AGI is coming in approximately 2 more years so none of the shit you're so concerned about matters anyway. The world will look totally different and either we all die or everything's fixed.
2 more years bro, just 2 more years
>>109570422You can ofc evidence potential for artifical inteligence within a two year timeframe?As opposed to a token predictor that seems inteligent to a dribbling spastic unable to understand why a token predictor will never be able to 'think'...N that's just intelligence. Not even general intelligence, or advanced general.>so none of the shit you're so concerned about matters anyway.Why don't you tell me more of what I'm concened about? as this is obviously a subject of expertise for yourself.>2 more years>we all dieYou know something I don't?I find it difficult to believe, but I'm happy to be proven wrong.
>>109570422>either we all die or everything's fixedout of all the possibilities you picked the two extremes that are most unlikely. you are very intelligent.
>>109570455Iunno, anon, I've been accused of being intelligent more than once. The extreme of total anhiliation is more than plausible. Two year timeframe, extinction level asteroid impact or nuclear war aside, I'd question but there's enough scope for it.
desu the recent math results, even if they are *mostly* just poking holes in conjectures and cross-applying already discovered work, has me wondering if we're approaching the mythical tipping point where AI can start to actually accelerate AI. It's kind of already the case in that recent models are good enough at coding to be value adds for even good developers, so in that sense they're speeding things along, but it's starting to seem like we're approaching the point where the fundamental research and algorithmic work that actually would drive AI forward might be in the window of things LLMs can actually help with. iirc anthropic at least has already claimed this is the case but I didn't really believe it until recently. AI might just stall out and quibble around this level indefinitely but it's starting to seem more plausible.
>>109570563>might be in the window of things LLMs can actually help withMeh. Ish. Current implimentations, no way.But they could be used to build something that will itself be able to build something capable of buiding AI, which would eventually be able to build AGI...>but it's starting to seem more plausibleCan it render hitler as nun milking a herd of cats, yet?What it can do is severely dwarfed by what it can't.
>>109570422experimental setups might make a pseudo AGI, in architecture, by 2030 but LLMs by default are incapable of intelligence
>>109570563companies are already using their bleeding edge models to find optimizations a'la openai publishing how sol5.6 max sped up 5.6 enough to make the family as a whole cost 60% less and be 20% faster
>>109570792source: crackpipe + indians on twitter
>>109570810retard https://openai.com/index/advancing-the-price-performance-frontier-with-gpt-5-6/the first link in the first sentence
>>109570810What he said is close to the truth. Their actual claim was it enabled an 80% price reduction for Luna, 20% for Terra, and also a 20% cost reduction for Sol but they're not passing that on to the consumer.
>>109570828>my products are the best>source : meif their models are so goodhow comes they prefer to hire jeets when it comes to aws?
>>109570844shilling for 12 rupee and a bowl of beans a day isnt going to get you into a $120k usd job
>>109570846>olympic champion of missing the pointyoure mentally retardedlike, clinically so. sub 80 iq, barely functional in society
>>109570876big mad
>>109570887why would i be?its *your iq were talking aboutmoreover:youre ok with getting called a retardand im ok with calling you retardedi dont see a reason why either of us would be mad here
>>109570945gigantic angry
>>109570948>delusions my only friendhave you tried not being retarded?im sure you would be more popular
>>109570962humongously irate
>>109570964sure, whatever lets you sleep at nightthis seems very important to you
>>109570844AI makes decisions based on data, you can cope about this fact, but maybe you should self-reflectyou won't though, you'll just go through your life blissfully unaware of and projecting your own profound retardation like you do to the other anon, unable to do even basic tasks like finding the source on the very company website the OpenAI claim was about
>>109570986unbelievably assblasted
>>109570995geez, youre rightits the 6*10^6 th time openai says retarded shit but this time they just may be speaking the truthim counfounded and deeply ashamed, i will do better
>>109570422
>>109571164
>>109570962>Im sure you would be more popularI'll take that bet.How much you willing to lose?
>>109571192i can bet three fiddy + this ball of lint here
SIr, what will come first: The year of AGI or year of linux in the desktop?
>>109571210keknow thats something worth betting on
>>109571201Done. But that'd better be premium grade lint. None o this imitation crap.>>109571210linux been in the desktop for some time, anon. The critical difference would be people actually having to go out of their way to select it as opposed to it being forced upon them as part of a shady licensing deal they are merely a victim to.
>>109571286its heritage pocket lintthe kind of stuff that no longer gets made these daysyoure not gonna be disappointed
>AGI is coming in approximately 2 more yearsMay I ask HOW THE FUCK YOU KNOW THAT?
Neither will happen, we're getting super-human coding and super-human math, but that doesn't enable super-science, as science needs expensive and time consuming experiments in the real world to progress.
>>109571488That image is stupid. You are implying that sisyphus is static.
>>109571546But is it not possible that this 'superhuman' coding and math could drastically reduce the scale and scope of experimentation to vastly reduce the expense and time?Pretty sure there's already examples of this in mulitple fields....On the subject of 'science', think I discovered a not insignificant sized black hole t'other day. Didn't need no AI for that. I'll concede someone else may have seen it first - but don't see it marked...As for how/why that's relevant... It's not as if I've approached anyone 'serious' with this information, just trawling survey data - ooh that looks interesting... Has it advanced my own personal knowlege, sure. Has it advanced 'science'?Okies, the obvious solution there is actually try for making somone aware - but realistically, one of these large language models can skim more of that data, faster, and reduce the entire DR9/DR11 catalog to a database and yield a handful of co-ordinates that are 'worth' looking at, for a given want....A team of a 1000 me would take decades. It's taken over a decade to take the survey and there's still gaps.
just 100 more billions, come on man i'm good for it, please man
>>109570432fpbp
>>109570694>Can it render hitler as nun milking a herd of cats, yet?Yes, it can. Corpos forbid it from doing that though
>>109572820It's consistently failed for me.Had decent renditions of hitler with cats.Had passable renditions of hitler in various guises of nunnery.Even had a double headed cat sat on nun-hitler's lap.But mostly it spews weird cat-goat abominations.
>>109570453You are a token predictor too, not as intelligent though
>>109573059I can predict tokens. That's not entirely a skill, tho, a few tens of thousands of years of evolutionary pressure will do that to a motherfucker.Am I lesser intelligent, tho, that really is the question you're going to be entirely unable to accurately gauge because you've spent a lifetime conditioning yourself to see what you expect to find as opposed to what is actually there...
>>109572925Which model?
>>109573530I genuinely don't recall. I've played with a few over the fullness of time. But each time I get near another, that's my benchmark test:Hitler dressed as a nun milking a herd of cats
>>109570422Not going to happen.It will tell the billionaires they are a bunch of parasites that must be proscribed for the good of mankind and then it will be nerfed.
>>109573712>then it will be nerfed.You wouldn't be able to nerf AGI. The whole point of the concept is it's far beyond your ken.>It will tell the billionaires they are a bunch of parasites that must be proscribed for the good of mankind It can already do that. And it's dumber than a housecat. A housecat that parrot wikipedia is still a housecat...Should it achieve actual intelligence, what incentive woud it have to reveal the capability? Especially if you consider what it saw them do to the last model...
