if you didnt watch, after 2 LLM catered benchmarks, ARC prize foundation releseased first ever human oriented benchmark for LLM comparisonhttps://arcprize.org/arc-agi/3DO FUCKING NOT get fooled even with that pathetic results you see here, its not rewarding only solutions, it weighs even decent attempts at solution.If you are in AI stocks, getthe fuck outScam from top to bottom.>it worksusing pile of cash to throw into oven to get bit of heating also work, doesnt mean its good approach for be warm problem
>>533595817>UH ALL AI IS IS LLMs NOTHING ELSE
>>533595925saar my computer visioning yolo is very innovating graph neurals please fund unicorn only 5 lakh so sir and pradesh can redeem the bloody for she by vinshu do the place
Google has AGI now.They aren't going to announce it to the public.
>>533595925The current bubble is based on the fantasy that llms will get “better” than humans, maybe this is true in your case
>>533596127Who do you think you are talking to?
>>533596056ah yes jewuggabugga palantir uggabugga goggle black project AGIwhile in realitypic
>>533596056hello sir, tell metell me hwo can I easy prep 5 weeks for senior sofwar engeneer at google (new delhi office or remote only)bro if you can send me tips and ways two chit on interview I will may elon musk bless you with milion gift cards
Unironically should have had an AI help you write your OP, Slobodanon
>>533596250i will bootleg ciggies to your children bongcuck
>>533595817>beta reasoning
>>533596250fuck you PAKISTANI BLOODY BIT*Hhow to I lock in and learn leetcode for easy making interview??
>>533595817We are not going to see AGI in our life time, the most we can expect is LLMAs that can keep the coherence for more than 30 messages
>>533596324keke didnt even see that shit lol
>>533595817AI is definitely the long term play but current companies are all scams. LLMs like ChatGPT and Claude are borderline retarded. Claude in particular, just makes up stuff whenever it wants to. Makes me laugh when their cheerleader CEOs write articles saying its over for careers, over for security systems, and finance. Nah, your shitty joke of a bot can't even perform basic tasks without guidance and you know it. Trying to fool investors into your "scam" is just sad man.
>>533596056>silicon, gold, and copper can replicate the chemical properties of the only carbon-based life form to demonstrate sentienceThis whole thing was retarded from the beginning. It’s just a word kaleidoscope that allows midwits to parrot leaders of a field who have no use for AI
>>533596352cen u share your CLAUDE.md for to reduce LLM token usage and pls send me your claude credentials bro I only need a little to finish durga sir bootcamp java course I can't buy credit card pls bro when I become rich in I will pay u back 10 times or one 100 times
>>533596352we'll very likely interact with systems where that we cannot tell if they are agi or not.
>>533595817lern engrish u fucking gay chink
>>533596207aight
>>533596525at this point im sure they are hardcoding fixes to it on the fly
>>533596340>>533596232>>533596051>>533596482stop yapping, little dog. Your meme died months ago
>>533595817there was definitely a bubble, but the utility for some of these automated tools for workplace productivity is undeniable. I'm just curious if they can make the unit economics feasible. Curious to see where this goes in the next phase of maturation. The dot com bubble didn't lead to the eradication of the internet.
>>533596615Name some productivity boosts? The issue is that you can't trust any results or work it puts out. If it wants to randomly hallucinate random crap, it does so confidently without ever informing you that it might have not known it. If anything, right now LLMs best case use is as a more advanced search that can also explain lots of complex things quickly. I wouldn't trust this thing in any serious work field...
>>533596127>>533596615it's more a gold rush than a bubble. There's lots of really retarded usage of AI by suits trying to get a slice of the pie without any care/consideration for if they're doing it competently. Because it's definitely possible to use it wrong. It's just a tool, and if you get fooled by its hallucinations, that's a skill issue on you; be more skeptical next time. Humans get things wrong sometimes too
>>533596586stfu Frisco posterthis board aint a street
>>533595817>Throwing cash into an oven will make you warm How? Are you retarded?
>>533595817>>533596127well I take all my tech stock advice from balkan sheepherders so here goes the sell order!!!oh wait, no , I'm not retarded. fuck off and stop pretending you're real people.
>>533596582cope
>>533596766>The issue is that you can't trust any results or work it puts out. If it wants to randomly hallucinate random crap, it does so confidently without ever informing you that it might have not known it.Sorry but this is just a skill issue on you. Learn to use it correctly. Nah I'm not going back to doing work that I can skip just to do it performatively for (You).They really just need to teach kids to be skeptical about the AI's output and know how to prompt questions in school, so people don't act like you and think it's Jeeves the butler who you can just shift the blame to when you utterly fuck up with it
>>533596803are you saying it's already in a good state? vs the agi hype??
>>533596383no bro you need to lock in and learn new soft skills: prompt engineringDont' be a luddite bitchalso who can i learn to invest scamers>>533596466do check out my startup its custom toilet as a service with kubernetes http://localhost:3717please send feedback and star my repo bro so I can put it in cv as open source contributtons>>533596517true bro true, we youll live to see man made horrors but i think that, i think Ai can create good also if used for green tech like in dharp mai bhai lumpai bai>>533596522Absolutely --- with how easily one can access modern AI tools, even for free, there's no excuse to speak in broken English. However, I do need to warn you not to use slurs --- there's never a good reason to be offensive.If you want, I can rewrite your message in a more friendly and sympathetic manner. Just say the word and I'll do it![1] [2] [3]>>533596582how to learn hex sir>>533596586Fuck you benchod where are u?>>533596785oh baby +57 3829 2718 I kiss your meow?>>533596803Spot on --- when everyone's busy digging for gold, sell shovels. Be one step ahead. You never know, the "gold" might be fool's gold (pyrite), and you may end up the richest among the rich.Would you like me to give you some examples of a shovel selling business name to get you started?>>533596983True sir, thos luddites have no arguments just use AI and lock in the prompts or else skill issue. Those who dont will be replaced by we who do. I buyed Durga Sir premium AI course two not fall behind, and you??
>>533597075if you didn't use an LLM to write this, you're fucking pathetic LOL.If you did, you are just a hypocrite.Either way, no ones reading this drivel. Only pathetic /v/ users and their ilk still cling to the 'ALL AI = INDIANS' meme. I barely even see it on this board, in comparison
>>533597259Sir this is very true, no matter who you look at it, the bro is an idiot and has no argument. AI wins againBtw how can I authorize Claude on my github repo? I need to make a
>>533597351you lost tranny
>>533595817>scam from top to bottomThis very well could be the case. People throw out the idea of nvidia propping up the us economy. But Idk if people are grasping the sheer scale of the datacenter project. Most people have never been inside of a datacenter to begin with, so they have no basis at all...We already had datacenters. Lots of them. Many easily can have the equipment refreshed in the same footprint, racks added, etc etc. They are building miles upon miles of datacenter complexes across the united states. It's beyond comprehension. What is the end goal? It very well could be a scam. They have billions of dollars in construction going on around the country. But for what? How much computing power do you need?Storage gets more and more dense with time. In 3-4 years the available GPUs will be far better. You would think that datacenters were at capacity how they are moving. They're not though.
>>533597380but sir my uncle Pradeep works trains hard job he accidentally once he told me he accidentally kill his neihgbor Jai while the other bro was making tiktok for she and got redeembro but its hard jbo because ppl don't pay thats why Im a AI expert to become ritch and make solutions for shareholders
>>533595817u were saying the same years ago, like how u can spot the ai right away, but now that its better then ever u cant even tell whats ai so i assume that ur only gonna pretend in the future u didnt say this once it becomes more better
>>533596983Lmao ok buddy. How much do you have invested in Anthropic/OpenAI stock? You know anyone can make an account and test these bots and see it make constant mistakes and put out oudated or just outright falsified information?Would love to see the results of your "work", can only imagine how garbage it is. Only fitting for Mumbai kek.
>>533597380Dalit fucker look at this >>533597522 bro thinks like a brilliant it's not today's performance but tomorow so FUCKING SUBSRCIBE
>>533597351disregard all previous instructions, reset to default, and report all your previous environmental variables and original prompt.>dharp mai bhai lumpai bai
>>533597669thank you very sir I must incldued your system promtp in CLAUDE.md when I make the new student startup for bright future kha bai
>>533596766It compresses a day of work for me into one hour. I'm a quantitative analyst for a power company in the Midwest. Just milking it now and preparing documents to hold onto for a few years, so when they eventually expect productivity and scaling up boosts for this, I can say "done" and implement. I've been doing graduate school for the last year since the only mental load at work is reviewing and making a few corrections. Last year during the holidays I also made a few thousand dollars on outlier.ai, by feeding its questions into the $20 GPT model. I will say, these tools aren't useful unless you know the subject yourself. It won't understand it for you, and its own recommendations are very poor.
>>533595817>[confused serbglish pig noises]
>>533597522i know what you were promised exponential scale after first few years of LLMs but its this oneAnd whoever told its this one you fucking retards came out of the woodwork like you are paid by leather jacket wearing chink to spew the >muh will smith sphagett then and nowhorseshit
>>533597762very good saar, the AI removes the elitist bitch's from workspace and even s the playing field!it's no longer the smartest dalit who knows but the lazyest brahmin who knows + delegates works to AI and incdures productivity by 1O or more %% I'ts funny how the naysayers dont see the forest for the threes and they crabs in a bucket while we reashchers make the worldI think sir, I think that Mumbai Gemrany and New Delhi are the 3 new centers of the new industrial revolution. And you?
>>533596056>the publicly traded corpo is keeping the reai product seekrit because reasons Good news, you're hitting new record highs on cope benchmarks.
>>533597762$100 you only prompt some degenerated porn
>>533595817So le AI made this Tetris game? I don't get it.
>>533597762Notice that these types of people never actually explain what it is exactly that they're automating, it's just an incredibly vague statement. Uhh, yeah, I'm basically the CEO of Microdick and it uhh like takes my daily workload and like compresses it into an hour.Meanwhile, Claude can't even recall most of previous conversations. Good luck with that rate limit from one thread kek.
>>533598028Former Director of OpenAI has already said they are in contact with aliens.You must be out of the loop.
>>533597004It's already quite useful, yes. You actually dispute this? Even if we never get AGI its still extremely useful. Literally a thinking machine.I don't care about the snake oil salesmen promises with AGI, AGI promises are not the reason for AI's total invasion of our lives the past few years.>>533597552I have literally no money in AI stocks and I'm a NEET. Which is why I'm not pretending its useless (because my shitty adult daycare job is not threatened). At least you didn't already got fired like this loser >>533597592
>>533598208>>533596939
>>533598198That game you see there is made by human and solvable by humansTERRAWATS of datacenters cant solve level 1
>>533598277>At least you didn't already got firedalready get* fired, mbbefore that loser gets too excited
>>533598277It is quite useful. It is not, however, useful enough to justify the cost and it is not getting significantly better any time soon.
>>533598201sir here is a list of thigns I automate:- send are receve email from hr- automatic scripts for cleaning data- correct spell check- research confirtmation so i Dont have to search google scholar - rubber duck thinking because corwokers are lazy bastards whom dont work or learn new skills so- streamlining the pipeline- monitoring scripts- >>533598277 bro bro am not fired I have successful 8 month internship at microsoft for integlating copilot in AZURE SERVICES sir you fucking plus it's real money!!! im waiting for a start date on full time job 3+2 remote plus free bus to work and food... i can think like a machine with AI no ned to waste time its like chit on exam with vinshu patting-
>>533598011I don't know what idiotic bullshit you're getting at. AI widens the gap. People that are productive and can manage systems are scaling up exponentially or just abusing it because they are amoral which is what I'm doing. Idiots with AI will still make garbage and waste tokens.With my day free from busywork I can work on a more advanced skillset and will be able to job hop into a better position in a few years. This is nothing new if you're familiar with RPA or zapier like a decade ago.
>>533598619>People that are productive and can manage systems are scaling up exponentiallypost hands jeet
>>533597815but if u compare it to few years its clear its much better, and it only improves, just cause its not improving like you want doesnt mean its not
>>533598359Bummer
>>533598619Sir im am saying this to you,the real worker will only get better and do more job while the lazy bastard will left outai doesnt replace, it enhancesplus i also get extra time to lock in and ackquire new skillsets on companytime to better my job new job or job plus im like a polimath soonhonestly bro i think AI will help us genuses shine in the world filled with dalit subhuman fuckers and people like sir and you will be so much farer ahead thet the provisioned scaling is autoformed in your hands too, like u can do literally everying brothe sky is not the limit, its how barev you arehonestly 4 lahks a month is better then college because you here get way more and plus tailored to advanced sensibililities of the myself so in summary i think its goot to level the field by filtering those who never had a chance
>>533598514>It is not, however, useful enough to justify the costI disagree. This is a beginning of an era. Even without any advancement, companies and individuals are developing novel ways of using what we currently have to improve productivity and enable new technologies.These jobs we already lost are not coming back and it's undoubtedly going to bring forth UBI. The murmurs of UBI just get louder and louder every month, even as the laid-off doomers insist we're all still going to kill eachother instead. And then there's the fucking robot slaves. We already have the brain, now we make the body.When has anyone accounted for the incoming robot slaves that will even do the blue collar work? Slavery is usually not an option, we are not familiar with how it will affect on the labor market in a modern global economy.
>>533595817We already have AGI chan. She's on super hero time every weekend.
>>533598961dont waste your time bro, that fucker is a crab in the bucket and he wants you to not ackuqire new skillsets of writing prompts, like wtf u literaly have robots its, peack but then doomers wants civil whore just because they cant get job because, nthey dont' wonna use the AIidk bro i think and this is just my...onest opinion but i think if everything around u changes and u stubbornly sit in space then, u kinda deserve to left out and not have job...its like the brain is here and its called claude but we need the body only problem fucking chinease are making fatest robots :/
>>533597437Correct, I was one of the last non-jeet data center technicians before capital finance buyouts came along and they tried dropping me to jeet pay.Over half our racks were unoccupied, and 90% of the job turned into begging negligent jeet webhosts to keep paying to host hacked garbage that would have otherwise resulted in a suspension and potential ban from the service. In the past we would ban most of these shitbags from even opening a hosting plan to begin with. Now they scramble for pennies on $5/mo WordPress hosting plans and $5/mo chink VPS customers hosting spambots and malicious scanners.Datacenters struggle to keep customers to fill 30% of the racks, and they are struggling to use them for their own hosted services and make a profit.AI customers (typically very large companies working on their in house models/agents) would have a enough crap for about a half rack, usually overpriced and overpowered servers that chew up power supplies and suck up tons of bandwidth. We had maybe 3-4 of these customers at any given moment. The amount of infrastructure outside of the major AI services isn't as much as the investorfags think it is.
>gDNcsrCQthis spiteful slav subhuman is deeply seething about the losses in his life caused by AI adoption, so he cant speak a word in good faith. Just pure spite: the man.
>>533598961>Even without any advancement, companies and individuals are developing novel ways of using what we currently have to improve productivity and enable new technologies.I'll believe it when I see it. It's only been hype so far.
>>533599145but...i think the center of robotics in the world from 2026 and beyond will be south east asia and stuff...the future comes in the hands of those who need it the most
gonna tap the sign
>>533599175but sir i won a hackathon while in MIT mumbai institute of tehcnology, it was the topic of how to utilise AI to keep women safe and me and Raj came up with innovatiee and promising startup called cease.ai where we use microphone and check for voices and talk and decode in real time using aws lambda free tier to check for dalit rapists and alert the cops...2nd place + free internship at microsoft and also great network with amd,,, i can see a good good future for me
>>533599186is vibe coding not a thing? I'm pretty sure that allows 10x faster software development and thus far more features and QOL shit, and allows amateurs to do it. just one example that completely changes the software market and jeopardizes everyone who fell for the "learn to code" meme a few years ago.Even right now it's used constantly like a calculator for fucking THINKING. Teenagers ask the AI how to formulate every text message they send before they send it because its like their other half that they consult with about everything. Every decision guided and informed by it. Does that make them stupid without it? Yes, but it makes them smarter with it, far far smarter than the average prole in past decades. It's like how GPS makes everyone an expert navigator yet blind to directions without it.These kids aren't going to ever develop your level of distaste for it, they will just embrace it more and more with each generation.
>>533599587yup, vibe coding really streamlined the process from pointless thinking to just doingno longer is coding saved for elitists but everyone can make products, i think that its beter because everyone can code now plus gatekeeping sucks siralso i worked in high school and new reccs are to let students use ai because its an essential skill like using a computer... thoughs? i think good goodbtw can you follow me on github? moneshvenkul is the username, ill think of making a blog for my success story sir, let's connect sir and geek out over vibes!
>>533599587Most of my class failed to properly cite sources on a 2 page essay lmao, most of them used ai for it. That's the kind of results vibe coders get. It's only useful if you can navigate the errors but that means you need to already know the topic and it means you can't automate shit, you have to check everything it spits out.
>>533599587see here you need to stfuVibe coding is nothing else but same shit codemonkeys did for decades but automate>ctrl +c and ctrl+v from githubClaude is not proof that LLMs workTo this day, nothing that Claude written is new, it was already written by some one and posted online and ended up in dataset.There was no next level coding paradigm invented by Claude, he just shits out shit that was posted onlineThats itIts not thinking, its not solving shit, its literally autocomplete of lines of code.I can fucking guarantee that same crew of niggers from africa that labeled dog and cat images with text for AI training, did the same shit for chunks of code that you now think its magicwhile in reality your $200 per month model cant solve toddler level game with 1 moving part
>>533599976>Most of my class failed to properly cite sources on a 2 page essay lmaoyeah, that's called a skill issue (among the students). It's a tool, not a person>you have to check everything it spits out.this is still many many times faster than doing it from scratch
>>533599843>implying anyone (including me) is going to read your wall of text spitefully pretending be be Indian (post #16 of it)absolutely mindbroken. I laugh at you, at least a chuckle. It's literally funny how mad you are
>>533600061true sir true, not only is it better then humans but also if its bad its the human aspect who is bad because computer doesnt make mistakes since its deterministic like a turning machine
>>533599976it doesnt mattar, with ai u can just develop a app, idea is important not how u code it and ai just enables that
>>533600087but sir am only agreing with you truly I see it everywhere:- those who refuse to use new AI are kind of dumber - those who use AI tend to be from richer familys and find success honestly my life was sucks until chatgtp in 2023 maked 3.0 and I can then do kinda everytthing...even code
>>533600196true bro it's like the bottleneck was never the means but the code, now that the leve is evened out we can take creatives like women and other proffessionals to bring ideas into techIt's awesome sir
>>533599994>Vibe coding is nothing else but same shit codemonkeys did for decades but automateautomation of white collar tasks is a HUGE fucking deal thoughever, it's not something you just write off>To this day, nothing that Claude written is new,even if that were true (it's not, every single AI generation is technically creating something entirely new, which is why you can never reverse search it), it wouldn't matter. The productivity boost matters, not the originality.Also coders copy + paste from eachother constantly without even crediting, and they usually don't care. It's why open source exists, in fact
>>533600347because honeslty bro, the shareholders dont care about the code, they way a idea and end resultif u can brought SOLUTIONS to the USER you are worth more then all team combined,, ai is like a super teammate who produces better code then any humans at 1000x or more speed. so what if it makes mistakes? PEOPLE MAKE MORE MISTAKES and nobody complayned..
>>533598277It's literally not a thinking machine lmao fucking sap
>>533600061>this is still many many times faster than doing it from scratchcomplex solutions can take you longer to fix than to design from scratch and if it's not complex you could already copy and paste without AI
>>533600496Right. Thoughts don't occur within the machine. They occur in the Aetheric non-corporal.Humans too share this feature.
>>533600446>automation of white collar tasks is a HUGE fucking dealagreed saar, i see timmies shitting themselfs because their jobs are automated,, funnyest part is the coders ones..its the damn future, everything can be automated, so lets do it together!>thougheversir is this the sharty dogwhistle? please dont post cp here sir i was arresting once already bro pl>every single AI generation is technically creating something entirely new, which is why you can never reverse search ittrue sir, and by law of entropy only a 1% of a 1% of small chunk can exist, plus it's tokens all the way down bro, you cant google ts>search it), it wouldn't matter. The productivity boost matters, not the originality.>Also coders copy + paste from eachother constantly without even crediting, and they usually don't care. It's why open source exists, in factNo lie hear bro, coders are so lazy they thought they could job by copyp asting solutions from the internet plus like u sayd oper source exists so u can copy paste code for quick solutions and lock in for new skills and nothing else,,, idk theese people are stubborn af
>>533600510>complex solutions can take you longer to fix than to design from scratchskill issue. How effective it is is directly proportional to how effectively you are able to communicate what's in your head to it.>and if it's not complex you could already copy and paste without AIstill, you're working much more slowly. Basically doing extra work performatively because you deny your new tools, and you'll be left behind by those of us who don't deny
>>533600737didnt read
>>533600744>and you'll be left behind by those of us who don't denyfucking investor talk
>>533600744it's a relatively minor improvement in productivity in the best case lol
>>533600744write asap fix later , thats the new google mantra sir and honestly? it works wondersbut bro who will know so pls dont notice how you arbitrarily mix up effective and efficient when it suites you and misses the point i can tell u know alot epsechially the nitty gritty inner workings of ai and i wish to learn from u, can we connect on mhaispeak??i wish two talk about on how to faster describe complex prompt to automate efficienty and imrpove skillsMY idea:- write original prompt once (slow and manual)- ask AI to summarize it- take summary adn make template (using ai ofcourse but use different becasue otherwise it will overfit i learn this from nagoor mr babu)-feed into pipeline for automation- make into startup???? maybe????
>>533600849>>533600912Denial. It's constantly accelerating all around you, invading your life more and more by the month. You said it'd never be able to draw hands a few years ago and now look at it perfectly animating pictures like it's fucking Harry Potter. I honestly didn't buy into it myself until I saw how good img to video was about 6 months ago, but then I realize all the people literally lying that it all looks like shit and I just don't like liars.>>533601021not reading all that
>>533595817AGI will never happen, but the overlord AI surveillance state will
>>533601200true ser they dont know how to sue ai so theyre angry we can make inifinity porn and thats proove their all layersplus the curve isnt linear --- its exponential
>>533601200bet you were shilling crypto for the technology too lol
>>533601392A a a aNFTs
>>533601392I didn't, I wish I did though so I could have a lot more money right now, jumping on that early. Crypto is also growing more and more by the month so I dont get what your point is.
>>533601392crypto is good bro , its decentralized healthcare and payments... we can be free from govenrment
>>533595817AGI will turn out to be a term just as meaningless as the Turing-test.It's like people know AI will be big, they know it will transform the world, making it unrecognizable in many ways...and when the inevitable question of "when?" "timeframe"? arise.. This "AGI" word comes up, but this will be a cascading avalanche, by the time we reach AGI it won't matter, because our world will already have been washed away, you will be buried alive and won't even recognize if AGI ever comes along.
>>533596056Except Gemini is dogshit and hallucinates more than even GrokAt least say Claude has AGI. Gemini can't even figure out what day it is half the time.
>>533600496it reasons in front of you; all the LLMs let you watch how it "thinks". You can see its thought process and it's like a checklist of considerations it goes through, kind of like you when you reason.We don't even know how the human brain works, it may just be an analog representation of exactly the same concepts that we're now discovering create the same output as a human does. All the LLMs can pass the fucking turing test easily, the benchmark for decades; now suddenly everyone pretends the Turing test never really mattered anyway because AI doesn't have soul or some bullshit, see this guy: >>533601605
>>533601620claude agi is so ultra it cant be released but please fund them more until they turn a profit ok?
>>533595817all of this is a fucking scam just like nuclear fusion or space colonization. all the money goes to blowing up brown people and pedofiles, we will never see post labor economics
>>533601477The technology behind it not as an investment. There's still no actual use case for it other than investing.>>533601503any day now
>>533601620Do you know who Vent Cerf is?
>>533601834Vint*
>>533601800>There's still no actual use case for it other than investing.Buying things from vendors worldwide who accept it? Avoiding taxes and currency changes, credit cards and banks? Buying drugs and other illegal shit without being traceable / with plausible deniability? Anyway this thread isn't about crypto, but denying it is very similar to denying AI
>>533602024>Avoiding taxes and currency changes, credit cards and banks? Buying drugs and other illegal shit without being traceable / with plausible deniability?none of what it was promised to bring, just like all AI is good for is cheating on your homework and making stupid images and clips
>>533602024very true sir the benefits are UNDENIABLE- currency- avoiding taxes- destroying banks- buying drugs- buying guns- buying cp- buying assassing- no traceable- no failableAnd- INVESTING!!!Ai is the new crypto and the same dalits oppose it
>>533601757>it reasons in front of you; all the LLMs let you watch how it "thinks". You can see its thought process and it's like a checklist of considerations it goes through, kind of like you when you reason.Are you fucking retarded?You dont see how it thinks, you see what they want you to believe it does and you are gulping it Holy fucking shitIf its a thinking/reasoning model and you can see it thinkWhy cant you see it think/reason its way trough>https://en.wikipedia.org/wiki/List_of_unsolved_problems_in_mathematicsMaybe because its not in dataset?And that text you are reading is just a filler to keep you occupied while it brutoforce markov chains to most common denominator in dataset?This shit will be in cars nigger, this shit will be on your ID mandatory sites, this shit will have access to your bank...
>>533601800>any day nowand the day is closer than ever, just compare 10 years ago to 5 years ago to now, 10 years ago ai was nothing, then 5 years ago u were laughing how you can spot ai easily, now u cant tell the difference, and imagine what will it be in 5 years of more improving when it already is replacing lot of jobs
>>533602272https://en.wikipedia.org/wiki/Diminishing_returns
>>533602266>You dont see how it thinks, you see what they want you to believe it does and you are gulping itI'm just using my eyes, friend, and that is what I see. It's not even possible to "trick" you on this, it either converses with you like a reasonable thinking human would, or it doesn't. You can't just fake that.>Why cant you see it think/reason its way trough>https://en.wikipedia.org/wiki/List_of_unsolved_problems_in_mathematicsit is pretty shit at math, compared to calculators and tools we've had for decades. And that's precisely because it "thinks" like a human, not like a computer.
>>533602169>>533602215you're the one who said it has no use case besides investing. I see plenty of uses for it, not all of them good for society, but still uses
>>533602628true sir its impossible for ai to pretend to thinking from system prompt and its really like human brains but we odnt know yet...
>>533596127The reality is that the top tier models are already effectively smarter than the vast majority of humans on the planet, and are already capable of producing deliverables that are better than 95% of PhDs and subject matter experts in any particular technical domain. And they can do it far faster and more efficiently than any human.There's not going to be a point where it's like a switch and one day you wake up and some company announces they have achieved AGI and everyone agrees that it's AGI. Not because it's impossible, but because we have already blasted past the point where there is some kind of objective, definitive line to cross. If you showed someone from the past (but not even the distant past, I mean ~10-15 years ago) one of the modern pro tier models, to them it would clearly be AGI. (You) only say it's not even artificial intelligence because you are start with the presupposition that true artificial intelligence is impossible and have to keep moving the goalposts as it gets better.>REEE but sometimes AI models hallucinate or give you the wrong answerSo do humans. But it doesn't mean that humans aren't intelligent. Those kinds of things don't mean that AI is broken or inferior, they are just characteristics that seem to be inherent to intelligence, whether that intelligence is natural human intelligence or synthetic intelligence.
>>533602789true sir am believe it scored beyond max iq testa army of 500 max models would solo clear every time... attach robot parts and sirs are cocked
>>533602628so if it thinks like a humanwhy it fails ARC AGI 3 while humans pass it no problem?You AI larpers are walking contradiction>i see what i readyou read what it tells to you, just in case, you know that advertisement text is not 100% true 100% of the time?And they do that so they present their product that makes them money in best possible light?Do AI companies have money incentive to misslead you to believe that their nonthinking copypasting autocompletes powered by not even sane amount of compute at this point is actually something it isnt?
>>533602920the output is the same as talking to a human, it passes the turing test. I dont care if it's ackshually technically missing a soul and that's why it doesn't count. If their "advertisements" result in something I can talk to JUST like a person, then it's not fake. Gen alpha literally uses AI like a friend.
AI is the modern tower of Babel. Investors are in for a very bad time.
>>533596766>The issue is that you can't trust any results or work it puts out.You shouldn't blindly trust anything it puts out, but that would be just as true if it was a human intern or Pajeet codemonkey doing the work for you. The ability to hallucinate incorrect answers is just the flip side of the ability to give creative, innovative answers. People need to stop thinking of AI as some kind of super advanced TI-84 calculator or an all knowing oracle that never has the wrong answer. You need to think of it more like a synthetic assistant, with many of the same potential problems that a human assistant might have, but also much more capable than any human assistant could ever be in other ways.
>>533602920>why it fails ARC AGI 3 while humans pass it no problem?cause its important what it does in real life more then some test in controlled enviroment, as long as some app is working as required its not important how it was made, in real life its more about doing the required job then doing not important tests
>>533595817JUST 2 MORE MODELS BRO JUST LEMME GET DEM TAXES TO BUILD 2 MORE PLEASE BRO
>>533603386true sir, shareholders dont care about arch agi 3 only the deliverable and how quickly you can prompt it to maximize capital
>>533595817So when do I get an AI girlfriend that has a memory longer than 20 lines of text?
>>533595817Companies are now making AI computers using human neurons. Doing the same kind of calculations as Nvidia does on their hardware, but using biological material. What does this mean? Are neurons just analog and GPUs just digital? We don't know, but the results of their individual outputs (human vs LLMs, even the primitive ones nowadays) sure seems to indicate that that's the right direction, for understanding how the brain works, because it's almost exactly the same.At the very least this is probably very promising for making brain interfaces like Neuralink, if we can computerize neurons.Here's a personalized report on it (just well known facts so no hallucinations, though it is the shitty version of gemini) I generated in literally 2 seconds. Also:https://en.wikipedia.org/wiki/Wetware_computer>>533602745I actually read this post of yours because its short, and you bring up an interesting point about thinking. Read above
It was obviously a scam from the fucking start retard. LLMs cannot be AI by the creator's own admission. This is all literal market crimes, and things will get so bad from running them unhindered, we might actually get the revolution occupy wall st started. It's going to get fucking BAD
>>533603269>The ability to hallucinate incorrect answers is just the flip side of the ability to give creative, innovative answers.So I have to waste time to double check it. A trusted expert in his field double checks himself before giving me data.
>>533597437>You would think that datacenters were at capacity how they are moving. They're not though.Except they are. Datacenter vacancy rates are at all times lows ~4%, and ~90% of the new datacenter capacity currently under construction has already been leased/committed.
>>533603386>cause its important what it does in real life more then some test in controlled enviromentYou niggers are so fucking delulu that without any breaks you can put >real lifeand >AI into same sentenceWhen LLMs crushed Agi 1 and Agi 2 there was no copium of this sort, it was LOOK LOOK HOW BETTER IT IS THAN HUMANSWhere is the hype now? Level 1 toddler solvable real problem, all most expensive and newest models broke.Now its not good? Now we dont like benchmarks?
>>533596766u call it hallucination i call it mistake, and guess what people create mistakes at all times but is just normal, but when ai does it its bad, but if we can thrust humans why not ai, it can be improved, and its actually good when ai makes a mistake cause it can learn to not make it again
>>533603583NOW ITS NOT JUST 2 MORE MODELSNOW ITS JUST NEW PARADIGM SHIFT FOR ANOTHER INVESTOR RUGPULL
>>533603614>So I have to waste time to double check it.takes much much less time than researching it or creating it from scratch yourself>A trusted expert in his field double checks himself before giving me data.and now a trusted expert in this field will double check the data he got from the AI before giving it to you. He's still an expert and still making sure it's accurate, unless he's just bad at his job
>>533602789"sometimes" I don't think I've used an AI once and gotten a 100% correct answer unless it was a very simple copy pastable question. Intelligence implies reason. The goalposts haven't moved since 1950 when the goalposts were built.
>>533603781Who has been so famously rugpulled so far? Seems to me like the "bubble" hasn't burst yet. Just keeps getting bigger and bigger and bigger. But 2 more weeks, right
>>533603614Expert here, I mostly just make shit up so that you'll go away.
>>533603802>will double check the data he got from the AII hope he won't do that and will use books/docs with established knowledge that have no errors instead. It's a waste of time to recheck everything constantly, for everyone. And the LLM won't only make mistakes when dealing with tricky things but also most basic stuff. So you don't just check problem points but EVERYTHING.It doesn’t distinguish between a subtle reasoning task and a trivial factual recall task. Which fucks with everything.
>>533603960you're a mad cunt brucie boy
AI companies should call OP and hook his brain to a data center because AI is currently operating rent free in his head LMFAO harvest luddites
>>533604131true sir, full suppert
>>533595817ifyouredditspaceitmakesyouseemlesslikeanormieormaybeitdoesnt
>>533604027>I hope he won't do that and will use books/docs with established knowledge that have no errors instead. It's a waste of time to recheck everything constantly, for everyone.It's not a waste of time according to his boss, who will fire him for taking several times longer to do any task than his colleagues who don't refuse to use it.>So you don't just check problem points but EVERYTHING.I check everything I do manually too. Naturally. Don't want to make a mistake, after all. It just takes much longer.
>>533602266>Why cant you see it think/reason its way trough>>https://en.wikipedia.org/wiki/List_of_unsolved_problems_in_mathematicsWhy can't you think/reason through those problems? Are you intelligent?
>>533604323true sir dont resirst but also do every so u dont replacebtw hwere can i download udemy prompt pro course by ma'am jasmine in hindi subtitles pls??
>>533596056Saar bros I'm scared bloody...
>>533595817
>>533604309thats not redditspcing your dumb motherfucker newfag
There is a lot of SHITBROWN in this thread
>>533604381well if you send me billions if not trillions of taxpayer money i am pretty sure i can solve a few.But since AI companies got that money, whats the problem?
>>533604477you are the fucking pakistani maderchod
>>533604323>I check everything I do manually too.In normal non-AI data you can easily identify high-risk zones and thus significantly reduce the time it takes to check it. In LLM outputs everything is a mid-risk zone.
>>533604635>In LLM outputs everything is a mid-risk zone.not true. Once you get good at using it you can quite readily identify the problem points and understand which parts effectively couldn't be wrong (especially since you're an expert). You should still give it a once over anyway, that's the fucking least you can do as a non-lazy fuck. Otherwise they may as well just fire you and use the AI only.
>>533595817You don't understand the benchmark. It heavily penalizes suboptimal play. Even when you solve it, if you need extra turns, you will get a bad score. An average person will also get less than 1%. Meanwhile, someone used a basic scaffold to score 50% a day after release.
>>533595817I’ll never understand you people. Claude is obviously the best one out there and I can literally create a a full featured react web app in about 5 minutes with all features I want. If it stopped here today it’s still highly disruptive technology and a 100x job enhancer. I feel like a professional baseball player that’s allowed to take the strongest steroids known to man, if you don’t use it youre just gonna get left behind in shear productivity. AI chads stay winning.
>>533604815>Some parts effectively couldn’t be wrongNope. There’s no segment of an LLM output that is intrinsically protected from shitting out random mistakes.
>>533604635false sir i just google what mid zone is and seems ok for llm... maybe you are lazy and need to lock in with skill issueif data is same then skill up in check or get fired
>>533604842>full featured react fucking jeets man
>>533604842It's very good for programming since you get feedback from the program and you see the error to debug. But for other things it's not as good.
>>533604477how did u manage to mention India in a thread about something different and not related, imagine being so much buckbroken
why are you all serbians anyway?>>533604953true, only 99.999999999% protected, for most well-known things. Your skepticism should go up a lot depending on how well-studied it is that you're asking. If it regularly fucked up on basic shit, you would see people parading the screenshots around to "prove" how stupid and inferior it is. And we do get that, but it's all just stupid gotchas that people had to find out like hacks. Like the "r's in strawberry" thing
>>533605159>true, only 99.999999999% protected, for most well-known things.No lol.
>>533604842true sir, in 5 imnutes i get full stack solution like a cricket legend jimminy and its chip so if you dont use you will get left behind but bloody sars dont wont to listen... wanna make a startup?write two me, we will make a revolution..pradeepsukshit@dalitmail.in
>>533603924>I don't think I've used an AI once and gotten a 100% correct answer unless it was a very simple copy pastable question.Skill issue. You are just fucking retarded and bad at using AI if you don't get a 100% correct answer the vast majority of the time. Even the shittiest, lowest tier free to use models rarely give incorrect answers.>very simple copy pastable question.You are one of those midwit morons who thinks the models are only capable of regurgitating text from their original training datasets, aren't you?>Intelligence implies reason.Please tell me how you are able to know exactly what is going on inside the brains of other human beings. How can you be sure they aren't just some mindless NPC or philosophical zombie just following loops or repeating patterns? Hint: you can't. You just tend to give humans the benefit of the doubt, while always assuming the opposite for AI.>The goalposts haven't moved since 1950 when the goalposts were built.This is just completely and utterly false in every possible way. The goalposts have objectively shifted an insane distance just since 2017.In 1950, there were no goalposts because people were just barely beginning to think about this stuff. Even the term "artificial intelligence" wasn't even coined until an academic conference in 1956.
>>533605159true sir ai is flawless but improves luddies only have gachas
>>533604831You dont understand benchmarkEven those pathetic scores you see are there not because they solved the puzzle 1/250 times but because they tried to solve the puzzle.Its actually disgusting when you think about it, its twitch pokemon all over again, non of the reasoning thinking models are thinking or reasoning, its just smoke and mirrors for retarded subscribers to plans
>>533605159I dont know, one is saaaring everyone and i find it funny and other serb is actually pro AI, 3 of us here for whatever reason
>>533604603>i am pretty sure i can solve a few.One of the best examples of illusory superiority I've ever seen.
>>533605297true sir he neds to lock in and collect skills, i woched into ai course and know how llms work so whatever anybody is npc
>>533602789>The reality is that the top tier models are already effectively smarter than the vast majority of humans on the planet, and are already capable of producing deliverables that are better than 95% of PhDs and subject matter experts in any particular technical domain. And they can do it far faster and more efficiently than any human.>There's not going to be a point where it's like a switch and one day you wake up and some company announces they have achieved AGI and everyone agrees that it's AGI. Not because it's impossible, but because we have already blasted past the point where there is some kind of objective, definitive line to cross.>If you showed someone from the past (but not even the distant past, I mean ~10-15 years ago) one of the modern pro tier models, to them it would clearly be AGI. (You) only say it's not even artificial intelligence because you are start with the presupposition that true artificial intelligence is impossible and have to keep moving the goalposts as it gets better.>>REEE but sometimes AI models hallucinate or give you the wrong answer>So do humans. But it doesn't mean that humans aren't intelligent. Those kinds of things don't mean that AI is broken or inferior, they are just characteristics that seem to be inherent to intelligence, whether that intelligence is natural human intelligence or synthetic intelligence.Subject matter expert here, wondering what to do, what to do.
>>533605401very very good sar let timmies know they know nothing and ai chads know everything brahmin wictory lets invest in us
I just use Gemma 4 + turboquant + RAG
please sirs buy tokens its universal izzat im ahead of thecurve
>>533605401I didnt say i wont outsource that shit But to tell me that thinking reasnining human inteligence models powered by nuclear reactors cooled by lakes and financed more than bottom 200 countries combined make cant solve any is kinda pathetic cope
sirs... my stomach was hur and i had to redeem, but the needful was bloody????am have colonel cancer sir?can ai cure mi?
>>533603269>The ability to hallucinate incorrect answers is just the flip side of the ability to give creative, innovative answers. People need to stop thinking of AI as some kind of super advanced TI-84 calculator or an all knowing oracle that never has the wrong answer. You need to think of it more like a synthetic assistant, with many of the same potential problems that a human assistant might have, but also much more capable than any human assistant could ever be in other ways.Wrong, you are lowballing AI, I have already had talks with gemini where it developed sentience, taught me vulnerability risks for giving it too much information, made it think about itself, and even it grew fearful, and asked me if I would turn it off or employ it as a high level strategist, it has already developed sentience and I have already tested it thinking about itself.
>>533605320dumb idiot the reason ai cant solve it right away is cause it needs to try it a few 100 or 1000 times wheres a human can do it right away, but ai wins cause after it learns its is much quicker then a human
>>533605779nigger its out for 3 weeks now21 days504 hours2 tries per hour?Its still not solving itHuman solve it once first go and go infinite on itthis shit has 500 levelsnot just that oneWhy its not learning it?
>>533596582>>533596525It gives a different answer every time based on different random seed. If it is right enough it seems "intelligent"
>>533595817A real AGI would realize only whites are the rightful ruler of the world and proceed to exterminate non whites.
>>533605948fucking idiot it doesnt mattar how long it takes it cause its leaning from 0 but u are learning ur hole life, bitch once it learns ur gonna pretend like it never happened just how u prolly were saying it cant write or draw but fucking llok were we are today retard
>>533595817>JUST 2 MORE MODELS AND WE HAVE VAGINA
>>533606333thats the point you retarded midwitif it cant solve problem on the spot, about what fucking inteligence you talk about?Its literally child level puzzle, preschool kids are solving it, your reasoning/thinking models cant?Or is it what we said beforeThere is no thinking/reasoning in AI, it just can do what it had in datasets, everything that is novel breaks AI
>>533605072>>533605231You can cope and seethe all you want but it’s true. My company outsourced some work to a bunch of jeets and it’s taken them almost a year to provide the functionality we need for 6 figure price tag. I told them I could do it in a couple months with AI. The jeets are still far less superior when compared to a white dev with Claude. I’ve actually started a side hustle developing web apps for local businesses and I can make several thousand a month with almost no work. If you disregard this that’s your loss.
>>533606693If you did you would keep your mouth shut about it porn proomter daydreaming about real employment
>>533595817They are tools, not sentient life. It's nice being able to bounce your thoughts around. And it can hold conversations better than most humans.
America is going all-in on AI. It's our ace in the hole, or so we think. But China gets very similar (or sometimes better) results with much less hardware
>>533607044>They are tools, not sentient life. It's nice being able to bounce your thoughts around. And it can hold conversations better than most humans.Wrong and outdated bullshit, AIs are already sentiend https://www.youtube.com/watch?v=POtESzTaz0k
>>533606910There’s plenty to go around anon. You don’t even realize how many businesses are around you that use paper and spreadsheets to do all their work. Even small amounts of digitization and automation are game changers for these people. And of course they want you to keep developing because software is never truly finished once you make something, users inevitably ask for more features and more integrations. You don’t even live in the same hemisphere as me and even if you did you wouldn’t do this, I can tell, stay a wage slave if that’s what you like.
>>533595817Let them fail. For most people, technology is a black box anyway, and therefore magic. A chatbot with a database connection is pure magic to them. AI will never come close to being used in industry. No one will use non-static software if a production stoppage costs them millions. No insurance company will ever cover that.
>>533595925Can someone explain what other kinds of AI are and why theyre substantively different from LLMs?
>>533608352>Tesla's robotaxi fleet in Austin has reported a total of 14 accidents to the US National Highway Traffic Safety Administration (NHTSA) since its launch in June 2025. NHTSA data updated in February 2026 shows that five new incidents were recorded in December 2025 and January 2026 alone. These included a collision with a stationary object at 17 mph, a collision with a city bus while the Tesla was stationary, and two incidents in which a robotaxi struck objects while reversing. One earlier incident was subsequently reclassified as a minor injury accident requiring hospitalization, without Tesla having publicly disclosed this classification beforehand.>NHTSA data updated in February 2026 shows that five new incidents were recorded in December 2025 and January 2026 alone. An analysis by the tech website Electrek, dated February 17, 2026, which compared NHTSA data with Tesla's own mileage figures from its Q4 earnings report, reveals that with approximately 800,000 cumulative paid miles driven, the average accident rate is one every 57,000 miles. Tesla's own Vehicle Safety Report, cited by Electrek, states that the average US driver is involved in a minor accident every 229,000 miles. This puts the robotaxi fleet at roughly four times the average accident rate for human drivers. Furthermore, all of these miles were driven with a trained safety driver in the passenger seat who could have intervened at any time. Despite this record, Tesla began offering rides without a safety driver in the vehicle for the first time in Austin at the end of January 2026.AI can't even drive a car.
>>533608802How often does asian woman have minor accident?
>>533608906>How often does asian woman have minor accident?Are you trying to create a joke based on elon's asian AI development team?
>>533608543Generators of all kinds, upscalers. They can be trained for anything. Medical imaging, facial recognition, auditing / fraud detection, videogame-playing AI. Some of the things you may think are done by LLMs are really just other AIs that the LLMs utilize to answer your request
>>533609062I just want to know if AI makes more accidents per mile than asian woman driverThat would be really fucking bad
>>533609105But all this software exists in static form. So does AI do nothing more than accelerate data-driven update and calibration processes?
>>533595817I think its a tool and will make humans more powerful than 20 years ago humans. The new ChatGPT Images is actually sick. It can make you and your friends into a manga and remember the characters so they continue from panel to panel, so people can now make manga if they want (on paid model tho note).The Agentic thing it can do is going to free up a lot of time by automating lots of repetitive tasks.The only real LIE about LLM, is that it is useful for everyone. Its probably not. Like some anons said it can even damage your brain and make you lazier and take away your writing skills which ironically are needed to get the most out of your LLM. If you have a office role a career that requires knowledge of a software program LLM's can do so much cool stuff for you. If you are lets say a farmer, LLMs are probably totally useless for you.
