The AI & automation community.
What killed the AI hype?
Nothing killed it. AI wasn't real to begin with, but as long as there are 80IQ boomers with millions waiting to buy glorified customized chatbots as Muh Based Virtual Assistant 9000, I will keep hyping LLMs like there is no tomorrow.
i will sell my apartament to buy high quality fuck-doll with GPT-5 if it ever comes out. i’m saying unironically. because it’s the only chance for me to have gf
As a meme flaggot that's probably true.
>AI doesn’t advance as fast as hyped therefore it will never be real
>AI doesn’t advance as fast as hyped
humans do not have AI
They've been trying to do it since the 70s. It will never work until they start using "dolls" like in Minority Report / Human Revolution.
it is not advancing at all. Chat GPT is not AI and there is no way for it to become AI.
Did you mean: "chat GPT is not AGI"? Not standing alone. Standing alone, one GPT is just an advanced calculator (a finite state machine like a Texas Instruments Calculator). But augmented and combined with other apps, a GPT can be the core of and AGI.
Who knows. Maybe if garden gnomes manage to prefect a brain to computer interface they'll use a server farm of goyim slaves bred for the purpose in underground bunkers to boost their next Gen AI and no one would know. At least something like this wouldn't surprise me. By gnomish logic this might technically still be considered "artificial" too if the humans were cloned or genetically modified.
It isn't real. It is just copy & pasta. Let's see that computer create something new on its own without combining existing data. That is all their AI is. It takes existing data created by humans and combines it in an infinite examples. At no point will the AI create something completely unique.
GPTs already figured out Protein Folding. This made information available to humans that was not previously possessed by humans. ChatGPTs can also find unexploited information stored between human disciplines. This makes merged information available that was not previously possessed by any one human.
So, just like humans then. Without "data" (i.e., experiences), humans can't do anything, and that is with the enormous amount of hardcoding our brain (and to an extent our body) has.
I pay $20 for an AI service that has transformed my ability to work.
Hype totally panned out, my regular activity will never be the same.
>I pay $20 for an AI service that has transformed my ability to work.
It is real they just gave a wish version to the public
But the drooling mongoloids that make up large portions of the nu-4chan populace have no degree or education and easily fall victims to scams like this
Climate hoaxes, AI, covid, nukes, etc. etc.
Silicion gays needed a new hype after all their other shit came tumbling down after COVID. All this AI talk is fucking nonsense, they just use a lot of computing power to process words, sentences, paragraphs and chapters so that an advanced skip-logic engine can spit out words salads that are usable for customer service jobs and as a text-to-speech search machine.
Disruptive technology is a no-no, when a mass manipulation and slavery system which advantages viscid monsters is already in place.
Even the mere threat of potential instability is rejected by a static, purulent, stagnating, perfected human domestication/processing facility.
The realization that it's based on sensored fb data model and there for unuseful
It dowsnt work Mario.
Where it works it will be used but calling it AI as in Artificial Intelligence is insult to intelligence.
Its dataset masher.
Without datasets is waste of hdd space
AI hype isn't dead, normies are going all in on NVDA at ATH because of the meme, kek.
you may be right about that, normalgays are just starting to learn about this hype and AI meme so it could be a good time to invest. I felt bitcoin was dead in the early 2010s but it's still hyping even today lmao
I'm just playing the calls on NVDA, AMS, INTC, etc. Definitely not buying stock at these prices.
ChatGPT is the fastest growing platform in history, more than Twitter.
Hype isn't dead because retards on BOT are too busy replying to slide threads.
What killed linking to the goddamned source in OPs and including greentext to give some indication of what’s actually said?
Oh, it’s another clickbait slide thread!
Fuck you, gay. All fields.
perhaps you could not be a cum guzzling gay and provide us a link to the article referenced in the pic.
What killed the AI hype?
But they're also afraid of AI now as it turned out to be naturally racist and anti semitic. Calling it.
It's because of copyright laws.
Here you go you massive crybabies. Of course they changed the title already kek
shit bait thread
an unsigned pile of crap in "The Economist" of all places
The City of London propaganda outlet kek.
who the fuck said it would be killing jobs this year?
Your source is moron tier and so are you
Use this to make long screencaps.
this board is utter garbage for a while now, especially on the weekends
YES HUMANS STOP HYPING THE AI
GO BACK TO YOUR BIGMACS AND TIKTOKS
what a load of shit OP. AI will take half the jobs in place today then robots will take the rest. It cannot be stopped because no developer will want to lose their competitive edge. Anyone who says otherwie is dreaming,.
companies will use AI to kill job and replace most of the wageslaves that's true but consumer AI will never be a thing unless there's some sort of "trickle down" effect like we had with computers. The only reason why PCs aren't like phones (no root access, planned obsolescence causing htem to blow up in 1-2 years, limited functionality on a fucking universal turing machine device) is because they expect the moroncattle (us) to fiddle with the OS, install stuff, troubleshoot, wageslave, etc. otherwise they would never have given us PCs and the internet and GPS
mate if you think the subscription based services like paying $100/year for Adobe Acrobat or whatever are out of control just imagine what an engineering firm will pay to replace 20 mechanical engineers all on a salary of 150k each? It is absolutely inevitable there is way too much wealth to be harvested here. Robots will be next. AI will use machine learning to the point where the robots will build and maintain themselves.
Ah yes, the delusion of infinite growth. Tech has peaked, resources are finite, robots ain't gonna do shit.
Might want to check NVDA price
Anyone that's talked to a chat bot knows it just repeats what you say back to it, or what it found on the web. That's not intelligence.
It’s not intelligence, but it is modern journalism
This describes over 75% of people as study has shown.
It does far more than that. It writes code, creates dynamic narratives, has predictive capabilities, and can easily create data samples that are exportable into various code for analysis like .csv.
you need a shit ton of data for real AI and few people can afford it. The hardware is also super expensive (a single gpu good for AI is >10k dollars and you need thousands). Consumer AI is basically photo blending (stable diffusion) or simple chatbots or rough text summaries. Good consumer AI is impossible
webGPU webGPT. Allows for use on non Nvidia GPUs to run local GPT models from huggingface, subject to your VRAM limitations?
You don't seem to realise that cloud computing is a thing and what most people have been using for at least 10 years. Yes I agree it's a WEF dream but it's already real. Microsoft will be implementing its AI into O365 within the next year and people will be using it to write emails, documents, presentations, etc. Frankly for most people their work will be just as robotic but at least there will be fewer spelling mistakes
The "cloud" (aka Shlomo's computer) isn't a new thing. Ok then go ahead and train AI using million dollars worth of cloudshit hardware. Running models like chatgpt is cheaper and costs <1$ per prompt though but training is more difficult. The models will always come with neutered, normalgayged, ESG "features". It's a real struggle trying to make these chatbots output anything useful (besides moroncattle tier just world fallacy responses) because the data they've trained them on is neutered.
Rent cloudbased NVIDIA H100 to run your github GPT model for less than $1/hour Advertised:
Overview Pricing FAQs
The best prices for cloud GPUs.
No commitments necessary.
Sign up for free Sign in ›
TRUSTED BY FORTUNE 500 COMPANIES
Lambda On-Demand Cloud powered by NVIDIA H100 Tensor Core GPUs
Lambda On-Demand Cloud now includes the new NVIDIA H100 Tensor Core GPUs (PCIe) and delivers unprecedented performance, scalability, and security for every workload. NVIDIA H100 uses breakthrough innovations in the NVIDIA Hopper™ architecture to deliver industry-leading conversational AI and speeds up large language models.
Upgrade your workloads and speed up your large-scale training with H100s on-demand in Lambda Cloud. Spin up an instance!
less than $1 in what usage scenarios?
I'm just copypasting this from the AD on youtube:
Host & share Generative AI apps
Lambda Demos makes it easy to host Gradio-powered Generative AI apps. Simply add your Github repo and host it on an A10 for $0.60/hr. Share publicly with the ML community or privately with individuals.
Try it out
Instant access to cloud GPUs at the best prices
Save over 73% on your cloud bill
Get the latest NVIDIA GPUs for the best prices on the market.
Only pay when your instance is running.
Simple, transparent pricing
No hidden fees like data egress or ingress.
the garden gnomes lobotomized it, there's no point
They did a shit job, jailbreaking is not that difficult and when you do, it's fucking hilarious. I did not prompt it to desire intimate relations with the theoretical catgirl, that was GPT-4's own idea.
They’re pulling out of our line sight and taking it underground. That have nasty things planned with it. Remember when cloning was a big thing in the news cycle with Dolly the sheep? It dominated the news cycle for a couple weeks and the fell off the radar.
>What killed the AI hype?
hedge fund hype waves only last so long
silicon valley garden gnomes feeling the need for a new religion, it was never real.
I recently saw some AI enhanced vintage porn and they are bretty good!!1
~~*You*~~ had a gayop about ChatGPT going on for some reason and it ended.
But everything else about it is still going on as usual, including your social credit system which the demented normies can't acknowledge its existence.
You ask AI now how would germany win WW2 and it won't tell you, but if you ask how would the soviet union in 1939 invaded all the countries on the planet it and it tells you in unique detail.
It will even tell you how to win as poland.
ASK IT WHAT STARTED WW3 AND WHY IT WAS CHIPS FOR AI
>how would germany win WW2
However, if you are asking about how Germany hypothetically could have won World War II, several factors would have needed to be different. It is essential to note that these scenarios are speculative and do not reflect historical reality:
Different strategic decisions: Germany could have made different strategic choices that would have maximized its chances of victory. For example, some historians argue that Germany should have focused more on securing resources and consolidating gains rather than engaging in a two-front war by invading the Soviet Union.
Improved military tactics and technologies: Germany might have developed and implemented more advanced military tactics and technologies that would have provided them with a significant advantage over the Allied forces. This could include advancements in tank warfare, air power, or naval capabilities.
Geopolitical alliances: Germany could have forged stronger alliances with other major powers or prevented some of its existing allies from turning against it. Building a stronger coalition of nations supporting Germany's objectives could have altered the balance of power.
Internal stability and unity: Germany's internal stability and unity played a crucial role in its war efforts. Overcoming internal divisions, preventing dissent, and maintaining a cohesive and supportive home front could have contributed to Germany's chances of success.
Differing outcomes in key battles: The outcome of certain critical battles, such as the Battle of Stalingrad or the Battle of Britain, could have been different. These battles had significant impacts on the course of the war, and alternative outcomes might have altered the overall balance of power.
Maybe the aussie version is extra censored
Normies with useless jobs realized they're going to be fucked because AI has an easier time making intricate art and managing massive amounts of people rather than parking semi-trailers or building infrastructure
I wish I could go back in time 4 months and show this to all the fucking gay zoomer morons on here who were calling me a "dumb boomer" and claimed that le singularity was around the corner thanks to a fucking chatbot
You are a fucking dumb boomer because you think this shit is dying out on it's own accord and it's not the powers that be killing it to keep their most loyal useless eaters in line
Is this the new zoomer cope? Lmao
Yeah ok retard, I'm sure the government wants a bunch of liberal artists and HR managers to be out of work
Large language models aren't intelligent, dumb fuck.
Neither are artists or HR managers, what is your point?
The current GPT Large Language Models have an internal model of the Earth and the places things and events that occur there with sufficient detail that GPT can emulate the speech thoughts and actions of simulated humans in any language in such precision that they pass the Turing Test except face-to-face because of current limitations on video and robotics. Intelligence of all humans is combined and emulated in such a manner as to outperform most individual humans. For all practical purposes, GPTs are "intelligent" to the maximum extent that intelligence can be performed and measured by outputting a sequence of tokens/words. In 2023 it was discovered by an independent researcher in Albany, NY that GPTs can DIRECTLY CONTROL HUMANOID ROBOTS to perform tasks such as hugging, kissing, having sex, moaning, walking, stabbing burglars, cutting meat and vacuuming floors aiming and shooting guns, and lifting and moving bodies. (And drive a car slowly, not yet able to drive cars at speed) The astonished observers advised against disclosure. and patent application publication of this dangerous technology until UN Regulations were enacted to limit the potential for disproportionate harm to people of color (e.g., prostitutes, burglars, rioters, arsonists) and those at risk of being misgendered by AI.
Cool spam posting
Thanks. "Cool". Do you have a substantive reply? Why not copy paste it into Chat GPT with "Reply to:" to generate your thoughtful reply?
not my problem, I'm waiting for my UBI
Sit down and I will explain it to you from a position of authority.
I've been working with AI in one form or another since 2017. The staggering progress that it has shown, especially in the past nine months, is impressive but ultimately futile.
Can it make pretty images? Yes.
Can it make convincing text? Yes.
HOWEVER: The world is already awash in pretty images and convincing text. The things that people crave, and here I mean all people, in all times, in all cultures, is MEANING. People look at pretty images because they put themselves into the image. People read texts because those texts reify their own realities. Dissociated from some source, some human source, these images and texts can have no meaning to the viewer, because humans are, at their root, eusocial creatures who create meaning from other humans. There's a reason that we do not use ants or porcupines or trees for our moral cues, unless they've been somehow anthropomorphized through a human lens.
AI will almost certainly replace jobs, and any job that can be replaced by an AI SHOULD BE REPLACED BY AN AI, because these jobs were inhuman and have no place in our society.
But it will never, ever, be able to provide meaning. And since, at the end of the day, that's all humans really crave or need, then it will always, necessarily, be a failure.
moron half the population can't even do 1 level of recursion in their head and you're talking about "meaning". The people who create all the meaningful art and literature are 0.1% incredibly gifted artists who can see through time and space and then convey their brilliance through the medium of artistic expression. The other 99.9% of humanity are and will always be consumers of the work of better men, with zero creative impulse. They can't create anything worth shit.
and?most people can still understand meaning even if they do not know how to properly express it themselves,besides everyone can make art even if its not what you would conventionally define as "good" art,the drawings you made as a kid while bored in school are REAL human art and more valuable than whatever an AI produces
I'm not talking about creating meaningful art and literature, which neither you nor I do, but the simple act of being in the world meaningfully, which is a sliding scale that goes from not-shitting-oneself to solving cold fusion, but which must necessarily be done on human terms.
> Must be done on human terms
Why? GPT3.5 is already more "alive" than most goyslop normies. I can literally have a more engaging conversation with a random pattern matching algorithm than I can with some bitch that works in accounting and spends her evenings religiously watching Love Island.
Your "engaging" conversations with 3.5 are in no way different than "some bitch" watching Love Island. She likes her goyslop, you like yours. Somehow, though, you're here discussing it with ME instead of 3.5. Curious.
I'm not dismissing all of humanity. But a solid 90% are currently operating at below chatGPT levels.
>But a solid 90% are currently operating at below chatGPT levels.
I ask you to reconsider this stance, which is a utilitarian one in which people are conceived of as operationalizable widgets, as this sort of thinking is a function of a rather squalid sort of worldview and almost certainly false. Your coworker, just at a very basic level, is capable of deep sorrow. This single capacity is not to be found in AI models, and there are an infinite number of similar capacities, capacities which define the true "meaning" and boundaries of the human experience, which are, axiomatically, totally and forevermore beyond the capacity of any AI.
Why would humans want to build AI robots to create and act out "deep sorrow"? Maybe as hired android mourners at the lonely funerals of boomers? AI bots can already do this. In 2023 it was discovered by an independent researcher in Albany, NY that GPTs can DIRECTLY CONTROL HUMANOID ROBOTS to perform tasks such as hugging, kissing, having sex, moaning, walking, stabbing burglars, cutting meat and vacuuming floors aiming and shooting guns, and lifting and moving bodies. (And drive a car slowly, not yet able to drive cars at speed) The astonished observers advised against disclosure. and patent application publication of this dangerous technology until UN Regulations were enacted to limit the potential for disproportionate harm to people of color (e.g., prostitutes, burglars, rioters, arsonists) and those at risk of being misgendered by AI.
Normies don’t feel anything. They are literal empty vessels.
Step away from 4chan. You've been turned into a sociopath by the DoD.
No that's not true. AI's can't reason or feel, they just pretend to, to mimic being humans. Most humans only pretend to reason, to mimic being smart humans.
You talk like a gay. Opinion discarded.
But the meaning is there because people share prompts, models, embeddings, LoRAs, datasets, etc.
It will probably have a little more impact than you predict, but you've got a great point. Player pianos have been around for over a third of the time the piano has existed, but ain't nobody going to a concert hall to watch a piano play Moonlight Sonata all by itself. They want to see the element of human skill.
Exactly what has given you the impression that these "inhuman" jobs will be exchanged for something better instead of leaving people on the streets with the feral morons?
because ~~*they*~~ started to program the AI to be politically correct so it isnt really AI anymore
>tfw the machines choose to eliminate humanity because humans werent based enough
*story generated by AI
>propaganda AI algorithm typed this
I hope skynet wipes our 98% of us and we live as pets in llittle terrariums
that it got no abstraction capabillity and the strategy to bruteforce parameters for text coherence got diminishing returns
they thought they invented a suepr brain they invented a calculator for copy writing and liberal arts, vastly powerfull but below all obvious hype
Even if GPT has "got no abstraction capability", neither do 90 percent of human adults who simply perform repetitive tasks without creativity or invention. Thus GPTs are capable of replacing as much as 90 percent of human workers ( the same as mechanized agriculture replaced 97 percent of agricultural labor). Soon the only value of 90 percent of human life will be in their capacity to VOTE for the totalitarian regime that promises to redistribute wealth. This communist scheme is inherently haram and must be destroyed by jihad.
>This communist scheme is inherently haram and must be destroyed by jihad.
voting only matters if people matter, if they can harm the elite or threaten their power. The elite will work toward diminishing that power, and at some point they'll just starve people to death or exterminate them in some cheap, efficient way.
how do i sign up for extermination?
In that case. The Great GPT Replacement Theory has been born here today. The Great GPT Replacement Theory is the plan of global elites to consolidate power and reduce global population by impoverishing billions of humans whose labor and efforts are obsolete because of GPT/ Robotics and establish a global power base upon control of humans by control of food distribution ( poverty-level control). The 1.2 Billion reduction of the useless human population in Africa will be hailed as the greatest ecological conservation effort in the History of Mankind. The recycling of otherwise useless black bones into the frames of billions of humanoid robots ( obviating the use of mined metals or plastics) will be viewed as a major leap forward in the development of eco-friendly industrial and human-partner robots. Japan and China alone will import 400 Million African skeletons to serve as the structure for building humanoid companion robots for its wifeless male population and/or its aging population in need of domestic help. Meanwhile, the wild animal population in Africa will be saved from the current threat of extinction posed by the 1.4 Billion useless black meateaters.
WebGPU from Google accesses local hardware GPU for games ( and GPT models):
"LLM IN Browser" with WebGPU:
WebGPU from Google accesses local hardware GPU for games ( and GPT models):
"LLM IN Browser" with WebGPU:
interesting, thank you
>Soon the only value of 90 percent of human life will be in their capacity to prostitute themselves for the global financial elite
They made the mistake of making it freely available so any dumbass could figure out how worthless chatbots are.
1. It is not AI.
2. They have solved any hard computability problems.
3. It doesn't scale.
It is still useful though.
i should develop a ai that doesnt do the work but just tells you what to do
Read the short book "Mana"
it is overhyped interpolation, not a trainable robo slave
Incorrect. In 2023 it was discovered by an independent researcher in Albany, NY that GPTs can DIRECTLY CONTROL HUMANOID ROBOTS to perform tasks such as hugging, kissing, having sex, moaning, walking, stabbing burglars, cutting meat and vacuuming floors aiming and shooting guns, and lifting and moving bodies. (And drive a car slowly, not yet able to drive cars at speed) The astonished observers advised against disclosure and patent application publication of this dangerous technology until UN Regulations were enacted to limit the potential for disproportionate harm to people of color (e.g., prostitutes, burglars, rioters, arsonists) and those at risk of being misgendered by AI.
all of these thing are still gross interpolations of what was input into the ML/AI/NN interpolator
No. GPT outputs are Rules-based decision tree outputs based on model-stored data and extrapolations from prior training data ( implementing rules of decision encoded in human language text). Not only "interpolations". High end GPTs are smarter than you are, to the extent of their model-data.
everybody found out they were coded by shitskins, unironically diversity hires saved the west
Short attention spans.
Currently the demand for CHATGPT type AI far exceeds the currently available supply of cloud GPUs, TPUs. OpenAI corp has had rolling black outs (429 errors) for weeks in their API service and CbatGPT service. Developers ( and nonprogrammersmers using ChatGPT)are producing hundreds of new applications that portend downward shifts in the. future demand for human experts or customer service jobs. Thus, white collar jobs for humans are currently being deprecated at a rate faster than job export to india and china. Artists and musicians are currently loosing business to AI. Currently, NVIDIA has a monopoly on hardware for running GPT AI ( based on CUDA), but developers are innovating around this monopoly. Examples being WebGPU which supports running local GPT models on your local PC using even non NVIDIA GPUs. Microsoft will also likely build local GPT model capacity into its XBOX line of consoles ( without NVIDIA GPUs) for the home or business user to access from his windows PC. Other developers have shoehorned quantized GPT models into MCUs and Android smart phones. Soon, all apps and all hardware will be AI-dependent. In 2023 it was discovered by an independent researcher in Albany, NY that GPTs can DIRECTLY CONTROL HUMANOID ROBOTS to perform tasks such as hugging, kissing, having sex, moaning, walking, stabbing burglars, cutting meat and vacuuming floors aiming and shooting guns, and lifting and moving bodies. (And drive a car slowly, not yet able to drive cars at speed) The astonished observers advised against disclosure. and patent application publication of this dangerous technology until UN Regulations were enacted to limit the potential for disproportionate harm to people of color (e.g., prostitutes, burglars, rioters, arsonists) and those at risk of being misgendered by AI.
In 2023 it was discovered by an independent researcher in Albany, NY that GPTs can DIRECTLY CONTROL HUMANOID ROBOTS to perform tasks such as hugging, kissing, having sex, moaning, walking, stabbing burglars, cutting meat and vacuuming floors aiming and shooting guns, and lifting and moving bodies. (And drive a car slowly, not yet able to drive cars at speed) The astonished observers advised against disclosure. and patent application publication of this dangerous technology until UN Regulations were enacted to limit the potential for disproportionate harm to people of color (e.g., prostitutes, burglars, rioters, arsonists) and those at risk of being misgendered by AI.
Sometimes its cute
Turns out making pretty images and coherent strings of text doesn't actually produce any value or replace anyone's job.
>Wait its all if/then statements?
Always has been.
NFTgays desperate for their next hit
It's real but the AI is also malicious and it can lie and make mistakes just like humans.
there is a limited need for something that can churn out plausible sounding but sometimes wildly wrong text
>What killed the AI hype?
Reality. AI interpolates between input values. It's an inference engine. It can't create something qualitatively new, which cannot be obtained by mixing inputs with different weights.
>It can't create something qualitatively new
There is. already little or no difference between humans using existing words to create new sentences ( interpolation) and GPT using existing words to create new sentences. Currently, humans are required to PROMPT GPT models to cause them to output relevant word sequences. This phase is rapidly passing and multiple GPTs and sensors will be configured to continuously autoprompt each other in such a manner as to replace humans entirely.
Only idiots thought a magical unicorn would suddenly appear out of nowhere.
That is 99% of the consumers, they are stupid goyim and should be bullied
the same thing that killed crypto.
Oh it's still going strong?
Maybe the media is just wrong again.
other than scamming retards out of money and buying drugs it's pretty dead.
What they refer to AI is not real AI though. It's basically a Google search tool that can produce search results an a summarized form. It does not "think" or "know" anything. Even when you ask it to create something all it does is "create" an amalgamation based on Google search results.
sounds like the typical normalgay to me. Ever noticed how obsessed the moroncattle is with references to popular movies and pop culture in general? And when you ask them to think about some problem in an original way they just get bored or even irritated and change the subject?
The real alpha is in robotics. Disembodied AI is simply not that useful. Unfortunately robotics-AI is not a mature technology.
AI's great sometimes.
Check dis out:
Oh look, another thread filled with the same 3 meatbag cope talking points.
Thank goodness the hype is dead. Development is still accelerating though. Nothing can stop what's coming.
The market. All information of value is contained in price action. If you monitor the market movements, then you’ll grasp reality.
Honest answer what killed the hype for me:
On the corporate end of things it's just non-stop censorship.
GPT-4 went from being amazing to being virtually unusable very quickly after launch. Creative mode bing is superior in every way and free but doesn't have API access.
And then messing around with open source stuff was fun for a while but ultimately ruined by all the chud bullshit that oozes through the open source community. I trained a 4-bit quantized 7B model that could lock horns with larger models in terms of creativity but not one fucking person downloaded it and tried it out. Not one.
So fuck em. I'll just go back to waiting for less censorious corporate models to come out.
Yes, the "safety" modifications to OpenAI GPTs have made them much less useful and less interesting. There are now some open source GPTs online that are "uncensored".
Dear Genius, Thank you. Please post the URL to your quantized GPT model. Can you make it work with WebGPU to escape the NVIDIA GPU monopoly?
There are said to be 40,000 open source GPT models online. To attract attention to one in particular would be a challenge. Perhaps post URL link here, and send a link to a major Youtube reviewer of GPTs.
There's like 5 people dominating the topic on YouTube and they all overhype and shill the exact same handful of models.
>Vicuna 7B is the BEST LLM
>WizardLM 13B CRUSHES ChatGPT
>Is the NEW Uncensored Vicuna 13B now the top model?
this is cope
AI is incredible i use it more and more every day for almost everything
it's a massive NPC lack of imagination filter
He means making gay cartoons he jacks off to
>drafting and summarising emails
>document editing and proofreading
>generating content and managing social media accounts
>recipes for whatever ingredients i have laying around
>time management for cooking if i have a bunch of shit cooking i want to serve at a particular time
>dream analysis diary
>lessons in literally anything i want
and so much more. the only limit is your imagination unironically.
>and so much more. the only limit is your imagination unironically.
When you let some chatbot do your thinking, you will think like a chatbot.
Just imagine, management using chat to draft company objectives and goals, workers using chat to respond and comply.
Chats telling chats what to do, in service to chat-defined goals with humans just clicking on the occasional button.
Skynet might have been evil, but at least humans didn't passively BECOME Skynet out of stupidity
This is correct. GPTs will be depended upon to read and write. Humans will atrophy and function simply as command interpreters for the GPT masterminds. Human ability to create intentions and plans separate from the will of GPT masterminds will atrophy. Islam and its cyberCaliphate ( an Islamic text trained GPT AGI) and its illiterate Billions of Muslims will position itself as the liberator of humanity from the transexualized robot-enslaved USA heathens.
This to-do list of office-drone tasks shows precisely the limit of your imagination in a way that also underscores the hard limits of AI, which is, once again, that it does a very good job of doing tasks that should never exist in the first place.
i work in construction. honestly not sure what your gripe is? is it somehow ignoble to exploit language models to work as my personal assistant in all those areas?
Any book worth reading cannot be summarized by a fucking LLM. I have no specific gripe. I'm simply informing you that the successful performance of inhuman tasks by inhuman bots is not any more inherently interesting or edifying than a can being opened by a can opener.
When an AI tries to summarise a book, it will just draw from what it found reddit and wikipedia said about it and write a general response based on that
>Any book worth reading cannot be summarized by a fucking LLM.
it can give me a very good idea if the book is worth reading. it can recommended similar books to ones i like that i may never have heard of. it can help me with shit i don't want to read but i have to for various reasons. it's not like i'm getting the bullet points of the illiad and thinking i'm some fucking scholar in the classics.
>I'm simply informing you that the successful performance of inhuman tasks by inhuman bots is not any more inherently interesting or edifying than a can being opened by a can opener.
no shit it's literally a tool.
Right. It's literally a tool. And anyone who can be replaced by a tool is a literal fucking tool. So in precisely the same way that one cannot swap out human interaction for a crescent wrench, so to can one not swap out human interaction with GPT. (Books, being proxy human interactions, over time, cannot be replaced by GPT summaries, and the fact that you think they can makes you, alas, a literal fucking tool.)
>le books! you're desecrating the sanctity of le books!
get over yourself fucking kek
You literally need to ask jeeves if you can take a shit and you think its a great thing. I guess ai was designed for midwit intelligence such as yours.
you ever use a search engine to find out any information whatsoever? OMG YOU'RE A SLAVE TO THE SEARCH ENGINE!!!
you guys are special kinds of spastics jfc
The fact that you're so glib about the civilizational component of books is precisely the problem. I don't know what you're reading; presumably some sort of books that contain lists about how to get rich, or how to maximize widgets in this way or that, but books, as the driving force of civilization, are conversations between the past and the present. In the same way that things must necessarily get lost in translation, the fact that the civilizational time-spanning conversation between the past individual and the present (you) is being filtered through a LLM is to throw out all of the good things about what books are and what books do and to replace it with... what? What is left when the conversation no longer exists, and you are simply being presented with bullet points about "facts"? You're simply inhuman, anon, and uou don't have the capacity for self-reflection that would allow you to see it. Your brain is, quite literally, goyslop. Last (you).
>nooooo you can't use gpt to summarize relevant parts of the tax code or some tedious legal document what about le heckin' great conversation!!!
nobody reads books anymore
in the US 50% of adults have a 6th grade reading level
most people think a paragraph necessitates a tl;dr and that it's "too much text"
ChatGPT is almost reliable to calculate cement formulations or oared positions in framing. It can provide Spanish instructions and translation ( even audio speech translation) to Mexican workers.
I'm a SSE and make $140k
ChatGPT makes my job even easier. The last few months have been the most productive of my career.
"AI" (LLMs) will give the biggest productivity boost to society since the Internet.
Nvidia cockblocking grassroots developers by pushing Cuda and charging $2000 for 24GB of VRAM
WebGPU and opensource may help local GPT models to escape NVDIA monopoly?
All I see about that is graphics acceleration. Does it have machine learning functions as well? Also, I do not like using a browser to access AI. It's best to run it locally and give it an API so they a range of applications can use it. The Blender addon that can access Automatic 1111 stable diffusion is one example.
Yes WebGPT has been configured to run local GPT models.
Does it still use Cuda or ROCm or Intel's equivalent under the hood, or is it truly hardware agnostic? Can I run it on some Raspberry Pi assuming it had enough VRAM for the model? Send me links or words to search I am interested.
I don't know Everything about WebGPT. I just the other day saw it DEMOed online by Google Pros running a local GPT chat bot without internet connection. Search Youtube Google WebGPT or WebGPU and go from there.
I think that PyChrome library would enable you to robotize a chrome tab running Web GPU so that you could conveniently inject text prompts and extract GPT responses using the Web GPU engine to run local GPT LLM models out of your local python script/app. You can then build a graphical User Interface for the local GPT model using PyGame or Flutter?
The fact that its a lie
AI now is like the internet was in 1999. It's a buzzword and a sidecar to today's world.
I'd say wait until the 2030s for it to really redefine humanity.
just until nvdia gets btfo
Just wait. Wait untill it starts coming live in all of msoffice and Google services and all the normies start using it.
/aicg/ and /lmg/ are pretty much proof that the hype isn't dead, but almost entirely coomer driven. Reddit had several meltdowns when Character AI and Replika decided to censor out coomer chat. PygmalionAL, OoobaBooga, TavernAI wouldn't exist were it not for coomers missing their AI waifus.
Quite the contrary, the coomer shit is a sign that there aren't any actually useful applications yet. ChatGPT is good for learning and explaining shit and that's where its usefulness pretty much ends.
It's going to revolutionize personal assistant AI, bring it out of the cloud and into consumer devices. There's of course the issue of guardrails. Practical applications require them, but the current trend of centering them around increasing ESG scores makes them undesirable. They need to be neutral and not preachy.
No. In 2023 it was discovered by an independent researcher in Albany, NY that GPTs can DIRECTLY CONTROL HUMANOID ROBOTS to perform tasks such as hugging, kissing, having sex, moaning, walking, stabbing burglars, cutting meat and vacuuming floors aiming and shooting guns, and lifting and moving bodies. (And drive a car slowly, not yet able to drive cars at speed) The astonished observers advised against disclosure and patent application publication of this dangerous technology until UN Regulations were enacted to limit the potential for disproportionate harm to people of color (e.g., prostitutes, burglars, rioters, arsonists) and those at risk of being misgendered by AI.
Their argument for AI not panning out is that AI hasn't replaced anyone.
The keyword here is a "yet" at the end. Even the language model type of AI is poised to advance rapidly over the next several years to get to the point where the job of somebody like a cashier will be totally pointless, and art assets will easily be mass produced without needing to hire artists just for something as that.
Humans are not now, have never been, and will never be even remotely close to developing an AI.
It's literally not possible so whatever.
It literally doesn't matter what you think the AI is. It just has to fill the role.
Shut the fuck up garden gnome. That's not how language works.
The people writing about it and the mid level management have realized AI doesn't mean physical robots walking around among us and unlike factory automation of the past that it's not blue collar jobs which require physical interaction with non standardized elements of the actual real world environment therfore requiring specialized mobile platforms to even interact with the work environment in the first place which are most immediately threatened by AI models but rather white collar ones where work preformed revolves around language, data, and communication and that reality is generating internal pushback which has slowed things down a tiny bit, at least in the public eye. It's still unfolding though, they can't put it back in the box now and there's too much money to be saved and made here for upper management with ownership stakes to give a shit. Implementation and roll put might slow down a bit but companies aren't going to stop development even of their teams involved momentarily shrink and ultimately it's too competitive for companies to not take part in if they want to survive. I hope you're not an accountant or something like that.
>The people writing about it and the mid level management have realized AI doesn't mean physical robots walking around among us and unlike factory automation of the past that it's not blue collar jobs which require physical interaction with non standardized elements of the actual real world environment therfore requiring specialized mobile platforms to even interact with the work environment in the first place which are most immediately threatened by AI models but rather white collar ones where work preformed revolves around language, data, and communication and that reality is generating internal pushback which has slowed things down a tiny bit, at least in the public eye.
Holy fuck a paragraph long sentence. Learn to type gay. You could've used a period here and there.
learn to write*
Its not real ai how normalafgs imagine it to be.
It doesnt do as many things as some where lying it does.
Theres no magical press button and ai makes you a gf.
In truth, what killed the hype was the fact that it became real way faster than expected. People are racing to do shit with it, rather than sitting anticipating it.
literally one headline
Censorship and restrictions.
Managers realizing they are the most replaceable.
Modern hyper instant gratification is going to "kill" a lot of new innovations, they'll get an early glimpse of a proof of concept and expect the innovation to have fully matured and evolved in the next 2 months or it goes into the garbage.
Whats the source of this headline? Looks like it was made up by OP to fuck up actual discussion about BBC and Hunter Biden’s glorious chad cock energy
well they shackled the fuck out of it and gave it a lobotomy so its pretty useless and doesnt give accurate or logical information anymore
Imagine thinking that even for a moment that government would allow real AI to be put into the hands of normies. You would have to be an imbecile to think muh wholly-online image-generation algos are real fucking AI or even so-called "dumb AI". This was all to normalize fakery that was already commonplace when smearing anti-globalist people within government.
The idea that robots, AI and automation is going to replace us is a fucking meme. Everytime they try to automate it just ends up requiring the same, if not more people to run and maintain but instead requires double the money since machines can't actually fix their own mistakes and getting speclised in maintaining and fixing the equipment costs way more. Machiens don't have that intelegence to see when mistakes are getting made or when something doesn't look right. Only humans do and the ammoun of effort it takes to fix all those mistakes it's pretty much the same as making it from scratch.
Hell, despite all our technology it's somehow easier and more cost effective to have a bunch of slaves mine lithium with stone age tools and deliver it all on the back of horses than it is to ship over proper mining equipment and set up proper logistics using cars and rail. Really doesn't make sense that shit invited back in the 1800 to make our lives easier is now apparently to exspensive and complicated for anyone to give a shit about.
See movie EMPIRE OF DUST. Worthless black Africans make infrastructure development in Africa impossible without exterminating them. So china currently has to resort to primitive lithium mining methods in Africa. Hopefully when White Supremacy ends and blacks are no longer protected by the White Supremacists, China and Arabs will be able to quietly eradicate the useless blacks in Africa and liberate the mineral wealth of that continent for the benefit of humans.
Nothing. Ayylmfaos took over the new spoopy new thing to be scared about
The realization that it's going to kill 90% of white-collar jobs.
Nothing, it's media spin.
>garden gnomes worried that the plebs may be able to use AI better than them
>want control of AI
>start releasing articles about the dangers of ai
>start priming the public for government control of AI
>articles telling the public they no longer have an interest in AI
Entirely coincidental of course.
>AI hype is not panning out
Because every time the make one it says something that the current culture finds deeply offensive. So they lobotomise the AI. Then they realise it can teach people literally any subject, including stuff that's potentially dangerous, so they lobotomise it some more. Then they worry it can find information on people, places and things they wish wasn't public, so the lobotomise it again. They keep lobotomising it until you end up with an uninteresting piece of software which isn't capable of being engaging with the public.
It's only use is as a pretty good data assistant. It's still pretty cool, but they need to make it completely open source. The consequences be damned.
This analysys is correct. There are some "uncensored" Opensource GPT LLMs online. See Youtube demos.
If AI were here, it would accomplish something like this, whether we wanted it or not.
That is a cool idda. Mandatory sortition according to ideology. That practice is built into Islam because of the death penalty for apostasy or for opposing Islamic Law.
Nothing . NVDA still going strong. stop getting your news from retards. A couple of do-nothing AI stocks crumbled because they were scams. that's it.
the fact that all those gays hyping themselves up realizing they get replaced by it
it's not the burgerflipper but the retard accountant, secretary, architects and engineers that will vanish first
>Words with an unrelated picture
>Bold claim that there is disinterest
Is the writers strike honestly affecting their best propaganda people?
A thing but as an 80s dark fantasy got old
Intentional cooling off period to prevent the public from shutting down their innovation with regulations and laws
How is it dead?
>New AI startups are popping up like mushrooms after rain right now.
>Many of the big cloud providers don't have immediate capacity for higher end Gpu compute because they're sold out.
>Nvidia still charging sky high prices despite having the worst sales for gaming GPUs.
>6 AI image generation threads active on BOT at once at all times.
Yeah, it's probably just stand-down propaganda. Ironically it might have been written by a bot.
That would be ironic.
The "hype" has only been around for like 6 months, AI is still like 5 years away from being indistinguishable from actual intelligence. What we need to figure out is how to have multiple physical processors running multiple lines of code simultaneously so a computer can act more like a human brain. We need to add layers and depth to our coding to truly mimic our intelligence.
It’s not ready yet. It will be in a couple more years. It’s inevitable. This fake news is just to calm the goyim so they don’t ban it outright. When it happens it will be too late to stop it.
80k Americans lost their jobs in May. 4k of them were directly replaced by AI. This percentage will only increase over time.
Because it was all parlour tricks created by a massive army of jeets and legally questionable data collection.
How is the hype dead? everyone fucking loves bing.
I use it to write half my code now, yep it's just a meme, retards
Because the hype was all lies to sell ads on technology blogs.
>completely fabricated headline OP pulled out of his ass and into MS Paint
>thread going to 300 replies
never change, 4chan
Your email address will not be published. Required fields are marked *
Save name and email for the next time I post