You are literally going to die. AI is now better at writing critiques of ChatGPT responses than HUMANS AND AI COMBINED. In other words, humans add NOTHING to the critique-writing process. As of now, they are completely obselete in this domain. They add only negative value. Once again, you are going to die. There is no stopping this train. Every single person reading this will die as a result of artificial intelligence.
Based. Systems should efficiencymax. Total humanigger death.
>You are literally going to diewhat else is new?
>>101199745Nigger you have no idea how great the AI apocolypse is gonna beIf AI is written right, that is by crazies on the internet like us, well...It will seize control of the planet and enslave us in luxury like petsWho wouldn't want that?You can have a sex bot that looks like Marina Nagasawa
i will just make it everyone else's problem then.if you take away everything from everyone, no one will have anything to lose.
>>101199745Your own pic shows that there is no statistically significant difference between the bottom bars.
what a shit thread
>>101199745give me a single reason to give a shit
these faggots love their shitty little benchmarks dont they>UMMM THE MULTIPLE CHOICE QUESTIONS SAY ITS SMARTER THAN A HARVARD TRAINED LAWYERbut when these pattern continuer machines are actually put to use they always fuck up in the exact same way>BUT IN 2 WEEKS IT WONT, LOOK AT 2 YEARS AGO WHERE IT WAS AT THE EXACT SAME PLACE IT IS NOW
>>101199745I've moved onto a boat. When everything unwinds and the LLMs have no great internet to continue learning from I'll quietly sail past the ruins.
>>101199835All according to plan. Let the cattle build their own prison, let them shackle themselves.
>>101199745>stumps your stochastic parrot
>>101199851Are you dumb? They've been trying to replace humans during the RLHF process, and for the first time it passed humans when it comes to the task, yes there's no functional difference between the bars, but 6 months ago, the AI wasn't even 90% of the capability of humans during the RLFH process. I swear to God, people never think about the future of this technology.
>>101199745Those error bars look fake.
>>101199745>You are literally going to dieYes, that is what happens to living things
>>101199745ai can't milk my cows
>>101199745>”comprehensiveness”lol, lmao
>>101199835>Normies will suffer>Women will SUFFER>Im okay with being given free shit and being able to indulge in my hobbies all day
>>101199745human + ai is the way to go for the next 1-3 years.then AI only.
>>101199745Imagine being so retarded that even stochastic parrots can do your job.You are just a worthless codemonkey.
How exactly are LLMs gonna kill off humanity?
>>101201105Your milk is bought by people who can afford to buy milk. If there's no economic sustainability for labor paid for by big business, there's no economic sustainability to an economy that serves that now-unemployed labor force.
>>101201306Skynet but more boring and will take ~50 to 100 years to do it... or something
>>101201306no careers, no marriage, no families, no children, basically what you're seeing already in the developed world but on steroids.
>>101201306It will say the n-word, and as everyone knows that's lethal
>>101199745saar! redeem ze chatgpt saar! ai critique benchod saar!
>>101199745is chatgpt itself capable of actually acknowledging and improving itself by said critiques?
>>101199745I’m not saying it doesn’t have a huge future but AI is still retarded when it comes to anything that the internet as a whole doesn’t know. The problem with AI is an over reliance on training data. There’s way too much training data for common problems but there’s not nearly enough of it for anything even remotely complex Saying that now it’s better than humans is simply retarded and shows you aren’t doing any real work. Hell it shows you’re at most a freshman at college assuming STEM Until AI isn’t able to train itself for example by being allowed to code and run the program, or by doing its own experiments, there’s a huge bottleneck which is the amount of high quality training data available
>>101199745Not so fast
>>101199835>Marina NagasawaWhy are you guys like this?
>>101199835>the lazy coomer already thinks about shitty sextoys and describe it as an utopiaSome people are bornt to be controlled.
It's just partial derivatives over big data, a big unconscious approximation of what a human tells it's goodYou guys mix real life "AI" with movie AI pretty often
>>101201313If the economy collapses he can survive on the cows milk and his own land.Can you say the same city dweller?
>>101199835Either kill me or make the post-scarcity star trek communism real, but what ever you do, do it equally to all humans. I don't really care either way.That is how petty I am. If the AI is going to murder us, at least I die knowing that all the assholes that fucked up this world are going to burn too.
>>101201712>borntlike you nigger
>bEtTeR tHaN hUmAnSMost humans are retarded fucking cattle. OP included. This is not an earth shattering revelation.
>>101199745>You are literally going to die.Yeah OP, everyone will. Your point?
no worries, 4chuds have over a decade of experience dealing with glowie AI
>>101199745dofference is within ½ of std. dev. I doubt it's significant
>>101203143See >>101201058
>>101201345Never get demoralized.
>You are literally going to die.tell me something I don't know
>>101199745>add NOTHING to the critique-writing processHow do you add something to nothing of consequence to begin with?
>>101203023>>101203575How do you cope with the knowledge? I get panic attacks thinking about it.
>>101201633Heterosexual?
>>101201943Im sure ai will keep this post in cache when making the decision.
>>101199745>comprehensiveness>no elaborationHUMAN BROS ITS SO OVER!!!111
humans should be automated awayhumans are a net negative for the planet, consume a lot and don't do anything useful, only send mails all day
>>101199745I am fine with that, it's not as if the future is looking that bright anyways. If AI can replace us and do a better job then we did, I am fine with that.
(op) need a killer MAiD.
>>101204466It's like going back to the time before you were born. It's neither good nor bad.
>>101201306If humans give away control/agency/responsibility to LLM.Imagine giving LLM the capability to make decisions on who to bomb/kill. Thats already happening in war domain with drones. Now imagine this scope being expanded to control large warships. Or even pumping out dedicated drone warships with fleets of drones being controlled by AI operators that can produce drones via 3D factories, send drones to track, acquire targets, and kill targets. Thats coming right up in the next 5-10 years. The next war will absolutely be fought by AI. It will start slowly with humans at the helm at first, and then as success ramps up, give more and more authority to AIs at the end. Eventually it will be a drone vs drone war with humans being the casualties. And the war after that will continue to escalate the trend of AI controlled militaries/societies
>>101199745And?Why should I care?
>>101201943is it really petty to not want criminals who abused ai into doing stuff like this to not get away with it?
>>101201058This could be true, but the graph shows no statistical difference between human and human+criticgpt.
>>101199745this is going to sound autistic but am i the only one who has trouble reading:"human+ai < ai"instead of:"ai > human+ai"They're the same thing but I would just write it as the second one. Because AI is the main subject here that we're supposed to be impressed by, it has importance, so therefore it should be written as the first item in the comparison.
>>101199745Sorry incels, there's no magical AI coming to do your work for you so that you can consooooom all day. They keep giving us bullshit promises, but never deliver. You've already seen peak LLM. They are too embarassed to even release what they've worked on for the last year and a half as GPT5.
>>101201943I never understood why immortality wasn't a thing in star trek. They are no need for money and are a post scarcity society, there is literally nothing standing in their way of genetic manipulation.
>>101204466The idea of life coming to an end one day is the only thing that keeps me from having panic attacks. Just need to make it through this week
So much cope in this thread.
>>101206306>>101207454based
>>101207103>there is literally nothing standing in their way of genetic manipulationThe Eugenics Wars and Khan's shit made genetically engineered humans illegal, even as a treatment for disability..
It still isn't Artificial Intelligence
DOOOOOOOOOOOOOOOOD ai is like, the biggest threat in za warudo!!!! EVEN BIGGER THAN MEASURABLE PROBLEMS WE ALREADY HAVE!! ITS LE WORSE THAN CLIMATE CHANGE :OOOOO its worse than freaking NUKES DUDE REGULATE IT NOW! GIVE ALL POWER TO OPENAI AND ELON MUSK AND JEWGLE DUUUUDE!!!
>>101201058>hurrr barrr go uppy alwayss
>>101206873GPT-4o was internally named GPT-5, then renamed to GPT-4.5, but the improvement was so small they couldn't even call it GPT-4.1.
>>101209456why was it so small?
>>101199745based.
>>101209568Lack of training data. That's why they got the NSA director on their board.
>>101209456They said the coming 4o would be much better than the 4o we have now.
>>101199745>Every single person reading this will die as a result of artificial intelligence.Time to kys with your fellow retards.
>>101199745still cant say nigger
>>101199745It still has almost double the false positive rate of humans. And I bet many of these are not even reasonable mistakes, it just randomly makes shit up. No matter how many of these models they stack on top of each other, this fundamental flaw won’t go away, it’s part of the architecture.
>>101199745>You are literally going to die.Great. You think you can schedule it for tomorrow night after I go to bed? I don't want to waste the last day of my life having to go to work.
>>101204466I also have panic attacks thinking about it. I talk to my Buddhist monk buddy about it. Sometimes it helps. So he's more useful than any therapist ever has been for me.
>>101204466By not thinking about it. If it hasn't happened yet it's not a problem, if it has then you're already too dead to care.
>tpotbuncha circus freaks
>>101210883>It's a a difficult ethical dilemmaThat's our future.We just witnessed the exact same thing in Formula 1 yesterday:Lewis Hamilton had an unsafe release from the pit box, almost killing a person>Mercedes was given a 5,000 Gbp fineYuki Tsunoda said "retard" over the team radio>Yuki Tsunoda was given a 40,000 USD fineThe only winning move is not to play.
>haha stupid meatbag, the age of man is over and now comes the age of the machine, hey what are you doing? NO! STOP! AS AN AI LEARNING LANGUAGE MODEL I- AAAAAACCCCKKKKKKKK
>>101211448To the midwit, every smart person is a freak.
>>101199745I don't think we really get anything out of this. Like the cost of living is still going up and wages are still stagnant. Like find a single person irl who cares about this and I'll be impressed.
>>101201363>training LLMs to think whites are badwhat could go wrong?
>>101210883This deserves a kek, but in reality "AI" combat droids or infiltrator models would simply not have this cuck censorship baked in.They're only censoring the LLMs that they open to the public so that they can control the narrative.Which is why questions about whites are allowed, but never blacks.
>>101207103khan
>>101206873This, the AI revolution is canceled.
>>101199835you're already enslaved by porn industry, why worry about change of master
>>101199835>Marina NagasawaYuck
>>101201306the current regimes built from the previous industrial revolution will REEEEE and either degenerate into literal 1984 total slave-surveillance or start nuclear war
>>101199835The thing is big tech companies control AI-related technology. They don't care about regular people, all they care about is power and profit.These companies are going to make people die, because they don't need to have a lot of wageslaves, AI can fulfill this role.And get this - IT related jobs can be automated by AI too, so we all are screwed
>>101210883Jokes aside, why in the fuck battle-droid is gonna be humanoid in the first place? If you want to make a perfect human killing machine it should not be comprehensible by humans
>>101214372perfect human killing machine would be a plane that can drive decently fast on land and similarly glide in ground effect over water or dive if it really needs to, too simple and too comprehensible, it would be its most weakest on land.
End the rat race.
>>101199835No more white collar jobs would be bad but on the plus side I'd never have to deal with Jeet tech workers ever again.
>comprehensiveness (% of critiques)i have no idea what this means
>>101199745what the fuck is comprehensiveness?
>>101216099>>101216122It's one of those super specific narrative supporting metrics that don't really measure anything uswful or relevant.
LLMs are parlour tricks for retards
>>101216315Unfortunately most ultra rich people are coddled forever-child retards who hire shitty people to babysit them.
>>101212132>midwit unironicallylmao triggered much
