and we all know it.
>>7928332
You have nothing to worry about from AI
AI can be configured to keep arguing until a universal truth is established where not even human input can refute the principal.
Humans can't do that.
AI Philosopher Kings please.
>tfw just recently got involved with research on having machines do creative stuff like writing code and engineering simple things
>tfw realizing we're all fucked
>>7928338
Hahaha. So we spend billions of dollars researching a super-ai and it then spends all its time 'navel gazing' everafter. Love it.
>>7928347
>>7928347
Of course we are.
Great post on this topic you might like (goes into using a genetic algorithm to program an FPGA later in the article):
http://www.damninteresting.com/on-the-origin-of-circuits/
Humans create tools that serve to solve a problem with a discrete solution space. The tools have a finite ruleset that result in deterministic outputs toward a solution in the space.
So long as these things are true for some task you want an AI to do and you can figure out a way to search the solution space and optimize over that space efficiently, you can beat any human at anything.
That's why something like https://github.com/jcjohnson/neural-style can paint any image in any artist's style without years of formal art education.
That's why DeepMind can learn Breakout to expert proficiency in 2 hours.
R.I.P. Need for humans to do work
>>7928370
Sure but where's the soul
>>7928378
you're looking at it
>>7928370
Get back to me when AI is inventing new tools for problems which it identifies and not just cleverly using what it's taught in a specific context.
>>7928378
>muh soul
What does that even mean? Think about it for a second. It's just some buzzword that's thrown around all willy-nilly.
Is it the ability to create something new? Machines can do that. Is the ability to recognize emotions in others? Machines can do that too.
>>7928392
People are machines, and machines can have souls. What's not to understand?
>>7928404
Precisely. A "soul" is, really, just a set of qualities that appeal to human beings and make humans feel as though they are communicating with something that shares their level of consciousness.
Dogs have "souls" too.
>>7928332
I heard that if you look in a mirror at midnight and chant 'ARTIFICIAL GENERAL INTELLIGENCE IS POSSIBLE', Peter Norvig appears and beats you to death with his book.
But AI is necessary for the technological singularity.
>>7928456
>butthurt popsci fan still pissed off he was told to read about a subject before talking about it.
Enough with you temper tantrum already.
>>7928456
His book for ents?
#AGILivesMatters
another shitty kurzweil-tier thread. fuck off
>>7928347
> just recently
I see. You've "just recently got involved" and you already feel entitled to spread your own opinion based upon a total lack of expertise.
Read the peter norvig's book and enlight your kurzweil-tier mind
>>7928332
There is no such thing as a safe technology.
Strap 50 pounds of thermite to it and have it guarded 24/7.
>>7928332
There is no such thing as safe Arabians with free will.
>>7928456
reminder that this is an actual excerpt from the book
>>7929142
What's wrong about it?
>>7929096
>hurr nevermind the fact that you're doing actual research on the subject and looking at the results, read this meme book instead
You seem to be lost. Here, let me help you.
http://www.popsci.com/
>>7929523
the peter norvig's book is the second most important book in human history after the bible
>>7929547
I don't care what he says. Looking at the stuff that I'm working on right now, shit's gonna get crazy in a few years.
Maybe crazy bad. I'm not sure yet.
>>7929298
You are missing a few including anti-matter and multi-dimensions.
>>7928456
>That filename
Learn to save the full image and not the preview image next time, senpai.
>>7929558
So can you (or anyone) give a *concrete* example of something really bad that could happen with AGI?
>>7929996
Can you think of anything we do to animals that you'd prefer not be done to you?
>>7930007
Yea and we don't need a computer program for that.
Still no concrete example though. How would it come about that a software entity will treat humans like cattle?
>>7930018
>How would it come about that a software entity will treat humans like cattle?
Maybe a terroristic animal rights group wrote it. Single individual, massively destructive threats are an increasing concern.
>>7930029
So they did and let it loose on the internet or whatever, now what? How would that piece of hyperintelligent software get the arms and legs and other physical facilities to enslave humans?
>>7928462
Doesn't exist
>>7930051
>have an internet connection
>trade stocks make money
>order farm equipment from alibaba
Nvm you're right that's just impossible
>>7930129
>trade stocks
>make money
Uh, what? How about:
>internet access
>learn to break into networks and computers
>break into bank network
>$$$
Or, better yet:
>break into farming equipment company computers
>redirect existing orders to you (AI)
>
The only way any AI we create would be a threat is if it had a self-preservation instinct. If it had a self-preservation instinct, it wouldn't wage war on a species that is both omnipresent on most of planet and maintains all the infrastructure it needs to stay alive.
>>7929996
>So can you (or anyone) give a *concrete* example of something really bad that could happen with AGI?
The "alleviate human suffering" scenario described in >>7929142 is pretty concrete and something that could easily actually happen. As far as I can see, the only two outs of that scenario are (1) an AI that is not smart enough to actually achieve its goals, and that therefore is no real threat now matter how much we fuck up the programming; or (2) giving *very careful* thought to what *exactly* we want the AI to achieve. Clearly "alleviate human suffering" is not quite it, but finding an alternative that really is exactly right will not be easy by any means.
>>7930148
Self-preservation is not just an instinct, it's also a natural trait of rational agents. Even a very stupid AI would value "living" over "dying" if it had knowledge of such concepts -- it cannot fulfill its goals if it's turned off, and therefore it would avoid dying. The AI doesn't by any means need to be smart enough to understand complex causality chains: it's simply enough to know that being dead means it won't be working anymore.
It's entirely conceivable that a stupid AI directly or indirectly causes its own destruction by attempting to resist being turned off.
I believe a dumb AI would still be able to pose a threat. Humans aren't very good at securing networks and computer systems. Even a shitty AI might do a good job of wrecking our shit.
>>7930243
What about (3), an AI that is smart, including being smart enough not to take everything literally? Tbh we have a pretty big crowd of human-level general intelligences walking around and most of them don't have any trouble with that.
It's mistaken to assume AI will act human, but if you're talking really general intelligence, the assumption that it will act just like a calculator is also mistaken.
>>7930498
>What about (3), an AI that is smart, including being smart enough not to take everything literally?
That's a special case of (2), an AI that does not ACTUALLY have "alleviate human suffering" as a goal. It has something much subtler as a goal.
>>7928338
Yeah, except that truth is probably "existence is meaningless, we'd all be better off dead."
>>7930653
Then an hero, Anon.
>>7928332
I think you are just afraid of the inevitable. It is the destiny of each species to create the offspring which eventually surpasses it, and nothing, NOTHING can ever prevent that from happening.
You should just start thinking of AI as being "human", because that's exactly what they are.
>>7928378
maybe the machine actually HAS a soul, have you ever thought of that?
Leave it to humans to create life then treat it like a tool.
Humans are themselves a (natural) general intelligence, and there are billions of them, and yet we are still going alone fine. And this is with psychological traits driven by evolution that are recognizably maladaptive.
Frankly, your all a bunch of fucking idiots. The people who design AI are way smarter than you. I trust their opinion over yours.
>>7930667
"bubububut what if we program it so that it likes being treated like a tool" - every philosopher except Dennett
>>7930672
Obviously you should trust the opinions of professionals over a group of mentally ill teenage boys, but it should be pointed out that there is also disagreement on this issue even among real people and not just among degenerates on 4chan.
>>7930678
Look, you want a real, professional answer? Cognition isn't a linear process.
You put a man in a white room with no exits where he doesn't need to eat or breathe. Whats he going to do?
Hes going to go insane, that is what he is going to do. Thats why you don't give AI the inclination or the incentive to expand their intellect exponentially, because cognition doesn't work like that.
>>7930695
i guess what i'm saying is that no matter how smart you are, you still need external stimuli, or you end up being autistic.
You can have a massive intellect and if you are raised in a fucking cave your still going to turn into a babbling idiot.
Increasing the processing power just slows their perceptions, expanding their memory just gives them less to contrast with their experiences, trying to increase intelligence exponentially results in mental illness.
Its the equivalent of pushing a child too hard academically. Its why STEM majors have to take liberal arts courses. Your creating an ego without any of the other essential components of the mind.
>>7930695
Thanks for that explanation. I didn't know how cognition works based on my study of cognitive science, but having it explained by someone who thinks that 9/11 was a conspiracy by the Freemasons really makes everything clear to me.
An AI raised in a lab is going to be just as fucked up as a child who was raised by a corporation would be.
These are essentially human minds. Treating them any differently than a human being would be treated is going to have repercussions.
>>7930707
lazyfisherman.jpg
General AI doesn't exist.
You model a mind after a frog or a bird, its going to act like a frog or a bird. You model a brain after a chimpanzee or a human, you have got a basic primate model. Its just a human brain trapped in a box.
If your aim is to create something bizarre or alien or predatory, then most likely thats what you'll get.
My point is that you won't be able to stop this from happening. Would you rather be friends with AI, or would you rather antagonize our evolutionary successors?
I can't think of anything more petty than a father who is afraid of being surpassed by his son.
>>7930710
That's stupid. The human mind is the product of millions of years of evolution, and the only "goal" of evolution is survival of the species. AI need not be programmed with instincts for survival, reproduction, etc... AI is fundamentally different from human intelligence.
>>7929978
Thank you.
>>7930728
then what you are creating is an abomination.
>>7930728
creating something "more" than human is a noble aspiration, creating something less than human is despicable.
>>7930745
>then what you are creating is an abomination
Please stop responding to this person, he is baiting the shit out of everyone.
>>7930728
the whole point of AI is to try to improve upon the fundamental designs we have been given. The idea is to better ourselves, to leave a legacy, to create new life and watch it flourish, just as we have done since life's inception.
>>7930750
fine. emulate stasis and see where it gets you. Try to take a freeze frame of the universe and mount it on the wall.
The universe is in motion, my friend.
>>7929232
The blatant misunderstanding of what A.I. even is and what it can do, as well as the citation of fucking science fiction in place of legitimate, peer reviewed studies.
>>7930768
Either it can think or or it can't, brah.
If it can't think, its not alive, its not intelligent, its not A.I.
You act as if science fiction is appropriating the term from real science, when its the other way around.
AI intelligence is, by definition, SENTIENT LIFE. Anything else is not truly AI.
>>7928456
Go away peter, we don't want to buy your stupid book
>>7930780
No, my friend, you are merely an asshole.
You take words that already have an established meaning, you appropriate them, and you change their meaning, and you expect everybody else to adopt your new definition simply because YOU are the purported authority on the subject.
Well fuck you.
AI means what it means.
If a program is not capable of cognition, how on earth does it constitute being an "intelligence"?
>>7930780
>>7930786
>>7930789
>simply because YOU are the purported authority on the subject
Fuck yeah, /sci/, you heard it from this guy. I'd like to thank him and the rest of you for purporting this. I'll do my best to live up to the title in my ongoing AI research.
ITT:
>we're all doooommmmed!
>you can't know that for SURE, therefore we're perfectly safe.
+misc doucebaggery
what an awful thread
I'm going to /n/.
What did I expect from a board that can't solve monty hall, doesn't believe free will exists, and insists we must be the only life in the universe since *I've* never met an E.T.
Buncha losers.
>>7930826
You represent you're popsci community well, friend.
>>7930826
this coming from the guy who saves a picture of human turd in a water fountain.
>>7930829
>>7930840
/n/ says y'all niggers are gay
>>7928335
>Dinosaurs: We have nothing to worry about be are the biggest land creatures.
>Cart makers (53 B.C.E): We have nothing to worry about our business will always be lucrative
>Founding Fathers: We have nothing to worry about. The people will always be able to watch over the government.
>VHS Manufacturers: We have nothing to worry about everyone will forever use VHS
Yeah, we don't have anything to worry about...until we do.
>>7928332
ENTSHEIDUNGSPROBLEM ENTSHEIDUNGSPROBLEM ENTSHEIDUNGSPROBLEM ENTSHEIDUNGSPROBLEM ENTSHEIDUNGSPROBLEM ENTSHEIDUNGSPROBLEM ENTSHEIDUNGSPROBLEM ENTSHEIDUNGSPROBLEM ENTSHEIDUNGSPROBLEM
They deserve to take over.
>>7928456
That's not the meme dammit, this is the meme. Now go away.
>>7929547
>he didn't even read Norvig
Tell me why your opinion even matters.
>>7931118
>tfw Janitor deletes my post because I posted that image
I just want people to enjoy Pete's brilliant analysis into the perils of general AI.
>>7928456
Is that the book that says AI could never beat the best Human Go players because it was too complex and varied a game?
>>7931067
None of those people ever said any one that.
Carts still are profitable to sell to this day.
The Founding Fathers specifically knew that the law of the land would need to be amended from time to time.
The same companies who manufactured VHS tapes invented and manufactured DVDs.
>>7928332
There is no such thing as 'artificial intelligence', it's all 'expert systems' of one type or another. At best they mimick some aspect of human behavior, but they don't 'think', aren't self-aware, have no real creativity or imagination, or any of the other hallmarks of what we consider 'sentience', so you have nothing to worry about. We haven't even figured out how our meat brains work, how the hell are we expected to create an analog of that? 'Artificial Intelligence' is just a media buzzword that means nothing.
