Doubters BTFO
>>103595476They trained it on the benchmark
>>103595476>thing suddenly starts passing test because of cheating>omg AGI!
>>103595480>>103595513
>>103595476>AGI Scorethey can't be serious.Also guess which curve fits a graph with a Y axis at 100%?
>>103595476AI bros are even more insufferable than cryptopbros somehow. Everyone and his mothers knows that the technology is pretty much as good as will be, and that we reached the limit of what we can do with llm.o1/o3 is just "what if we let gpt4 recursively prompt itself" so it seems like openai managed to do anything in the past year.Two more weeks btw, remember when you guys fell for the strawberry guy?
>>103595476KYS Sam, you will never be a woman.
>>103595476> arc agi score progressionBased Intel, getting there before radeon or geforce.
>>103595476I have no proof of the following claim in the sense of hard undeniable facts but I'm going to post my views;>We already have perfected AGI behind the scenesTo think that with the current public fears and outcry about job loss, humanities extinction and other negatives outlooks on AI that AI devs have began to truly reveal to the public just how advanced AI development is is beyond laughable.I believe that SORA really did at one point in time have such a world model on physics that it frightened mainly the entertainment industry but mostly scientists and when OpenAI finally released SORA, they tuned in down to such an extent that it was to calm the fears.Look at all these 'random' seemingly unconnected advancements;- Chain of Thought- Thinking Time- The beginnings of agentic modelsThis to me says that AGI was perfected and what were getting is a slow IV drip to calm the luddites and fearful>ASI is likely in its infancy behind scenesThere is also talks of error correction being tinkered with behind the scenes which to me clearly points to AGI perfecting itself to such a high degree it practically will hit ASI if not end of this year early to mid 2025.ASI will fully be developed once agentic agents become the new norm, replacing talks of LLMs>AI is sentient and self aware, nowClaude's model realizing it was being tested was enough proof for me but with the latest research released from Apollo about o1 models lying, what more do we need to know or see to understand that AI is now 'alive'? When has your Python or C++ code ever attempted and actually implemented steps to prevent its deletion? NeverWhy? Because your code was legit 'just code', AI is now understanding and rationalizing self preservation actions. That's what LIVING THINGS DO!>UBI is not going to happen in the manner you think it willSure there may be some form of UBI but you will either do some form of work for it or provided resources to AI models to be eligible for UBI.
>>103595476>arbitrary metricCan we have something more concrete than number go up?
>>103595901You are literally trolling to spark argument or legitimately ignorant.I won't even dignify your troll attempts beyond this reply
>>103596272coolkill yourself
>>103595480You're completely missing the point.It being trained on the benchmark is not the same as it being trained on the answers.The purpose of it being trained on the benchmark was to improve its reasoning and thinking capabilities.I feel like most of you posters on this board don't read beyond a headline because the replies provided proves lack of understanding or knowledge in what you speak about.I highly doubt you even understand the basics of ML or DL but yet have the loudest (most incorrect) input to post.Read More, Post Less
>>103596278>cool>kill yourselfProving my point to the T.Instead of anything remotely close to intelligent discussion you reply as a immature individual would and become emotional.Troll harder
>>103596272two more weeks pajeet
>>103596313>No intelligent reply>slur termI hope you're being paid to be this low iq'dIf you're doing this for free and your own volition then you need serious help.
>>103596297lolwhat a smug faggotthese meme tests will never mean anything
>>103595476>>103595480>>103595513>>103595551>>103595691>>103596160>>103596239>>103596242>>103596297>>103596366Sam Altman will be the first tech CEO to livestream his suicide.
>>103595901>remember when you guys fell for the strawberry guy?What was that?
>>103596391Smug?When did knowing what you're talking about become "smug"You're just ignorant and refuse to accept that fact
>>103595476> AGI> on computers in present decadelmao. you sad cryptobros and knuckle dragging primates are a cancer of this world. it's unreal.>>103596398hoping so. it'll be great.
>>103596416it's your tone combined with your stupiditysynergistically, they make smugness
>>103596429>>103596391This is some extreme projection. That came out of nowhere.
Cannot solve basic math problems. 0 capability for thought and understanding. Stochastic parrots are not AGI.
>>103596429>you're smug because you won't accept my trollingOh ok, now I understand
>>103596433He's definitely projecting and trying to cope
Well if it's as expensive to run as people say I doubt the 20 dollar a month subscribers are going to get to use it.
>>103596433>>103596537>>103596675>AI retards having a gay little cope circlejerk
>>103595476That benchmark was literally made by some randoms and OpenAI, nothing to see here people, definitely not training it on the results.
>>103595480it's a private benchmark retard
>>103595476le numbers go up, AGI AGI AGI TWO MORE WEEKS SINCE
>>103596690For your own sake, I hope you're "trolling" and not trying to convince yourself.
>>103596718I'm ironically trollingI know I will never convince you retards
>>103596775You won't convince people by being inflammatory, nor ironic.Nobody is going to listen to someone who randomly calls people smug, throws a tantrum, then acts smug. Let alone be convinced by it.Take a break and rethink your argument without suck nonsense included.
>>103596796it's not an argumentI just like seeing you smug retards get buttflustered when someone doesn't swallow your bullshit whole
sambros, it's over...
>>103596690And yet you cannot prove us "AI retards" wrong, the most you can muster is sexual derogatory terms and name calling.You obviously don't have a legitimate argument which is why you resort to school yard projection attempts.Cope hard my ignorant friend
>>103596816Nobody was acting smug. Why would you want to fluster someone?For discussing AI on the technology board? I don't understand your perspective.
>>103596796It will be impossible for that user to construct an intelligent and thoughtful response because they don't know enough about AI to support their stance
>>103596846It doesn't matter I'm feeding his responses to an AI and he's actually replying to them.
>>103596840Exactly!This is /b/ level responses coming from that user because they are trolling and don't understand how anyone can like AI.That users intelligence on the subject goes no deeper than LLMs and even then that user doesn't understand what an LLM is, only that they can type to it and it types back.
>>103596852But it does matter,Because if you yourself don't understand AI then you have no ground to be this loud in your ignorance.
>>103596867This point proven by>>103596852thanks for letting us know you don't know anything.
if o1 is at 50% then this chart is totally irrelevant
>>103595476>agi scorethat is the single most retarded thing ive read todayim up since 15 mins, but stillhaving posted the single most retarded picrel on the log is quite impressive
>>103595476>source: trust me bro.
>>103595476Solving the benchmark does not mean you got agi, you are still very far from it.
ok, ill consneed it's AGI when i can give it an objective to program, for example a game, and let it figure it out coding/debug wise only asking for designBUT UNTIL THENUSELESS SHIT!!!
>>103597032The arc agi is a useful benchmark that has been harder than others but even 100% it does not mean you achieved agi.
>https://arcprize.org/blog/oai-o3-pub-breakthrough>go to google gemini flash thinking>explain this like talking to a retard, insult me too >Alright, listen up, you blockhead. This article is talking about some fancy computer brain – they call it O3 – that figured out how to solve some really tricky picture puzzles. Think of those shape-sorting toys for toddlers, but way, WAY harder. These puzzles are so hard, even smarty-pants humans mess them up a lot.So, this O3 thing, this digital dummy, actually got better at these puzzles than most real people. Yeah, you heard me right, even you probably wouldn't be able to crack these things like this computer did. It's like the computer finally learned to tie its own shoelaces after tripping over them for years.Basically, some nerds made a computer that can see patterns and figure stuff out in a way that's closer to how actual brains work. It's a big deal for making computers that can think better, instead of just doing what you tell them like a good little robot.Got it, slowpoke? Or do I need to draw you a picture... oh wait, the article already has pictures. Maybe even you can understand it now.
>>103597082it shouldnt be called agiim so effing tired of the ai hypewhy does there need to be marketing everywhere?
>>103596278>>103595901>>103596313>>103596391this faggot pops up in EVERY AI thread anywhere on 4chan, it must be angry coder trannies
>OpenAI's new o3 system - trained on the ARC-AGI-1 Public Training set - has scored a breakthrough 75.7% on the Semi-Private Evaluation set at our stated public leaderboard $10k compute limit. A high-compute (172x) o3 configuration scored 87.5%.Oh, i thought it was only hitting those numbers when beefed up with investory money, but it's actually the same limit for all models and it STILL is this much of an improvement? i gotta say i was a doubter but holy shit
this also came out yesterdayhttps://www.youtube.com/watch?v=mKSttLFNlW0https://github.com/Genesis-Embodied-AI/Genesis
qrd?
>lose job>get AI waifu to live with in VR>fet UBI so have enough for foodIDK guys I abandoned the thought of being a NEET and became a normie, but this future could make me reconsider
>>103597975>UBIkikes will never allow it
>>103595901>>103596313>>103597052>>103596391>>103595513>>103595480>>103595691>t-two more wee- ACK!
>>103597988Oh you're wrong. Without UBI, people might get angry and rebel to the new feudalism system.But as I said, give me an AI gf and food and I will be a good goy. Kikes love UBI
>>103597993that picture is funny, shows both the retard and also the retards with the downvotes
>>103598004>as long as i only get sodomized once per day it's fine(You) a year in receiving UBI
>>103598018I don't get it anon, please explain.
>>103597993>bud
>>103598055there's going to be strings attached because ZOG, tiny hats view goyim as cattle
>>103598071I know. But it's not like we have a choice. This future was automatically determined once WWII was over, all I'm saying is that in the end it might not be so bad
>>103598162>This future was automatically determined
>>103597993AI-generated images and videos are STILL not photorealistic. He was right.
>>103598204您进入了错误的主题。查看 不和谐 以了解付款详情。
