Shifting computing around and giving another title to the CEO is now "AGI Development". At least that gay jew knows how to trick VCs into buying his claims.
>>108452098>Shifting [Headcanon]
>>108452036everything is cope and smoke and mirrors. Ever since the first openai SOTA model was publicaly available all the improvements have been marginally more accurate responses, larger context windows, or quicker responses. The only optimizations these corpos are making are for more compute so they can support larger models. There is no AGI. We will get to a point where the growth is no longer possible (much like transistor sizes) and it will all stagnate. We are clearly reaching this point with how the latest releases are performing and im sure we'll see companies IPOing either late this year or early next year to exit
>>108452036>Shovel salesman says things are moving faster than expected for selling shovels
>>108452036woooow so nothing to do with them bleeding money>surely if we call it AGI it will solve all problemsMeanwhile Claude can't fix the bug in my 1500 lines project after half a day of proompting. I had to resort to make it create autistic debugging tools for me instead. But then i noticed the debugging code was almost as long as the actual code and subtly buggy. I uninstalled Cursor out of contempt. Waste of time
>>108452036>world simulationSick!>focused on roboticsehh. Was kinda hoping they meant something akin to Google's Genie modelshttps://deepmind.google/models/genie/
>>108452036Between OpenAI shutting down Sora and Anthropic slashing majority of usage for max plan users, this is just the saddest cope possible.>b-but luddite!soon they won't even have rupees to throw you
hmmm
>>108452317you're conflating incremental improvements in one direction with a plateau in capability overall. better accuracy, larger context windows, and faster inference solve different problems in the capability space. the fact that scaling one particular metric has hit diminishing returns says nothing about whether the entire tech tree is exhausted. openai abandoning sora is a resource allocation decision based on where they think the biggest impact towards advancing the tech tree.you're prtty much saying that we'll never have computers because electricity isn't possible to generate and distribute at a scale. they're different problems and once you solve one you can solve the other even though they're not directly related on the tech tree.
>AGI in 2 weeksAltman-chan this is the 581,719,634,672nd time you've said this
>>108452036dilate
>>108454479And what is this tech branch we are not seeing
>>108452036mediagen like sora never made sense to meresult look shit, negative PR effect, unresolved copyright and the only real use being the internet memes
>>108452036>this just in, the richest 5 people in the world have decided to cut down ALL trees and have already gotten enough governments to approve their plan and have begun the process.
>>108452036>Things are moving faster than expected >We don't yet have the capabilities we said we would by now >But we are moving towards them faster than expected >At this rate we will have these capabilities in 6 months >6 months later >Things are moving faster than expected >We don't yet have the capabilities we said we would by now >But we are moving towards them faster than expected >At this rate we will have these capabilities in 6 months >6 months later >Things are moving faster than expected >We don't yet have the capabilities we said we would by now >But we are moving towards them faster than expected >At this rate we will have these capabilities in 6 months >6 months later >Things are moving faster than expected >We don't yet have the capabilities we said we would by now >But we are moving towards them faster than expected >At this rate we will have these capabilities in 6 months >6 months later >Things are moving faster than expected >We don't yet have the capabilities we said we would by now >But we are moving towards them faster than expected >At this rate we will have these capabilities in 6 months >6 months later >Things are moving faster than expected >We don't yet have the capabilities we said we would by now >But we are moving towards them faster than expected >At this rate we will have these capabilities in 6 months >6 months later >Things are moving faster than expected >We don't yet have the capabilities we said we would by now >But we are moving towards them faster than expected >At this rate we will have these capabilities in 6 months >6 months later >Things are moving faster than expected >We don't yet have the capabilities we said we would by now >But we are moving towards them faster than expected >At this rate we will have these capabilities in 6 months
>>108452036AIGODS keep winning. Ludditetroons seething rn
>>108455610I don't know, I generated that comment with claude opus.
>>108455762Holy Luddite copeola.
>>108452036>faster than many expectedtransformers are going nowhere
>>108452317Lol in meantime Moores law ran from 1960 to 2010 or so. So I guess we get improvement until 2080 or so?
>>108455762
>Selling goyslop generator to consoomers didnt work cause consoonmers are broke>Lets pivot to corpos instead! by copying elon grift>Corpos gave out all the money and are now pulling in debt and laying off ppl to get "AI">All while corpos like coreweave are building giga projects on soon to be outdated blackwell GPUsGEEG
>>108457139Cute.
>>108452036>accelerate the economyToward collapse?
>>108457134The improvements to the size of a microchip and the performance of a neural network arent comparable in the same ways
>>108452036Damage control
Who taught indians the word "Luddite?"
leaked yesterday btw
>>108460479Did that goof really generate his posts with AI? Either he's a gigaretard or it's all fake
>>108452036looks like they are going down
>>108452036>Jew>wild claimsWhat are you even doing OP?
>>108457111What an attention whore.
>>108459944NFT grifters.
Spud can eat two breakfasts
6 more monthstrust the plandiamond handshodl
>>108452036>brockmanjew jew jew jew jew jew jew jew jew and jews
>>108452036>hurr durr ludditeAnyone who uses this word in context of AI should hang themselves.
>>108460479even if this shit is true, which probably it isn't given that this looks like an ai genned text(not to mention that dozens of """""""leaks"""""""" get posted on /pol/ every day), it will at best be a lot of hype again and then they'll continue to not be profitable.Sora unironically could do quite well if they just scaled it down and repurposed it for specialized usecases like generating placeholder footage for media enterprises or some shit, instead of giving public access to literally fucking everyone in order to waste on useless slop. However the people who manage OpenAI seem to be somehow even worse at their jobs than the bigwigs at Microslop. The way they fumbled the bag they got with ChatGPT 3 is going to go down in history right next to formerly famous dotcom companies like AOL.the real truth imo is that the government told them that they'll need the computation for something else.
>>108463555You can already hook agents up to robots. It's cool but given how bad people are at build/using/deploying agents and how bad people are at building/using/deploying robots it's just not going to get used.It took these AI companies years to catch up with what I built at home on my own and that's with easy software. The robotics stuff just isn't happening.>but businesses-Dude I literally work at a company building stuff with robots. Do you have any idea how slow businesses move? We're way faster than most too. Then after all that there's the idea that the model is actually useful beyond mundane/gimmicky stuff. Meh. No. It could probably replace Amazon warehouse workers but they were already doing that with classical robots. This just speeds that up.
