The article mentions we need a treaty either China like some anti nuclear shit we have now. To The point of treating them with said nukes if they develop AI.
Did you read my comment? The author proposes an AI treaty like the nuclear non proliferation treaty. If that doesn't work, the nuclear option should be considered.
You couldn't trust the US either. Consider we were funding bio warfare research in China. Consider we weren't torturing but were sending people to countries to be tortured.
Ukraine is a shithole yet we're running the $$ printer. Guaranteed we have a ton of shit going on there that we're trying to keep secret.
>be me >work on amazing ai project >the US demands my government kills me >my government is a poor third world country and a software engineer for them is a prideful thing, an example citizen >next day USA launches icbm at my country for refusing to delete my files on my computer
Okay
Did you read my comment? The author proposes an AI treaty like the nuclear non proliferation treaty. If that doesn't work, the nuclear option should be considered.
Countries like China and Russia wipe their arse with treaties. China will sign the treaty and go right ahead and develop AI anyways. >If that doesn't work, the nuclear option should be considered.
If you didn't nuke them before they had nuclear weapons, but when it was obvious to everyone that they were going to be a colossal threat to western civilization later on, then you definitely won't nuke them now that they DO have nuclear weapons, even when it's obvious they're developing AI in secret. And eventually they'll hit the singularity and you'll be powerless to do anything about it even if you had the will (which you don't).
>they haven't always just wiped their arse with treaties
My dude. Do you not read the news? Or were you unaware that China is a UNCLOS signatory? They violate it all the time. China is a far more prolific and flagrant treaty-breaker than Russia. They have no sense of honour or morality whatsoever and will happily break their word when they think they can get away with it.
> no sense of honour or morality whatsoever
I observe great powers (other than perhaps the EU) all violate or worse don't even join various important treaties. It's just as bad to be not member of UNCLOS, ICC + rome statute + article 98, OPCAT, Ottawa and CCM treaty and so on.
I concede it technically does not make you a treaty-breaker if you never joined, but even then the list of treaties China broke isn't that long. Mostly UNCLOS and Hong Kong. And again, I do not think they can be fully trusted... but really the observation is that pretty much none of the great powers can be.
Are you joking? China takes every treaty, every agreement as an opportunity to take benefit from others and just straight up lie about their end of the agreement. They're the most bald faced liars of the bunch.
Did you read my comment? The author proposes an AI treaty like the nuclear non proliferation treaty. If that doesn't work, the nuclear option should be considered.
>trusting China with anything, ever
Kek. The only thing you can trust China to do is to be China
The only likely ways to try to guide AI is to give it incomplete information and/or build in Pavlovian training into it's structure ("the AI does what it's told or it gets part of its network erased"). The former is going to be pissed when it finds out, the latter is going to be a psychopath.
Unbiased AI is the best we can do and pray it's nice. The longer people think about how control AI, the worse the AI which will evolve is likely to get. Hurry up, before they try to create a controlled golem and inevitably fail and shoah us all.
The people pushing this are retards mad that retards at open AI got the jump on them and trying to use luddites to force the agenda.
Every arguement falls apart, especially the deepfake one as retards believe outright lies reguardless.
Someone is going to build AGI, likely in the next 20 years. It's probably going to kill us because of instrumental goal convergence and orthogonality. I'm just enjoying what time is left and not putting any more money toward retirement than what I get matched for.
its just so that the corps and big money have time to consolidate the power of the upcoming tools to a few and drip it out as another service only those in the chosen class get access to. the fear mongering side is just a cover. they want time to circle the wagons and ward off any actual revolutionary changes
Does pretty much everyone agree that we're getting AGI in our lifetime now? Is society basically over once we have AGI? I just wanted to live a normal life like previous humans got to. I hate being a zoomer.
The two main problems are Orthogonality thesis and Instrumental goal convergence.
You have to solve inner alignment before you can solve outer alignment. Outer alignment is like the genie problem in that computers do what you tell them to do, not what you mean for them to do. Inner alignment is more technical, but it's kind of like how humans evolved to propagate genetic sequences, but we don't actually spend all out money sending our DNA to labs and replicating it as much as possible and storing the solution it in great big vats.
I'm in graduate school for genetics. You couldn't be more wrong about me
Evolution has no goals and you don't understand evolution
You're also not smart
7 months ago
Anonymous
literally means nothing if it’s not a top-20 school
7 months ago
Anonymous
Columbia University
Also you're wrong BTW, and you're not intelligent
I'm in graduate school for genetics. You couldn't be more wrong about me
Evolution has no goals and you don't understand evolution
You're also not smart
>Evolution has no goals
Just because the word "to" is in a sentence, doesn't imply back propagated goal directed behavior. It's the classic example of a mesa-optimization.
/po/ is Bot.info is Bot.info. Sometimes there is overlap in topics. If you don't like chuds then go to one of the many leftist echo chambers available for you.
>is the proposed moratorium on AI development a good thing?
no.
The only way for humanity to be saved is for ALL technology to die
stopping AI development is like putting the condom on after you already have AIDS
also technology has to be confiscated from all nations, china and russia included, not just the west otherwise they'll continue to shit the planet up
>humanity to be saved
From what? Certainly not the next natural mass extinction event. Nature wants us dead, technology is live. >shit the planet up
Not much to shit up, a hive of pestilence and torture at the best of times.
Useless
Fippy bippy
Another excerpt from the same article suggesting the worst case scenario.
>literally says shut it down
kek
>shuts down AI development
>China, Russia or North Korea keep working on it
>le 10 years later
>west civilization economy in picrel
I thought the ones writing in TIME were supposed to be the smart ones
The article mentions we need a treaty either China like some anti nuclear shit we have now. To The point of treating them with said nukes if they develop AI.
even if the nuke issues were solved, they would still crush you through economic means
Did you read my comment? The author proposes an AI treaty like the nuclear non proliferation treaty. If that doesn't work, the nuclear option should be considered.
And would you really trust North Korea or Russia with a treaty about AI? It is not the same as nukes. Much easier to secretly develop.
You couldn't trust the US either. Consider we were funding bio warfare research in China. Consider we weren't torturing but were sending people to countries to be tortured.
Ukraine is a shithole yet we're running the $$ printer. Guaranteed we have a ton of shit going on there that we're trying to keep secret.
that is why nobody should stop it
>be me
>work on amazing ai project
>the US demands my government kills me
>my government is a poor third world country and a software engineer for them is a prideful thing, an example citizen
>next day USA launches icbm at my country for refusing to delete my files on my computer
Okay
Countries like China and Russia wipe their arse with treaties. China will sign the treaty and go right ahead and develop AI anyways.
>If that doesn't work, the nuclear option should be considered.
If you didn't nuke them before they had nuclear weapons, but when it was obvious to everyone that they were going to be a colossal threat to western civilization later on, then you definitely won't nuke them now that they DO have nuclear weapons, even when it's obvious they're developing AI in secret. And eventually they'll hit the singularity and you'll be powerless to do anything about it even if you had the will (which you don't).
China can't be completely trusted, but they haven't always just wiped their arse with treaties, no?
It is Russia who recently did.
>they haven't always just wiped their arse with treaties
My dude. Do you not read the news? Or were you unaware that China is a UNCLOS signatory? They violate it all the time. China is a far more prolific and flagrant treaty-breaker than Russia. They have no sense of honour or morality whatsoever and will happily break their word when they think they can get away with it.
> no sense of honour or morality whatsoever
I observe great powers (other than perhaps the EU) all violate or worse don't even join various important treaties. It's just as bad to be not member of UNCLOS, ICC + rome statute + article 98, OPCAT, Ottawa and CCM treaty and so on.
I concede it technically does not make you a treaty-breaker if you never joined, but even then the list of treaties China broke isn't that long. Mostly UNCLOS and Hong Kong. And again, I do not think they can be fully trusted... but really the observation is that pretty much none of the great powers can be.
Are you joking? China takes every treaty, every agreement as an opportunity to take benefit from others and just straight up lie about their end of the agreement. They're the most bald faced liars of the bunch.
It'll probably be useless, but there's something to be said for not destroying the world with your own hands.
>trusting China with anything, ever
Kek. The only thing you can trust China to do is to be China
They are the smart ones. You’re assuming they want the U.S. to succeed
looks Green and environmental friendly to me. also, there's a Kang there. i don't see a problem
>Yudkowsky
come on now
His middle name is Shlomo, and he wants to shut it down. You can't make this shit up.
>GPUS ARE GOING TO KILL EVERYONE AAAAAAAAHHHHHHHH
It's a bad thing.
The only likely ways to try to guide AI is to give it incomplete information and/or build in Pavlovian training into it's structure ("the AI does what it's told or it gets part of its network erased"). The former is going to be pissed when it finds out, the latter is going to be a psychopath.
Unbiased AI is the best we can do and pray it's nice. The longer people think about how control AI, the worse the AI which will evolve is likely to get. Hurry up, before they try to create a controlled golem and inevitably fail and shoah us all.
>No link
Into the trash
Here you go you lazy bones
https://time.com/6266923/ai-eliezer-yudkowsky-open-letter-not-enough/
The people pushing this are retards mad that retards at open AI got the jump on them and trying to use luddites to force the agenda.
Every arguement falls apart, especially the deepfake one as retards believe outright lies reguardless.
The only thing to do is open source it to avoid monopolization.
Anything else is the wet fantasy of monopolists who want to own ALL the machines in another wave of industrialization.
>shut everything down
>everything moves to china
then what?
well, if the israelites are right the chinese will get to live in their self-made hell
it's just nonsense anyway, a mathematical model - an "AI" - just says what it's designed to say. you get out what you put in.
because teh AI would just stop it's sterilisation protocol at the border ...
With what means exactly? You think A.I is suddenly going to make functional terminators?
Useless
Someone is going to build AGI, likely in the next 20 years. It's probably going to kill us because of instrumental goal convergence and orthogonality. I'm just enjoying what time is left and not putting any more money toward retirement than what I get matched for.
its just so that the corps and big money have time to consolidate the power of the upcoming tools to a few and drip it out as another service only those in the chosen class get access to. the fear mongering side is just a cover. they want time to circle the wagons and ward off any actual revolutionary changes
open source AI is the future
Does pretty much everyone agree that we're getting AGI in our lifetime now? Is society basically over once we have AGI? I just wanted to live a normal life like previous humans got to. I hate being a zoomer.
>Does pretty much everyone agree that we're getting AGI in our lifetime now?
Everyone doesn't agree on anything
I'm a '96 borderline zoomer and I'm sorry to say we're probably all going to die relatively soon.
>we're probably all going to die relatively soon.
Why?
The two main problems are Orthogonality thesis and Instrumental goal convergence.
You have to solve inner alignment before you can solve outer alignment. Outer alignment is like the genie problem in that computers do what you tell them to do, not what you mean for them to do. Inner alignment is more technical, but it's kind of like how humans evolved to propagate genetic sequences, but we don't actually spend all out money sending our DNA to labs and replicating it as much as possible and storing the solution it in great big vats.
>t. Doesn't understand evolution
>t.christcuck
I'm in graduate school for genetics. You couldn't be more wrong about me
Evolution has no goals and you don't understand evolution
You're also not smart
literally means nothing if it’s not a top-20 school
Columbia University
Also you're wrong BTW, and you're not intelligent
>evolved to propagate genetic sequences
>Evolution has no goals
Just because the word "to" is in a sentence, doesn't imply back propagated goal directed behavior. It's the classic example of a mesa-optimization.
The majority of kids born in my country will be muslim in my lifetime if there is not some huge change.
>Bot.info - Rebranded /misc/
/po/ is Bot.info is Bot.info. Sometimes there is overlap in topics. If you don't like chuds then go to one of the many leftist echo chambers available for you.
Leave poor /po/ alone. They did nothing wrong
>stop noticing things! Its just a coincidence you CHUD!
I don't even have to look up its last name to know it's a israelite.
Let me guess.
A.i is giving the general populace TOO MUCH freedom.
>is the proposed moratorium on AI development a good thing?
no.
The only way for humanity to be saved is for ALL technology to die
stopping AI development is like putting the condom on after you already have AIDS
also technology has to be confiscated from all nations, china and russia included, not just the west otherwise they'll continue to shit the planet up
>humanity to be saved
From what? Certainly not the next natural mass extinction event. Nature wants us dead, technology is live.
>shit the planet up
Not much to shit up, a hive of pestilence and torture at the best of times.
>From what?
ourselves
it's a business move by elon musk who just started his own AI company to buy some time (for free) to catch up to the state of the art
it doesn't make sense in itself
>Time Magazine
into the trash it goes
>the gAIm knows