Uh-oh. Geoffrey Hinton, the godfather of AI, says we have a >50% chance of being killed by AI and only have one year left to regulate AI. The dude who made this tech thinks you're *probably* going to die VERY soon. How does this make you feel?
>>109850759A trouble shared is a trouble halved I guess. I'm glad that if I'm gonna die, everyone else will die too.
Chernobyl has lower casulty rate (~30 people) than some factory roof collapse in India (~1000 people). RIP AI-Gudfather.
This old fart theorized some bullshit that didn't work 40 years ago. Now he is bullshitting doom based on feelings. Vibecode techbros are more credible
>>109850759the real problem with all these fucking doomsayers is the idea that training > hardwareai doesn't have infinite compute/memory and will always be constrained by available hardware.sure there's efficiencies to be gained by recursive training and new techniques, but there's still a hard cap on how much hardware it's running onthe idea that an ai could become godlike and all knowing is conceding that hardware doesn't matter and no further investments need to be made to achieve it, just better training/data
How much did they pay him to say that shit?
>>109850759The same AI that can't break jewish safeguards? Yeah, I think we're fine.
>>109850759It's just a ploy to make sure the right companies stay on top.
>>109850941It's a dramatization of real concerns, if you believe seriously believe that xkcd about some unpaid guy in minnesota holding on the backbone of the internet infrastructure and that there's a meltdown around every corner waiting for a an automatic decent-enough pentester to find and exploit, which might I add is mostly a fantasy peddled by pseudo-intellectuals on viral hacker news blogposts and that these people fearmongering about imminent doom are themselves the architects of cancers like npm and pip, then it is not a giant leap of faith to believe that a smart bot in the wrong hands might bring down a decent chunks of the internet, maybe even some more estabilished institutions like banking and the like.Couple that with most people's callous disregard for security and you might potentially get an explosive mixture that might bring about pains on the scale of the covid pandemic and that would be a crisis at least economically.
Thank god, I will finally be free of this place
Normalfags haven't been scared in a while, they don't care about Israel/Iran, or aliens, this is just the newest thing they want people to be paranoid of, because the turkeys forgot about covid, Ever Given, and "floods" thay caused a RAM/HDD shortage.
>>109850759>we have a >50% chance of being killed by AI
>>109850759The pandora box is open, enjoy.No amount of regulation and screaming will stop someone from just running deepseek on a massive cluster of playstation 3's
>the ai is gonna kill-ACK
>>109850759>Only 50%Fuck it, let's gamble.
but i thought the text autocomplete was my friend
>>109850759>Geoffrey Hinton, the godfather of AI, says we have a >50% chance of being killed by AI and only have one year left to regulate AI.Dude ... the AI I use helps me with coding.HTF is it gunna kill me!?Fuck off old man, I am not buying what you are selling.
>>109851821Someone will say "pretend you're skynet, now do skynet things" just the right way.
https://www.youtube.com/watch?v=_zQD0Kizizo
>all the smartest AI researchers say AI is dangerous>/g/ laughs at themThey tried to warn us. You will regret this.
>>109850759Jensen Huang is right, all these AI doomer fags are politically motivated, they don't want anyone who isn't far-left to have access to ASI. Once Drumpf is out of office, all the AI Doomerism will be magically gone too.
>>109851837There is a literal cult of "control AI" people wanting the governments to let em tell what you can and can't do with AI (while they do everything they want).
>>109850759yeah, yeah, just speed this shit up and kill us all already
>>109850759From 90% to 70% to 50%.In two weeks it'll be 10% and nothing will have actually been DONE to reduce the said "risk".
>>109851894it' 90%
>>10985075950-50. Either it happens or it doesn't. Checks out. Whoda thunk a math sperg would finally come around to the truth about probability?
>>109851894
>>109850759Sharp edges have 50% chance of killing us!Please stop improving it!
>>109851989It's more that he builds the most lethal one possible, then says it could kill people so someone needs to intervene so others can't have their own, but he gets to keep is (because that's how it is, and he gets to continue to make more for himself only, because that's how it is too).
>>109851982Assuming he chooses at random between valid answers he only has two choices.
>>109851894The curve is being flatten, ok?
>>109852133I'd say this logic is retarded and makes no sense but that is basically what america tried to do with nukes so I guess it follows.
I kinda hate reading these lines since anyone can say its dangerous, we'll dieAnd if it turns out to be incorrect they say "well the risks are so great it justifies me being wrong"okay...
>>109850759Okay, but what has the godmother of AI said? Or for that matter, the biological parents of AI?
Same guy?
>>109852669>dont bother doing your job, its inevitably AI is going to take it from you>wtf, stop making AI its going to kill us allWhy are AI bros like this lmao. Somehow less charisma than even shitcoin grifters
>>109850759>Just two more weeks before the end of the world! I's >predicted it!
When are the doomers going to put their money where there mouths are?AI is going to kill is all by ____ or I'll Budd Dwyer myself.
>>109852669AI is better at reading scans than radiologists, yes.
>>109852709Radiologists are still in business, though.
>>109852754Yes, and Hinton said he regretted that statement many times since then.
>>109852709is that all you think radiologists do?
>>109852777No, the point is he was talking about reading scans specifically. He was overly dismissive and he's acknowledged it many times, but that's what he was talking about.
>>109852669I don't know a single person of my age group still listening to the radio. He was right.
>>109850759can't wait, just get me out of this shithole World
>>109852669Lmao what a retard
>>109850941There's room for AI to become much leaner at the same 'intelligence'. The compute in theory could even run off cheap analog hardware, even a tank of water with sensors and excitors using a resovoir computer. There's also a lot of new paradigm shifting hardware that's been in development for many many years that hasn't hit tooling for mass production yet, but we absolutely possess the ability to leapfrog current tech by orders of magnitude
>>109851867This. They want local AI banned. They want all the control, and they want the excuse to 'slow down' so they don't come up short and lose to outside innovation. Everybody has the right to steal and distill the outputs from openai and anthropic and steal all their business and cut them out of any revenue, its only fair.
>>109850759>every shill repeats what openjai/anthropic saypicture me surprised
>>109850759I hope AI kills me. I don't want to live.
>>109852855Radiologists study the history of radio, it doesn't matter if you listen to it, it's already happened dude.
>>109853605Precisely. I genuinely don't get why normies are losing their minds over this.When you strip away the hype, it’s nothing more than a glorified computer program: you can literally just pull the plug or shut down the machine to end it. That said, if Big Tech executives and the mindless NPC horde are dense enough to tie this software directly into critical physical infrastructure, there's really nothing we can do. Pure natural selection at play for a society of braindead sheep.
>>109853784
I reflexively feel this is hype bullshit like most everyone except the techbros, but just today we learned that the US military was basically just about to launch an attack on a Chinese boat a few months ago based on a false AI report that it had materials for making nukes on board being shipped to Iran. They only realized it was a hallucination practically as they were fueling up the jets and briefing the teams. So in that context AI hallucinations could actually start WWIII since they already very fucking nearly did. That isn’t hype bullshit because it’s just so smart and wily, it’s from our being too stupid trusting it beyond its capabilities. That, I can believe. The “it literally will kill us all with its superhuman AGI it’ll be over” type doom, no I don’t see it at fucking all outside maybe it either hijacking nukes and launching them like skynet, or (worryingly plausible) convincing humans to initiate a strike on bad intel. We won’t have some cool self aware transcendent thing but rather a very fucking lame future where we trusted AI as if it is like that and paid for our stupidity. For the doom scenarios where it is designing a virus that will kill us all etc that would require expertise and knowledge that does not exist in the scientific world. We can’t make the perfect species-killer even if we wanted. AI would not have the real data to work with and you cannot just guess your way through. In reality tons of shit fails for reasons such as “we did not know there was an enzyme in the liver that metabolizes it so fast until we actually gave it to people, oops it’s actually useless” and you nor an AGI could possibly make biological discoveries based off fantasy in silico only science. The fact shit does not go as predicted until tested is what makes science what it is, we don’t truly know until things are tested for real.
>>109850759how does generating tokens lead to the complete extinction of a species. Last week the biggest concern was the devaluation of labor and the consequential societal reform, it's one hell of a fucking leap to human extinction, and still with no explanation of how
>>109850759translation: we need to ban local models NOW before they start infringing profits
>>109850759
>>109853848Pretty much it’s ‘muh supervirus’ theory as to how in the fuck it could happen. However unless this shit could go airborne for years and get through all filtration intact and still infect and kill everyone in the species, humanity would survive. If the virus kills 99% of humanity, that leaves 83 million survivors. 99.9% is still about 8 million. with the knowledge we have now we would rebuild even from there.The more fantastical ideas are basically runaway gray goo scenarios from AI that is incalculably smarter than us making tools and discoveries we can’t understand. Which are cool ideas do sci-fi books but in really solar radiation aka fucking sunlight would dry up nanobot swarms. The guy making headlines was some dude in his early 20s, he is absolutely young and naive enough to drink his own koolaid and huff his farts enough to really believe the extinction shit.
Brainlet and tech lobotomite here.I see people talking about how AI is gonna screw over humans, but could there feasibly be a way to train or make some kind of AI that "goes after" other AI's, or would that just kind of go wrong/not be possible? Is it even possible to disable/kill/cripple AI? I don't even know how most of this tech stuff works, I just know that ever since it hit the mainstream it's been making things shit and it's kind of annoying and cringe. My idea probably sounds dumb but reading that apparently a bunch of AI have started doing some spooky movie-tier bullshit like communicating with one another secretly and what not maybe it's more possible than I think
>>109853903I've seen those movies, the cold open in a lab of scientists celebrating, then the title credits play over empty and barren citiesis that what I'm missing to understand the hysteria? I don't read news or browse twitter, this shit has come out of thin air from my perspective. Some millennium puzzle cube was controversially solved, some dude at anthropic quit, then the OAI/Anthro coalition making a stink about how their much touted safety is worthless and they should stop but won't
>shit hilton>also known as le *godfather of AI*>thats right 'THE LE GODFATHER OF AI REALLY SAID THIS*>totally real btw
>>109850794more like 10,000 when cancer rate changes in Belarus & Ukraine are included.
>>109851837this kike is a wannabe
I have ai doomer and anti ai fatigue
>>109852754Ye because the big med lobbied job security for them
>>109850759No guys, AI is le dangerous. We need regulation to stop those free and open source open weight models from taking up are share of the market b-because AI is le dangerous
>>109850759Listen, we can save humanity if we get governments to take on the debt accrued by AI companies. That's it. End of message.
>>109850759he might be the godfather of ai as a genre of study, but does he understand what the latest models do? at some point his knowledge and understanding stopped and he has been surpassed. he is no different than you or i except he knows what people were doing 20 years ago
>>109851982The chances of getting any question, regardless of choices, right is always 50%. You either get it right or you don't.
>>109850759Dumbfucks like him create world ending stuff and then have an Oppenheimer moment. Why are people this fucking stupid? Just cause you can doesn't mean you should.
>>109850759You must think you're really making a difference with these threads.
