https://www.youtube.com/watch?v=g5u2y0BwRJ0uh fellow snailcats? our response?
>>109747004>uh fellow snailcats? our response?
>>109747004ok it figured out the most basic puzzle game in existence. the fact that ai couldn't even play video games for children was an embarrassment, fixing that particular hole doesn't make them "intelligent".
still cant beat pokemon
I don't fucking care. I am tired of this dumb clearly inorganic spam.
>>109747004why are there random cuts in the """gameplay""" """footage"""
>>109747029It's not spam. You guys were claiming AI was flatlining and would never improve.You were proven wrong now deal with the aftermath.
>>109747053AI is flatlining. There is a fundamental barrier to the transformer architecture. I don't know why you don't understand this.
>>109747059See what I mean? You guys are still in denial.>There is a fundamental barrier to the transformer architecture. There is exponential progress every year. Models are getting faster, cheaper and more powerful all the time.
>>109747004Perfect.I'm going to make a shitload of these videos and spam upload them to speedrun.com under the TAS category. Like my unmanned AI channel on YouTube, I'll keep churning out videos and none of you will stop me.
>>109747004Why would I want to watch an AI play the game instead of doing it myself? It's just as gay as watching a streamer do it.
>>109747065You're in denial that there are people who are more talented than you, objectively smarter and more evolved than you. I absolutely cannot wait for the day you break down crying and finally decide to hang yourself. :)
>>109747004cool. i did that when i was like 10. fucking retard
>>109747080>You're in denial that there are people who are more talented than you, objectively smarter and more evolved than you.And AI is smarter than all of them so what's your point?
There is absolutely nothing that "AI" can do at this point that will be impressive when considering the amount of money that has been spent on it. It would literally have to create Star Trek IRL for it to have been worth it.
>>109747004>we spent trillions of dollars to create a robot that can play through a short 19 year old puzzle game for the low price of $500>you're not supposed to think that this is a huge waste of time and moneyHow about instead of a robot that plays my video games you make a robot that cleans my house? Why the fuck would I pay for 500 fucking dollars worth of tokens to make a robot play through a video game? That fucking robot should be the one paying me to play my video games. Fuck off clanker go do the dishes i'm not letting you use my steam account
Video games have had AI players for many decades. They are usually better than this.
>>109747131>>109747176Very little money has been spent though. OpenAI has lost a few tens of billions at most with inference profit margins around 70%-90%, so it would only take a few years for them to recover their losses and they know this, which is why they’re re-investing so much money for training. Oh and there’s the fact that they let a billion people use the service for free. I know you want to cope and believe that the scary AI will go away but it won’t happen.
wake me up when it can beat the talos principle
For god's sake, can't you guys just be happy that something cool is being made?
>>109747245NoNext dumbass question you stupid nigger
>>109747245these people thought it was so cool when Tony Stark asked Jarvis to write code. They bought little toy figures of r2d2 and C-3PO. But now that this tech is actually emerging they are seething for some reason. Like, did they think this was never gonna happen? That it was just a fun fantasy to indulge in while watching their cape fiction? And do they actually think that humanity will collectively forget about this in a few years?? These people aren’t real I tell ya
>>109747295They didn't realize they'd lose money.
>>109747352>research and development = losing moneyInference is profitable and has been for about a year now. Astra is extremely token-efficient and even if its price for customers were doubled then it would still be cheaper than paying humans to do the same tasks. And the Vera Rubin and Jalapeño chips haven’t even been in operation yet. We’re just getting started. They’re “losing money” now, but soon they’ll be printing $100B+ in profits every year
>>109747295> But now that this tech is actually emerging they are seething for some reason. Like, did they think this was never gonna happen? That's exactly why I find the anti-ai movement so weird.Did these people never watch The Jetsons growing up?We had so much entertainment showing robots doing everything. Now that it exists, people are bitching about?Come on.
>>109747295>when Tony Stark asked Jarvis to write code. They bought little toy figures of r2d2 and C-3POOnions overload.Leddit might be a bit more your speed
>>109747387Did you read what I responded to? I'm talking about the people seething over robots.
>>109747486that’s what you get for vagueposting
>>109747410>We had so much entertainment showing robots doing everything. Now that it existsWhere can I buy my maidbot that does all my chores? What companies are offering george jetson's short work week without a reduction in pay and benefits?
>>109747069You will lose thousands of dollars only to be near the bottom of the leaderboard with no earnings.
>>109747410Because the writers of these were dumbass optimists. Most of the tech in Jetsons is only good because of worker protections and seemingly infinite natural resources, but in real life if AI can automate half of your job successfully then half of your coworkers are fired and you do double the work. Also AI targeting the creative fields made most people instantly turn against it because everyone imagined post scarcity utopia where art was one of the main activities humans still do, instead it’s the only thing AI can do without major fuckups mostly because when it does major fuckups it does not cause any damages, and even then 99% of AI output is the m mf ost generic mediocre boring shit imaginable, making the sloppiest of human creation like Marvel movies look like masterpieces in comparison. Art is the one thing you must not automate as you did it anyway so you made people forever mad.
>a chatbot "played" a videogameNot technology.
I wonder how long it'll be until this can happen in real-time
>>109747059>There is a fundamental barrier to the transformer architecture.So? Doesn't means they can't make improvements or buiold a brand new architecture, we're still just on year 4 of AI, year 4 of planes still looked like this.
>>109747004oh hey a strawman argument>total retard death
gpt-1 played dota2 competitivelywe're not just flatliningwe're regressing
>>109747789no, retard, gpt-1 was not the dota AI
>>109747646>but in real life if AI can automate half of your job successfully then half of your coworkers are fired and you do double the work.That's literally a skill issue dude.GPT6 came out and it's doing most work FOR humans. Not the other way around. > Also AI targeting the creative fields made most people instantly turn against it because everyone imagined post scarcity utopia where art was one of the main activities humans still doThat's retarded.If a robot could cure cancer or vacuum, why couldn't it draw and paint as well? Do you think handing a paintbrush was suppose to make a machine explode? > Art is the one thing you must not automate as you did it anyway so you made people forever mad.Boo hoo.
>>109747801>it's doing most work FOR humans. i can tell that you have a fake email job
>>109747592Money means nothing, it's the message it'll send.
This is using TAS a billions of tokens, maybe it's impressive for absolute retards like OP. OpenAI beat Dota 2 pros live during TI, a game orders of magnitude more complex than portal 2 and that was before the launch of GPT 4 and without TAS. Also transformers are hitting their cap it's unsustainable
>>109747801>W-weell b-boo-hoo!1what I don't understand is why you haven't put a bullet in your head yet
>>109747023Actually it can, in 18-20 hours. Sol already could, but in 100 hours.
I fucking hate AI, god kill Sam Altman
I'm not saying AI won't transform society, but this is no different than GPT 4 getting 98% on law school tests for which it was trained on the answer key. GPT 6 is literally trained on playthrough videos including some that will include inputs on screen.
You know its a clanker when it used the cube instead of the chairs in this part
>>109747789>>109748064>>109748235Sweet Jesus this board is so low IQ
>>109747713With an LLM? A very long time because the architecture is really not suited for interactivity: more complex inputs take more steps to interpret, and more complex outputs take more steps to generate.
>>109747789>>109748064Jesus fucking Christ.OpenAI Five was deep-trained on Dota specifically and could only play Dota. Do you guys also think that StockFish is somehow smarter than GPT-6 because it's the best in the world at chess?
>>109747059>There is a fundamental barrier to the transformer architecture. I don't know why you don't understand this.why do you insist on getting raped repeatedly
>>109748259>>109748301TAS is literally easy mode with pauses, shut the fuck up
>>109748247I used the PCs from the offices
>>109748322Mate, the point here is that the LLM models are made to do many different tasks and GPT-6 likely wasn't trained to play Portal at all. Comparing them to a model deep-trained to only do one specific thing is so pointless that it makes you sound really retarded.
>>109747446pavlov's dog brainhears the verboten words and starts to bark, no understanding of what's going on
