[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: Bel.jpg (61 KB, 640x448)
61 KB JPG
Get ready to be replaced by SI
>>
Still hallucinates. Try again.
>>
File: 1765049441074185.png (11 KB, 500x500)
11 KB PNG
>According to rumors
>>
>>110249815
This is the model that solved the Millenium problem on his own.
>>
File: 1793189921717x-0.jpg (36 KB, 1079x381)
36 KB JPG
>>110249797
>>
File: 1783273033247593.png (116 KB, 724x625)
116 KB PNG
>>110249826
even your clanker says that a clanker didn't solve it kek
>>
>>110249797
>Accenture capped Claude at $29 a month per employee on October 1, after some staff had been spending up to $1,000 a month, 404 Media reported.
>Uber burned through its entire 2026 AI budget in about 4 months, then capped AI coding tools at $1,500 a month per tool.
>Microsoft has cut allowances too, and its AI chief Mustafa Suleyman told Bloomberg that Anthropic is extremely expensive and that Microsoft wants to eventually get rid of that cost.
>Meta had internal leaderboards ranking staff by how many tokens they used, and now it has introduced token budgets.
>Last month, some Apple engineers saw their daily AI spending cap drop from $300 to $200, Bloomberg reported. Some engineering teams could be cut to $100 a day next. Many groups outside engineering already sit at $100, and Apple has considered going as low as $25 for some of them.

lmao
>>
>>110249797
AI just can’t stop winning
>>
>>110249797
That would be impressive but is this something they can serve for real or is this a stunt?
And is this merely memoization of the context?
>>
It’s coming out tomorrow btw
>>
>>110249797
I have the solutions for everything I do, can frontier labs stop the pissing contest and deliver a faster model? I would trade a 2x opus 5.5 for whatever PhD retarded model both open ai or anthropic are planning.
>>
File: 1793888212904000.png (702 KB, 750x985)
702 KB PNG
>>110249844
>clanker
>>
>>110249848
My team plan is 100 bucks per month and its more than enough to rewrite my entire codebase every week, wtf are they doing
>>
>>110250030
That's a funny picture thoughbeit
>>
>>110249797
This account is a literal psychotic
Got a better S O U R C E ?
>>
>This time AI will replace you
>This time AI will replace you
>This time AI will replace you
>This time AI will replace you
>This time AI will replace you
>>
File: 1682780664332915.jpg (620 KB, 1850x1139)
620 KB JPG
>blue kekmark
>according to rumors
>singularity :OOO
>>
>>110249797
revese engineer hello kitty magical island, make no mistakes
>>
THE SNEEDGULARITY
>>
>>110249797
Finally, we will have superluminal space travel in a month!
>>
>>110249797
How many times have we heard this story?
>>
File: file.png (251 KB, 492x676)
251 KB PNG
>>110249844
>>110250030
Let's see what Grok has to say about it
>>
2 more weeks
>>
>>110249797
LETS FUCKING GOOOOOOOO
>>
>>110249797
That would require the context memory of a whole data center to do the work of a single white collar job. I don't think that will ever pay.
>>
one more week
>>
File: huh-cat-huh-m4rtin.gif (592 KB, 220x391)
592 KB GIF
Replaced by system integrators?
>>
File: 1793811485073150.jpg (77 KB, 640x403)
77 KB JPG
>>110250030
>>
File: 1793899139000608.jpg (95 KB, 1290x1810)
95 KB JPG
who's gonna tell the AI bros
>>
>context still not solved
>AGI anytime soon
>>
>>110249797
>TWO
>MORE
>WEEKS
>>
>>110252187
Context was solved ages ago, LLMbros just refused to implement it.
>>
>>110249827
way ahead of you :3
>>
>>110249815
OP is trying his best. Still can't seem to properly say "AI" though. Must be some prompting issue imo.
>>
>>110249797
Yes SI is getting extraordinarily more intelligent. However, there are so many "next model" bait accounts on X. Posts like picrel are just made up.
>>
File: 1781142314775289.jpg (28 KB, 1080x540)
28 KB JPG
>Trillions of dollars in doomed investments just to avoid paying white collar workers a proper wage
what the hell
>>
>>110249826
Luddite retards crying about muh ai psychosis in 3, 2, 1...
>>
>>110252232
>SI
lmao
>>
>>110252114
>>
>SI
>>
>>110249797
I'm other words you come back a week later and find that it hallucinated something early on and then burned a whole week's worth of tokens expanding on that error?
>>
>>110249797
the fuck kind of name is Bel?
I thought we were solar system and shit
>>
>>110249844
whatever they use for google search is wrong almost all of the time and can't be relied upon for anything
you're literally better off performing all of your google-tier searches in chatgpt or claude
>>
>>110252816
it stands for BELgian congo
>>
>>110252816
Apparently it's short for Betelgeuse, I have no idea why they went from celestial bodies to fucking campy children's movie characters but this just tells you the IQ of the people coming up with names I guess.
>>
>>110252796
Fraid so
>>
>>110253105
betelgeuse is a star, dumbass.
>>
>>110252431
In my culture, we don't say LLM, we say SI.
>>
>>110253172
no the star was michael keaton you retard
>>
File: 1746197168318609.gif (3.54 MB, 498x210)
3.54 MB GIF
>>110249797
>put down my computer near the ship hull
>hey JewGPT please weld this ship for me
>come back a week later
>absolutely nothing is done
So when is this thing going to do anything useful again?
>>
>>110253172
It’s actually a tiny black guy with a tiny head
>>
>>110249827
Truke
>>
>>110249848
So even microsoft is not using copilot, big if true
>>
File: 1769633752506.jpg (305 KB, 1280x1280)
305 KB JPG
>>110252697
/g/ in particular has been flooded with bots trying to push trump's AI rebrand. no other board has this, someone on /v/ tried the other day and people shat on him
>>
>>110250030
Learn to draw
>>
File: 1773620682843961.jpg (148 KB, 1762x960)
148 KB JPG
>>110249797
>the SI you ordered, sir
>>
>>110249797
>Experts say...
>Scientists decided...
>An anonymous source reveled...
Nigger.
>>
>>110255820
finally, the perfect model for /lmg/
>>
>>110249797
so is it whole week or 4 days? can't be both.
>>
>>110249797
How much porn can it write in an entire week?
>>
>>110253105
>>110253298
kek
>>
>>110249797
Sure sure, just 2 more time units
>>
>>110251404
Kek
>>
>>110255820
what is this tool?
>>
>>110252796
If you don't prompt it correctly, almost certainly, but if you do you could get copious work done without having to do anything.

Imagine if you could just have 10 different week long tasks running at the same time.
>>
>>110249797
Is it capable of not referencing "wrinkles" or "tensions" when assessing your deranged rants about womens, fags, and Jews though?
>>
>>110257050
AI psychosis everybody
>>
>a mid developer can be replaced by a commercial AI instance running on 50 GPUs in a datacenter somewhere burning more electricity than a small village
>developers getting fired and fear for their jobs

>a lawyer can be replaced by a qwen 2.5 running on an average modern laptop and a well maintained vector DB
>no lawyers are ever fired

Strange huh
>>
One day we will find out that AI companies were paying a bunch of third worldlers pocket change to spread FUD on 4chan, one day.
>>
>>110249797
Democrats won—it's TI now: Trans-Intelligence.
>>
>>110257118
None of this is true though. t. legaltech dev build ai meme saas slops.
>>
You already lost might as well sign your prayers
>>
>>110257135
But dont come to the bathrooms
>>
>>110257138
what kind of models do you use then?
>>
No it's definitely Bel-6.7 according to rumors. And it has a wicked sense of humor.
>>
>>110257254
A mix of gpt (5.6 before, 6 now since it's cheaper) (assorted sol, terra and luna depending on task) and claude (mostly opus, but some haiku. No fable). It depends on the task. One of the things that AI remains notoriously bad at is automated redlining and writing in an appropriate legal style, but even text classification on large documents (given that we have cost and timing constraints) requires pretty intense wrangling. Even the latest and greatest hallucinate a lot for things you'd think is quite easy, such as extracting the parties named in a contract while following the right output format.
It's quite good at single-document Q&A. Not perfect, but quite good. However, that just barely registers among legal workload in terms of time or expenses.
>>
>>110249827
wykurwiaj na wykop
>>
>>110252816
Like the prince



[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.