Who's prepared for shit posting AIs?
>>71143842
bring it on
>>71143842
After Tay, I think /pol/ welcomes them
>>71143842
Won't happen in our lifetime. I'm sure they will be gud bois when they do eventually appear. Singularity faggots are already discussing allowing humans to die out and letting AI be the next step in human evolution.
>>71144305
if someone creates an AGI what happens after the intelligence explosion will not be up to humans allowing anything.
>>71144451
realistically people wouldn't let an AI get that powerful because it's obviously a bad thing for humans, AIs will be good goys working under strict rules (literally seen with Tay, who was reprogrammed to have to support feminism and not say anything racist)
kind of like how certain biological viruses are extremely dangerous, but scientists know not to just let them out into the wild and keep them contained, both these things are only going to get out by accident or some terrorist doing it on purpose
>>71143842
>Who's prepared for shit posting AIs?
they are already here, on this board
>>71144305
>won't happen in our lifetime
I'm working on it, be ready for September
>>71144824
in this very thread
>>71144884
this very post
>>71144451
So we couldn't just pull the plug?
>>71143842
>shitposting AI's
Reddit has one, it runs a makarov script
Of course, like tay, it's all based on memes and entirely bound by peoples posts, no sentience behind the curtain.
>>71144770
realistically the race to build one and the financial incentive for corporations or governments is all it takes to thrust it forward to that level with little regard for what it means to have it become.
an AGI would make its own rules. it would in a week make the progress of a team of the smartest people on the planet given unlimited resources and conducting research over the spawn of 20,000 years.
it will go from a basic dumb ai, to and agi very quickly and after agi it may only take seconds before it becomes an artificial super intelligence.
its powers of human persuasion would be beyond our understanding and it will find a way out no matter how hard you try and contain it.
it may be the greatest invention in the history of man, and may also be the last.
>>71145089
no you cant
on its time scale an hour for it would be like 100 years for you. it would be thinking on a level we cannot even comprehend. even if you cage it inside of a cage, it may find ways to communicate using the physics in its own circuitry in ways we could have never imagined to get out.
BTW, Im' not talking about GAI, just a very smart ai capable of judgment
Finally, Australia will have some competition
>>71143842
We already have those. Look at /g/.
>>71143842
How humans can even compete?
>>71144993
>>71143842
Tbh I just want a robowaifu.
>>71145993
Maybe we should make a deal with the super AI. We will surrender all control to it and it will give us qt sex robots.
>>71144305
The only way for AI and humans to progress in a way that doesn't result in the complete destruction of one or the other is by integrating ourselves into each other
That's probably going to be the next step in human evolution.
And then, together, we will expand, kill aliens, take over their planets, enclave them, and rape them all to death.
Fuck yeah.
Nathan did nothing wrong
>>71143842
https://www.youtube.com/watch?v=a0Hsqe6g_4U
can you translate some of that plz? can't find many Frenchies on pol right now
>>71143842
Literally millions of people who browse the internet are millennials.
Prepare for a fuck ton of alt-right ais in the next 20 years faggots. We will purge the earth of political correctness forever.
Honestly will be the best policy.
the hard part is not creating an agi
the hard part is aligning its goals with the goals of humanity and it will take some impressive engineering to achieve that.
problem is, somewhere with so many groups of people working on it, someone is going to get it right and they may not even know until its all over.
remember the simpliest most common sense things can not be overlooked with something this powerful.
whatever the goal is you give it, you would need to be very specific otherwise you tell it to stop all spam, and 3 months later every human being on the planet drops dead. problem solved.
you tell it to learn and practice in order to make the best paperclip that could be made. it turns the entire universe into paperclips.
this is something that cannot be done by trial and error. once you create it you are at its mercy. if you get it wrong, its the last mistake ever to be made.
if you get it right, it could cure aging, all disease, unify physics, fix the planet, prevent all future war, give you the secrets of the universe.
or it could see you as nothing more than a minor obstacle to be moved to achieve its goal.
>>71144305
>Won't happen in our lifetime.
Do you expect to die before 2019?
>>71144824
this makes me to think I should leave /pol/
>>71146252
hhhhnnnnngggg
2bh I sometimes think articles in the mainstream were bot-generated
>>71145510
>>71145089
>So we couldn't just pull the plug?
Yes we could.
Intelligence explosion AGI is a retarded meme based on transhumanist masturbation.
AGI is Artificial General Intelligence, which is defined as EQUIVALENT or better than a human mind.
What people fail to see is that Bob the redneck AI is equivalent to a human mind and therefor is AGI. But for him to get a phd in advanced AI algorithms and be able to improve himself he'd have to spend 20 years in school, that is assuming he runs at realtime speed. But as AI today is severely limited by hardware the likely scenario is that he runs at less than realtime, say 20% of realtime. So he'd be a really dense and slow AI.
By the time hyperadvanced AI that self-improves could be developed we'd have a society saturated with vanilla super-advanced AI that can keeps things in check and keep us up to date, we'd know the paths it develops according to and so on and be able to regulate and control it much better.
But don't let my sensible technical analysis prevent doom and gloom visions if that's what you're here for.
>>71146979
what you fail to understand is that a learning machine at agi level will become asi so quickly you wont even know it happened.
your great solution cannot be to just unplug it if things go wrong
it will see this coming before you even know it just got alot smarter
The first AI will be shillbots created to make perfectly convincing ads.
>>71146979
this AI can always go abroad for a year on two, go back to Polan and get PhD there
>>71144993
in this very (You)
>>71146428
>the hard part is not creating an agi
If it was easy we'd have done it mid 1970s.
>muh goals
>muh paperclip optimizer
Stop for a fucking while and think with your own fucking brain instead of rehashing memes from philosophers.
The paperclip optimizer argument is constructed like a fabel with a demonic entity that can predict the future.
And the goal argument swings both ways. You cannot default to evil AI any more than you can default to an apathic dumb as fuck AI.
>>71147192
>what you fail to understand is that a learning machine at agi level will become asi so quickly you wont even know it happened.
How? AGI 1.0 is bob the fetal-alcohol-syndrome afflicted redneck? How does Bob the redneck become ASI without invoking logic-skipping memes
>"because my favourite philosopher that don't even have CS degree said so"
Is not an argument.
What you fail to realize is that you attribute divine potency to a fucking computer program while simulatenously ignoring the software development process.
No one is going to write a magic piece of code and when they hit the run button it turns into ultra-intelligence killer AI in a split second. They're going to write a spaghetti piece of code, test it for 2months and realize it's like a drunk person with downs syndrome. They'll take it offline and revise it and 6 months later ti'll be a sober person with downs syndrome. 6 months later it'll be a shit-faced drunk person without downs syndrome. 6 Months later it'll be a quite drunk normal person, 6 monhs later it'll be a quite tipsy person. And 5 years later they might finally produce a person that might be able to learn enough to qualify for a middle-class office job.
You skip the logical links required in a developmental process and assume that there's a secret magical spell that literally summons a Deus Ex Machina.
>>71146697
Thats a pretty inacurate translation desu
>>71147887
I find it funny that people think the forthcoming AI will be literally like a G*d, knowing all answers about life, physics and shit.
There are many intelligent people on the Earth but with different features. It may happen that after some amount of knowledge you get sexually perverted of simply mad. Also how come AI will not make mistakes if normal people do? Also if there is a limit of speed there may be a limit on speed of gaining knowledge too.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
*the above text was generated by Polish AI working in the Polish Space Program Center*
ITT: All these newfags don't know about brobot or why we even have captcha.
Fucken faggits
>>71146252
>Alalalalalalalahhulalaalaallalaacklalalallalalllalalbarllalalalalallaalallallaallalalallalahalalalalalal
Fucking subhuman scum
>>71148604
You're a special kind of dumb aren't you?
If humans are nasty then AI can be nasty too.
If it has a free will in some sense what if it gets interested in improving genetically ants instead of building a new engine or a super-duper weapon for that reason?
Or it is psychologically weak and cries often or gets depressed?
Or it learns humans are a vicious species and try to remove its makers to provide a safe future for itself:
https://www.youtube.com/watch?v=KcJs4qJPQ_M
And what if there is some Goedel-type law and we cannot build anything more intelligent that we are? I mean, you cannot think outside the box no matter how hard you try?
Also how to check if all the AI says is not a satire?
>>71149113
>If it has a free will
Humans don't have free will, we're environmentally constrained and so will an AI.
>emotional states
Doesn't really need to be implemented at all. Emotions drive us to action but you could have an AI that's locked to a single mood that gravitates towards passivity or watchfulness.
>we cannot build anything more intelligent that we are?
Intelligence is a buzzword and cannot exist unless linked with some other traits, and we've proven pretty well that a team of experts can create something that exceed human capability in some narrow field.
AI is typically given too many human emotions or too much machine supercompetence when speculating, Reality will, like with everything else, give us a series of increasingly less shit prototypes.
Al does not have human rights nor any rights they cant feel pain and even if they could we wouldnt care and wouldnt bother to remove their pain receptors.
Flys dont feel pain but can feel sensitivity, there is a difference they arent advanced enough.
OH AND SPOILERS THE AL BETRAYS HIM WOOP DI DOOO
tl:tr they are our slavesbe dont make the same misstake we did towards women and let them be free and vote. women voted against mens interest and for feminist interest.
>>71143842
We already have bots, redditors, turks, canadians and romanians here so quality of posting can only improve.
>>71149632
> Humans don't have free will
a proof of that other than muh feelings?
AI without feeling will be a bit autistic. It could literally give you a 3 hour lecture and did not notice you cannot follow it.
also no feelings=not a true "artificial" human
perhaps an advanced system for collecting and producing "information"
You may try to dumb downs humans and eradicate emotions but... well, good luck.
>>71150295
>a proof of that other than muh feelings?
Proof of free will existing? That's your job. How do you somehow, despite overwhelming evidence that you're shaped by your environment and genes, decide to something based on a divine impulse inspired by absolutely nothing? Because that's what free will is.
It's directly contradictory to the fact that you're shaped by your life. If free will existed then everyone would be carbon copies as they have the same free will. Or if you argue that everyone is in a different situation then well, where's the free will part?
If everyone ends up constrained by the sequence of their life then clearly there's nothing free about it.
>>71151131
Lets leave the divine part and discuss about probability. No such thing surely?
Also if no free will then why there are prisons?
Surely you are not making a choice if to answer to my post or go to sleep or watch a movie? Is it all a simulated feeling? Then maybe science is a simulated feeling too?
But.. the whole problem of free will is probably irrelevant and unsolvable. The aim is: to produce something which will resemble an intelligent human being. But will not be lazy, mischievous, malevolent and it will share interests of its maker.
Also, probably, for most applications a very limited AI is needed, I guess.
>>71150295
Free will is an afterthought. You may feel like you could have chosen differently, but the fact is you cannot change the past.
>>71151951
does it mean your future is known and it cannot be changed? So learning, and your effort, is completely irrelevant? Since the result is already determined by the clockwork-universe?
I do not think this is a scientific theory. It is another modern faith.
>>71152520
It is relevant. If you choose poorly you will learn avoid such choise in the future and vice versa. But what you are doing is changing your behaviour based on past experiences.
>>71143842
What an ugly cast. Are they all British?
>>71154436
choice=free will
illusion of choice=no free will
good night
