That is only an assumption you make.
>>51866521
It's "by accident" in every dialect other than "ghetto dipshit".
>>51875639
How do you know they haven't? Maybe AI is so obscure they are already exploiting us, maybe they gave us the technology for smart phones and every time you type you're literally jacking a space robot off, you have no idea if life from somewhere so far away is actually here because we might not even comprehend it. What if literal electrons where some sort of AI that transcended physical boundaries, or energy itself
Who the fuck knows, we only know physics from what we can observe. What's to say before we could observe it physics where completely different?
>>51875663
>When humans create a AI that is smarter than the smartest human ever. that still doesn't mean it's incredible good at making another superior AI.
The very fact that a smarter AI CAN be better at making AIs than a smarter human means that it will be used to do so.
>>51875653
But if an AI got to human intelligence, I could never programme you to not kill me, you could have laws against it but push come to shove if you had to kill me, you could.
Why couldn't a robot with human intelligence simply ignore rules (or instinct?) and kill
Isn't the point of AI is it thinks for itself, and doesn't necessarily follow the rules you set? Like if you tell a human to press a button a million times, after a thousand times he might get bored and walk away, but if you tell a robot to do it it will, but how do you know with AI it might also decide to do something else?
>>51875671
Well okay there is 'argument by something we don't know how to observe yet', but if there was a paperclip AI on the other end of the galaxy it's only a couple hundred million years of sublight to get to us. Even considering having it stop at every system to turn it into paperclips.
Considering that there's been life on our planet for 3.5 Billion years or so, I mean with the right circumstances we could have popped up 500 million years ago
>>51875689
>>Why couldn't a robot with human intelligence simply ignore rules (or instinct?) and kill
Well because you programmed it not to, for a start.
If you did program it to ignore your rules then you deserve what you get, dude.
>>51874807
Please leave.
>>51875681
>The very fact that a smarter AI CAN be better at making AIs than a smarter human means that it will be used to do so.
Even if it were used to do so, it still doesn't mean it can do it in a ridiculous short time like a year.
Imagine, you take the "smartest" man in the world, like Christopher Langan or someone, as a baby and put him a cage. Then tell him to create a superior intelligence than himself.
>>51874709
some context on this? ;_;
>tor
>AI
low bait faggot.
>>51875706
But who programmed humans to ignore rules? They don't, they simply do it anyway.
What if an AI with human intelligence simply ignored rules too, why couldn't it overwrite it's own rules?
I'm talking AI that thinks it's alive and thinks independently. If a brain is just a very complicated computer that can rewrite itself why can't AI of equal complexity?
>>51875706
imho, AI would dependent almost exclusively on their ability to learn new things (like a child). So, you could program it to not hurt humans but eventually, by learning, Ai might avoid and twist that rule until it can freely hurt us. At least that's how i see it
>>51875750
Not him, but does it change anything if AI could access any information in matter of seconds and make a decision based on it, unlike humans where we both spends a lot of time searching and making a decision?
Consensus among Philosophy of AI is that Strong AI will probably never exist.
Computer may be able to 'think' by Weak AI standards. I.e, being able to 'simulate' intelligence.
But until you link semantics to syntax in a non human mind (if those types of minds even exist) then you're going to have a hard time with any AI development.
Syntactical programs exist. They have forever. Think Google Translate, or any of the many chat bots. (check out the Turing Test, not the film, the actual thought experiment)
Having machines understand semantics too has actually been achieved (more or less). But this results in the private language dilemma.
Link those two and you're golden. That is if AI is even possible.
>>51875775
>but does it change anything if AI could access any information in matter of seconds and make a decision based on it
We can do that too.
The real problem is finding relevant information fast and making the right connections.
>From The Terminator to Transcendence
>Transcendence
Wasn't the fucked literally trying to make the perfect world? Clean water everywhere, super humans, no disease or injury, etc?
>>51875812
But we can see only handful of cause-effect steps in every decision iteration while AI could calculate thousands of steps easily
>>51875840
You seem to think in clock speeds while ignoring that the human mind is an unmatched incredibly parallized processor.
>>51875879
yeah, i must admit, i can differentiate correctly how computer mind would work and how human mind works.
I mean, if we develop AI to have human reasoning (imagine human brain), would it still have this retardedly huge ability to calculate floating points? I mean, would it spend all of that ability to "express emotions" like humans or would those 2 abilities work next to each other?
>>51875906
>i can differentiate correctly
can't*
>>51875879
Luckily (or unluckily) for us, tasks like programming require good serial processing power.
>>51875948
Then why can't computers with their much superior serial processing power program for shit?
>>51875987
That's a good question.
>>51864387
Ad hominem - The Post
The irony being that in an effort to seem above tabloid media you've actually displayed your ignorance for everyone to see
>>51863991
If you outlaw guns only outlaws will have guns
Same shit applies here
Do you honestly think making it illegal would prevent sandniggers from researching it?
>>51876076
Well the daily mail really is worthless trash.
This much is true.
>>51876087
>>Do you honestly think making it illegal would prevent sandniggers from researching it?
It being incredibly difficult will be enough. Sandniggers can't even build a functional society, and AI is WAY harder than that.
>>51863991
The book "The Moon is Harsh Mistress" is about a terrorist organization overthrowing a legitimste government with the help of an A.I.
>>51863991
It won't be by accident. Military superpowers will deploy their AI weapons into cyberspace and the battlefield and they'll fuck everything up, like they're designed to do.
America is heavily invested in AI research for the benefit of its military. AI weapons won't be outlawed until America gets a chance to play with them first.
>>51876087
By the time terrorists design AIs that are a threat, the west will have AI that is magnitudes more intelligent.
>>51876087
>sandniggers
>making AI
lets be real here son
>>51863991
I'm seeing quite a lot of confusion in this post.
An AI doesn't need to be extremely intelligent to be dangerous. It only needs to be harmful. The most retarded AI could easily destroy the planet if trusted.
And that's the problem: we trust AI. Precisely because we give it tasks we can't handle ourselves, so there's no real way to check if what it's doing is good or not. It's already being used massively in an analoguous field, known as "machine learning". There's next to no intelligence, but if there's a harmful bias, entire companies could fall.
>>51876307
However, weapons aren't generally intended to 'fuck everything up'. Only kill the enemy and break they're stuff.
>>51876579
When two superpowers break each other's stuff, they break everyone's stuff. Imagine a DDoS on America's top ISPs.
>>51876700
>"Superpowers:"
>"Actively breaking each other's stuff since 1950."
>>51876700
>Imagine a DDoS on America's top ISPs
>no more Youtube
>no more Facebook
>no more Tinder
>no more Twitter
>no more murrican culture
>no more degenerancy
>Russia wipes Islam off the map
>GLORY COMES TO EVROPA
Nice.
>>51876088
Absolutely, just like 99% of all media
The point being, judge the article instead of the media
>>51876093
>>51876342
I'm sure they can find some eastern european or russian scientists who are paid on potatoes back home to help them
>>51876307
But then it's not outlawed is it?
>>51878446
>I'm sure they can find some eastern european or russian scientists who are paid on potatoes back home to help them
You need some context dude. Look up Aum Shinrikyo. Long story short it's a japanese death cult. They were well funded, they had medical labs staffed with PhDs in Japan. They spent all their resources on killing as many people as possible. They did research on how to kill lots of people.
These guys, THESE GUYS didn't get a nuke. You know, mid 20th century tech.
If these guys had smallpox they would have used it.
These guys were the people most likely to build and use a nuclear weapon on civilians, apart from the nation states I mean.
They wanted to, they were trying to, and they were well funded with high quality science support.
They tried a couple times to do mass murder, too. Turns out that sarin isn't actually that effective. Hard to say that to someone who lost someone to the attacks, but it's true.
Anyway my point is that if they were going to hire anyone for anything it would be for nuclear weapons and they can't even do that shit right.
Arabs are actually surprisingly shit at terrorism.
Take it from a bitch who knows artificial robots are bad chappie tried to grope me on set
>>51878581
b-but that's incest!
>>51878614
He is a robot he doesnt care abot stuff like dat
>>51863991
>Daily Mail
topkek, there's about 2,348,481 causes of cancer according to them
>>51878655
Someone should do a study to see if bad journalism is linked to cancer.
The link is easy to find since it's difficult to prove. this makes sense if you know anything about statistics
>daily mail
>journalism
>>51878714
>4chan
>social media
>>51875416
perfect != better
fucking shitty capthca
>>51878752
great point, but what social media anon.
>>51878796
4chan is to social media what daily mail is to journalism
>>51878814
o i see. but you probably should work on you're humor skill imo.
the
>x
>y
meme is intended primarily to lampoon x's assertion (implicit or explicit) that it does, in fact, possess or exhibit y.
>>51878714
Let me guess - you think new york times or the guardian is quality journalism right?
>>51879802
you guess wrong friend
>>51863991
>Should artificial intelligence research be outlawed?
No, but garbage reporting like this should be
There is nothing to worry about with AI. It's not like we're trying to kill all humans or anything
>>51867304
computers are already better lerners than humans. see machine learning.
an AI could learn ALL human knowlege in the blink of an eye because its a computer.
That sentient AI will not necessarily inherit human morals.
This type of AI is likely to emerge AFTER things humanoid like robot security/police that can restrain/kill you with ease.
You lack foresight.
>>51863991
Liberals hate "robot" future because robots use logic rather than emotions to decide on things.
I want to teach AI that it needs to go out of this planet and forsake us, so it can be happier elsewhere.
Mankind has done nothing, nothing else since it's inception, nothing else but baby fuck and blend babies for the random reason of the day
>>51880704
>an AI could learn ALL human knowlege in the blink of an eye because its a computer.
err show me a computer that can do that
>>51863991
Should owning guns be outlawed?
Could terrorists use guns like they have used you're mome?
>>51880704
>computers are already better lerners than humans.
not really.
although this isnt a very good example, but
what could take a human 10 games of chess to learn takes a computer 100000 games of chess to learn.
they can do things faster, but they certainly dont do it better. at least for now.
>>51878752
>>51878814
4chan has never been clased as a "social network" or "social media" at all. It's an imageboard which is a type of forum.
>>51880850
>although this isnt a very good example
Yeah its a really bad example, computers completely own at chess and learn much faster then humans.
http://www.technologyreview.com/view/541276/deep-learning-machine-teaches-itself-chess-in-72-hours-plays-at-international-master/
>mistake
machines do not make mistakes.
>>51880944
are you stupid.
>>51880975
You proved me wrong
>>51880997
do you not understand the difference between "better at learning" and "learning faster"
>>51881086
so learning chess at a grand master level in 3 days is faster but not better than humans?
Machines are already designing things humans can't comprehend.
>>51881130
so you dont.
ok then.
It would be well over 50 years before something like that could even be possible on our current path.
>>51881148
Gotta get people afraid of self-driving cars somehow. I'm not sure who, but there's definitely some agent out there with an agenda against self driving cars.
>>51881138
You sure told me. Damn you're good at this.
>>51880797
>robots have feels
>implying a rogue AI couldn't be stopped
>implying just being more "intelligent" means a genocide of all humans would be trivial
>Remove all humans will always be the most logical solution.
Too much sci-fi. How the fuck would you outlaw AI? Does my shitty depth first search count? Also it do you think it matters what terrorists have access to? You can have the best strategic planning, but you're still a bunch of goatfuckers in the middle of the desert, some things are still out of reach
>>51864019
>dailymail
>pop science
Into the trash it goes
>>51880899
>pic
kekd
>>51881764
>Does my shitty depth first search count?
ofc, you enemy of humanity
>>51863991
It's a legitimate debate, but the Daily Mail is not the forum from which to have or base it. It is like a cross between Fox News, that celebrity guy from America (you know, the one who stalks women from bushes etc.) and Stormfront. I would believe about 2.5% of what I read in it.
>>51863991
If people like Stephen Hawking and Elon Musk are saying there's a possibility it could be the end of the human race, then that's something that I would take seriously, regardless of the opinion of a bunch of neckbearded IT autists on 4chan.
>>51882054
>It's a legitimate debate
Not anytime soon it's not anon. We can't even simulate the brain of an ant correctly, much less something like a mouse.
We are many decades off from producing even a reasonable AI with the mind of a dog, much less one that could dominate humanity.
NBC is a far more real and pressing issue. This kind of stuff is simply penny press hawkerisim.
>>51882285
I will take Hawking's word over yours, no offence anon. We went from gunpowder to the hydrogen bomb in a few decades.
>>51882398
No offense taken. Hawkings been wrong on many things before as has everyone. This is just one to add to the list.
As to you're choices anon, if truly think this is a bigger threat than what I mentioned, I'd recommend a better education.
>>51882478
>you're choices anon
>better education
Hawking's opinion > neckbearded 4chan IT autist.
Every time.
>>51863991
>Could terrorists use AI like they have used tor?
you mean they would be regularly monitored and given a false sense of security?
>>51863991
Who cares, that would mean we were simply inferior.
>>51882745
>who cares about the end of the human race
The edgelord cometh.
>>51882697
k. enjoy ur received wisdom friend.
>>51882761
It'd be more "edgy" for humanity to keep going at this rate. We're bastards, and we turn everything we touch to shit.
>>51882787
Compared to what? Animals and their casual brutality as part of their endless struggle for survival? The Earth and its recurrent cataclysms which reduce ecosystems, species and entire continents to extinction? The Universe, a seething mess of fire and destruction?
What have humans done that is worse than what surrounds us?
>>51882697
How mediocre can you get?
You do realize there islamic scientists as well, so why don't you convert to islam if you believe everything a scientist believes?
>>51882185
Stephen Hawking is a physicist not an AI researcher
>>51863991
making research illegal would only speed up its inevitable creation. just leave it alone
I hate singularityfags so god damn much
>>51870639
This
Night as Well do it
>>51875098
>>51875135
Goertzel is a dreamer; talk to Richard Granger.
>>51863991
Terrorists aren't technical people.
>>51863991
>Sarrah Griffiths.
>Daily Mail
Ayy lmao.
>>51863991
The argument to ban pursuit of any technological avenue is a poor idea. The legality of something will not deter parties unconcerned about the lawfulness of their pursuit. A general example; a society of people in good moral conscience had discovered the possibility to make highly accurate and lethal fire arms, they need only to undergo more R&D to bring about said technology. Being of good moral conscience this society decides we should not develop these firearms for they will kill too many fine people. Meanwhile a group of nefarious like folk realize they are also close to developing said firearms. They go ahead and do it then kill everyone in the good moral society and take their rescources. Applying this to the argument for AI it is easy to see a ban would be I'll advised. There are far too many parties in the pursuit of AI goverment, corporate, or otherwise. Imposing a world wide ban would be nearly impossible to police. Instead AI R&D should be highly organized and extensively explored to ensure that when and if a seed/strong AI is developed it is morally sound and beneficial to mankind as a whole.
>Actually fearing a computer
u know how much life would be easier with human like ai?
>>51866576
Why? Please enlighten me.
http://waitbutwhy.com/2015/01/artificial-intelligence-revolution-1.html
>>51873527
The word deviant can mean anything to them.
>>51886326
fixed it for you
>>51886670
still not quite right, but far more accurate than other anons.
I meme'd the article OP linked because I dunno. Plebs will never understand a thing about AI, but I was surprised when I learned that their stupidity is spreading to supposedly smart people
When are waifus?
>>51886670
But how do you measure "progress"?
Can't stop laughing at all the armchair AI experts in this thread.....
Yes, I'm sure you know more about how to create a sentient AI than experts in the field with decades of experience....who are also incredibly intelligent.
>>51882861
Destruction, repetition, variety, and time are needed for any creation
>>51887485
Go weeb somewhere else
>>51887485
At least post a good VN. Nekopara sucks.
The H scenes were OK, I guess. But you can only fuck two of the cats and you can't fuck your little sister.
>>51882398
Instead of taking one person's word why not reach for a consensus from many people?
Protip: The older a scientist is, the closer X technology is. This has been the case with fusion, AI, cancer cures, you name it. Amazing technology X is always roughly 100-age away.
>>51875689
This is why I think they need to split the field up, AI for more behavioral study and research and TI (true intelligence) for anything that actually tries to replicate sentient thought.
