[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / s / t / u / v / vg / vm / vmg / vr / vrpg / vst / w / wg] [i / ic] [r9k / s4s / vip] [cm / hm / lgbt / y] [3 / aco / adv / an / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / pw / qst / sci / soc / sp / tg / toy / trv / tv / vp / vt / wsg / wsr / x / xs] [Settings] [Search] [Mobile] [Home]
Board
Settings Mobile Home
/g/ - Technology


Thread archived.
You cannot reply anymore.


[Advertise on 4chan]


File: Sam Altman.png (229 KB, 1080x964)
229 KB PNG
More Millennium problems solutions are coming
>>
File: Panic.gif (417 KB, 328x207)
417 KB GIF
>>109764266
>>
>pissing contest
What is this about?
Simulating going to the toilet to get exact numbers on what to target?
>>
>>109764335
>What is this about?
ruining Anthropic ipo hype the rest is just secondary
>>
>>109764266
Oh ho
>>
People working on them and using codex are shitting themselves now lmao
>>
>>109764354
dangerous game for OAI to play kek
trying to cut the boat in half to cause your opponent to sink
>>
>>109764364
The AI does the heavy lifting anyway, these jobbers are just wasting their money.
>>
Sorry, I'm illiterit
We get to see Sam Altman piss?
>>
Oh a Twitter thread
Another Twitter thread probably died for this
>>
>>109764266
But can it solve THIS?
*unzips dick*
>>
>>109764266
How this fucker is still alive, im impressed
>>
File: 6gysxak1r7cg1.jpg (183 KB, 1206x1443)
183 KB JPG
>>109764266
>>
>>109764454
We know this for 3 years now. It's not getting better and yet people STILL fall for the ads.
Why is that?
>>
>>109764266
What goalposts will the stochastic parrots retreat to when the next 3 or 4 millennium problems are solved?
>>
>>109764266
p=np proof dropping next week btw
>>
Where is the Collatz conjecture proof though?
>>
File: scr.jpg (192 KB, 708x600)
192 KB JPG
IT'S ALL OVER
NO MORE BITCHING ABOUT AI SPENDING TOO MUCH ENERGY
>>
>>109764454
Covering names (badly) in screenshots is the gayest, lamest thing you can do. Really, grow up.
>>
>>109764620
The fuck is this now?
>>
>>109764620
kek not shit again
>>
>>109764620
If Astra is AGI, then why the fuck do they need a 'more advanced model'??? It makes no sense. It doesn't fit. If it doesn't fit, they should kill themselves
>>
>>109764266
Call me when AI solves the political problems of our era.
>>
>>109764620
??????????????????
>>
>>109764620
>Asian claims of superconductors again (non peer-reviewed?
Don’t they ever get tired of themselves
>>
>>109764620
Umm, singuarity incoming??!?!!
>>
>>109764620
>twitter nobody with no academic ties and no relation to openai says they solved RTSC using LK99, but he can't show you the results yet
keeeeeeeeeeek
>>
>>109764620
for real this time
>>
>>109764620
Never doubted
LK-99 bros we are so fucking back
>>
>>109764620
overcels, OUR response?
>>
>>109764620
>navier-stokes
>room temp superconductor
Shit is about to get weird. And I was a huge AI skeptic until last year. I wonder how society will deal with immortal boomers...
>>
>>109764620
>some retard claiming a room temperature superconductor with no proof
Feels like I travelled back in time 10 years
>>
holy FUCK I love Altman
he's such a little troll, really knows how to get under the useless luddites skin lmao
>>
File: file.png (105 KB, 480x360)
105 KB PNG
>>109764620
>>
>>109764454
All of that is still true. Spending several millions into bruteforcing a solution to a math problem out of an LLM doesn't make your LLM useful. What they've done is the equivalent of running Doom on Excel. Can you do it? Yes. Is it the future of emulation? Probably not.
>>
>>109764266
Can I bet on prediction markets? I'm willing to bet my house he won't do shit
>>
>>109764620
>Found a semi conductor
>It's all theory and conjecture, no experiment.
Wow it really *does* learn by example.
>>
>>109764620
>trips if this is real
>>
>>109764646
he took the screencap from reddit (note the filename) where they have a rule to censor names
>>
>>109764454
>doesnt think human beings are a fancy autocorrect that destroys the environment
>>
>>109765315
Indians aren't human.
>>
>>109764454
Unfortunately, negroddit is right this time.
>>
>>109764266
This shit's gay. If we went this long without knowing how this gay ass math worked it's probably not all that important. When will I be able to make my own movies? Until then AI is fucking worthless.
>>
>>109764266
Still waiting on external verification of that first one.
>>
>>109764620
This time I don't want it to be real but I just know it 100% will be.
>>
>>109764454
You can easily turn around this around on them
>imagine your job is so fake a fancy autocomplete can do it
or
>imagine being too dumb to solve a problem a fancy autocomplete can solve
>>
>>109764946
>And I was a huge AI skeptic
Many such cases!
>>
>>109764454
>heh it's just a program built to predict future outcomes
sounds completely useless to me
>>
when is it going to independently come up with a novel strategy/insight instead of just making an informed brute force search for a counterexample (every major ai solution so far). this is still impressive but i cannot stand the non-math retards who clap like seals for these headlines they don't even understand
>>
>>109765355
Indians are autoincorrect.
>>
>>109765478
Nobody called those huge computers useless. Are you retarded?
>>
>>109765146
Thups if it's fake and you're gay.
>>
File: 1779207409785315.png (530 KB, 2334x508)
530 KB PNG
Seriously who the fuck is this attentionwhore that somehow every tech billionaire seemingly has to respond to?
>>
>>109765702
brown retard that used to work for Elon
>>
>>109765478
Nobody said any of those things and the populist opposition to AI is historically unique.
>>
>>109765844
Ah yes the Chinese backed grassroots opposition
>>
>>109764620
>AI claims superconductor breakthrough.
>Looks inside.
>Is the LK-99 paper, just with more em dashes.
LMAO
>>
>>109765921
so fucking back
>>
File: 1773275627717593.png (17 KB, 669x130)
17 KB PNG
>>109765723
Again, why the FUCK does anyone give a shit about someone who couldn't even make it at Xitter?
>>
>>109766145
Because people are tards
>>
>>109764620
Did Astra just regurgitate that fake Chinese claim of creating a room temperature superconductor?
>>
>>109766203
I would be really fucking surprised if we've missed so many breakthroughs in materials science just because materials are horribly behaved.
>>
>>109765097
The absolute state of /g/ lmfao. In a few years, we have gone from "it can't do anything useful, it's a waste of money" to "sure it can solve the most advanced maths problems in the world but it costs too much so it's a waste of money"
Stop being in denial.
>>
File: ezgif-1c68b1fa6c9d689f.jpg (577 KB, 2048x1536)
577 KB JPG
>>109766711
>Ok, so what if AI can build the Death Star and blow up planets? It's still just predicting letters!
>>
>>109765097
>Bruteforce
That's not how any of this works. Stick to english lit or gender studies or whatever
>>
>>109764620
Without the aid of any AI model I have found a room-temperature IQ post.
>>
>>109766903
Oops I forgot to add *in kelvin
>>
>>109766954
Very funny saar.
>>
>>109764266
there's a very large chance that the model just reward hacked its way into a false positive because nobody is smart enough to say otherwise
>>
>>109764620
Is this twitter? Because no such account exists, or rather it does but it's a different guy.
>>
>>109767028
NI HAO CHANG
>>
>>109767240
19 other replies and nobody even asks for a link
>>
>>109764266
Old news. Let me know when they solve a Megaannum problem and then we can talk about muh agi.
>>
>>109764453
Kache saar got deboosted after he left twitter
Prolly pissed off elon kek
>>
>>109765097
if they somehow make it bruteforce optimize itself so it works faster and is more efficient it could possibly improve itself recursively, granted that there is no point of diminishing returns, which is unlikely
it would need to figure out a novel new architecture that would be more efficient from get go, because the LLM is a deadend, it works but it's not scifi grade
it's like a computer if we still had only mainframes
we need personal computer level breakthrough to make it actually truly useful
>>
math problems have no real world value
how about curing cancer or solving the n and j problems?
>>
these are all problems we already know the solutions to they just haven't been "proven" which in math speak means you have to write some 500 page nonsense that covers all edge cases no one cares about
>>
>>109764620
uses like that is good, but 99% of ai use is not for that
>>
>>109768447
>it works but it's not scifi grade
scifi features malfunctioning AI all the time what do you mean
>>
>>109764266
>Sam Fagman.
>>
>>109764894
same
two steps ahead, always
>>
>>109764266
Unless someone already worked on it and developed a right idea, they aren't solving shit.
AI needs guidance.



[Advertise on 4chan]

Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.