>he doesn't know
they hardmogged the israeli fraud on all fronts
the only thing openai has in their advantage is name recognition
that's about good for two major lawsuits
I mean it's pretty clear that GPT-5 is just going to be functionally moronic. It will beat everyone at benchmarks again (as usual) but when it reaches the LMSys Arena it'll just hover around LLAMA-3-70b-instruct. Honestly I believe at this point that OAI's run out of ideas or they're too longhoused to release anything that's a mile ahead of everyone else. >b-but the generative videos >t-they said they solved generative audio and voice cloning a year ago
ok, I'll believe it when they actually release those as consumer products and not just demos.
3 weeks ago
Anonymous
>ok, I'll believe it when they actually release those as consumer products and not just demos.
Generative audio already seems to have other smaller players dominating this area. The only way OpenAI can beat them is give the complete experience i.e VIdeo+Audio+Text generator all at once
How much computing power will GPT-5 need to decide where or not letting 30 million people die in a nuclear detonation vs typing Black person as a password to disarm said nuke?
b***h ass, L3 70B is very close and with community effort it will become something truly great.
Also 400B model is coming.
3 weeks ago
Anonymous
>moAr CorEs!!11!!! >MoaR pARamEteRS1!!!11
3 weeks ago
Anonymous
>Also 400B model is coming.
So.. Would love to run it and I'm sure I have enough power laying around, what would it take to create a power extraction tech circle device, that I just pile all my tech on and it transfers the power it to my main computer? Kind of like a wirless charger, but a siphon device instead.
Lmao what a cope. Anthropic released Claude 3 Opus thinking they would be the new kings, only for OpenAI to release a new GPT4 days later that was just good enough to overtake it by a little. Absolute chad move
Not him, but I'm a millennial and knew what mogging is. I think you gave a good definition of it but I would also say it's when something is so superior to something else, mainly because I haven't seen BTFO used in ages.
What are you using it for? All i really use it for is code. For the image generators it looks like the local models might just win. The customization and fact that they are uncensored just mogs anything any company puts out. I don't think they'll let us have shit even as risque as what gets posted in /sdg/ till it's so commonplace no one even bats an eye, in like 5 years.
I use gpt4 for translation. The adding context helps a lot, and makes the translation sound more natural.
However, getting it to translate certain passages that remotely hint at violence and or sex is a pita. Frick that trash. Still hoping for a local uncensored model of that, no mistral is not there, neither is llama.
>Didn't they say 2025 they are releasing GPT5 or am i misremembering?
The CEO said OpenAI was going to release a new model in 2024, but he didn't know whether they would call it "GPT-5".
that's why they're delaying the release of new GPT versions as much as possible, they don't want to reach GPT6.66
fugg
3 weeks ago
Anonymous
Working as an phone order taker I once had to generate an entirely new order for a customer because her transaction number ended in 666. Doing so required canceling her order and doing the whole thing again. Was faster the 2nd time because I could reference the list from the canceled order, but people need to realize that sequential numbering doesn't mean shit when it comes to their superstitions. And skyscrapers skip the 13th floor. It's so dumb.
The MSFT drama was clearly a last-ditch effort to cash out while the company was still hot. Musk cucked that hard. It was preceded by a media blitz about "safety" whereby OAI tried to argue that they should have some sort of monopoly on AI products for "safety"reasons, and culminated in in some kabuki with Saltman in Washington DC. OAI really is over, the big players with huge datasets smell blood in the water and they have a lot more resources to throw at AI stuff than OAI and none of the principled baggage.
This isn't even about math, it's entirely wrong, even as an approximation. UART can't transfer just a single bit because there are no single bit registers, depending on the chip and everything usually you have a start bit, 1 B data and a stop bit with optional parity. This means that you need to transfer 10/11 bits with the baud rate frequency (which is same on Tx and Rx) and this is in ideal conditions (eg. the receiver CPU can instantly read from the UART driver buffer in which the character is written to.
>2 more weeks to AGI/replacing all tech workers/replacing all office workers/replacing all support/replacing all translators/replacing all teachers/OpenAI becoming profitable since 2022
Just 2 more weeks and 1 trillion $ in funding, trust the gay femboy israelite, he wouldnt lie
how did he do it?
who is he?
That’s Claude Shannon the inventor of the claude-3 LLM
qrd?
>he doesn't know
they hardmogged the israeli fraud on all fronts
the only thing openai has in their advantage is name recognition
that's about good for two major lawsuits
You might want to check who's behind Claude, Anon
QRD:
same story as usual
Lizardisraelite > gay israelite
simple as
They still havent released GPT-5 yet so we dont know how truly ahead they are with SOTA AI
GPT-5 should be released in this year and it will become clear where OpenAI stands in AI race
I mean it's pretty clear that GPT-5 is just going to be functionally moronic. It will beat everyone at benchmarks again (as usual) but when it reaches the LMSys Arena it'll just hover around LLAMA-3-70b-instruct. Honestly I believe at this point that OAI's run out of ideas or they're too longhoused to release anything that's a mile ahead of everyone else.
>b-but the generative videos
>t-they said they solved generative audio and voice cloning a year ago
ok, I'll believe it when they actually release those as consumer products and not just demos.
>ok, I'll believe it when they actually release those as consumer products and not just demos.
Generative audio already seems to have other smaller players dominating this area. The only way OpenAI can beat them is give the complete experience i.e VIdeo+Audio+Text generator all at once
How much computing power will GPT-5 need to decide where or not letting 30 million people die in a nuclear detonation vs typing Black person as a password to disarm said nuke?
1.21 gigawatts
I assume he's talking about everyone being impressed by Claude Opus and Llama3 and OpenAI not having released anything interesting in a while
(sora looks cool but people lost interest in it fast because they haven't released it and clearly are not going to)
ehh Opus is comparable but I wouldn't say it's better. Anyone saying L3 70B is comparable is coping hard
b***h ass, L3 70B is very close and with community effort it will become something truly great.
Also 400B model is coming.
>moAr CorEs!!11!!!
>MoaR pARamEteRS1!!!11
>Also 400B model is coming.
So.. Would love to run it and I'm sure I have enough power laying around, what would it take to create a power extraction tech circle device, that I just pile all my tech on and it transfers the power it to my main computer? Kind of like a wirless charger, but a siphon device instead.
Lmao what a cope. Anthropic released Claude 3 Opus thinking they would be the new kings, only for OpenAI to release a new GPT4 days later that was just good enough to overtake it by a little. Absolute chad move
>i've never seen a tech company mogged so hard so fast
That logo looks awfully similar to the Star of Remphan.
what's "mogged" and "yapping"? asking for myself
hello zoomer
mogging is when you btfo someone
tyvm
Not him, but I'm a millennial and knew what mogging is. I think you gave a good definition of it but I would also say it's when something is so superior to something else, mainly because I haven't seen BTFO used in ages.
holy newbie
Didn't they say 2025 they are releasing GPT5 or am i misremembering?
They still need to drop 4.5, it's probably coming June per that blog post leak and will become the new best model for a while again.
No, they amount of guardrails they have on this thing now has functionally moronic it
What are you using it for? All i really use it for is code. For the image generators it looks like the local models might just win. The customization and fact that they are uncensored just mogs anything any company puts out. I don't think they'll let us have shit even as risque as what gets posted in /sdg/ till it's so commonplace no one even bats an eye, in like 5 years.
I use gpt4 for translation. The adding context helps a lot, and makes the translation sound more natural.
However, getting it to translate certain passages that remotely hint at violence and or sex is a pita. Frick that trash. Still hoping for a local uncensored model of that, no mistral is not there, neither is llama.
>Didn't they say 2025 they are releasing GPT5 or am i misremembering?
The CEO said OpenAI was going to release a new model in 2024, but he didn't know whether they would call it "GPT-5".
Reminded that Sam Altman raped his 4 year old sister and people let it slide and memory hole it because he's a rich tech israelite.
Delete your post, what you said was extremely antisemitic.
satan knows
SAm alTmAN
whoa
that's why they're delaying the release of new GPT versions as much as possible, they don't want to reach GPT6.66
fugg
Working as an phone order taker I once had to generate an entirely new order for a customer because her transaction number ended in 666. Doing so required canceling her order and doing the whole thing again. Was faster the 2nd time because I could reference the list from the canceled order, but people need to realize that sequential numbering doesn't mean shit when it comes to their superstitions. And skyscrapers skip the 13th floor. It's so dumb.
>No fun allowed
is this the gentoo that smelly incel nerds on IQfy can't stop talking about?
The MSFT drama was clearly a last-ditch effort to cash out while the company was still hot. Musk cucked that hard. It was preceded by a media blitz about "safety" whereby OAI tried to argue that they should have some sort of monopoly on AI products for "safety"reasons, and culminated in in some kabuki with Saltman in Washington DC. OAI really is over, the big players with huge datasets smell blood in the water and they have a lot more resources to throw at AI stuff than OAI and none of the principled baggage.
what about Juicero?
WOW I LOVE CHATGPT.
Programmers will be obsolete soon!!! HAHAHA CS gayS BTFO
>Can't even answer the most basic questions
GPT has always sucked ass at math thoughever.
it hallucinates all the time
sucks at math = sucks at any task that requires real understanding.
This isn't even about math, it's entirely wrong, even as an approximation. UART can't transfer just a single bit because there are no single bit registers, depending on the chip and everything usually you have a start bit, 1 B data and a stop bit with optional parity. This means that you need to transfer 10/11 bits with the baud rate frequency (which is same on Tx and Rx) and this is in ideal conditions (eg. the receiver CPU can instantly read from the UART driver buffer in which the character is written to.
ChatGPT is not meant to replace programmers but there are definitely dedicated coding AI that are on the path to replace programmers
>2 more weeks to AGI/replacing all tech workers/replacing all office workers/replacing all support/replacing all translators/replacing all teachers/OpenAI becoming profitable since 2022
Just 2 more weeks and 1 trillion $ in funding, trust the gay femboy israelite, he wouldnt lie
(You) will be replaced whether you like it or not
only midwits are coping about it at this point
I mean, yeah, competition managed to catch up with GPT 4 after 1 year.
but don't forget OpenAI is 1 year ahead and is about to release ChatGPT 5.