Here's your AGI bro.

bro

This is just the weakest version of the model, Sam has promised even stronger versions they weren’t able to unleash yet. Be patient

holy shit

gpt-6 will fix this

a LLM with the mathematical abilities of a PhD level researcher... it's truly over for humanity...

We're in the age of computers that suck at math. What a time to be alive

Holy shit.

>>106192875
It can’t tell that 5.9 and 5.90 are the same

>>106192890
That's honestly kinda expected considering LLM architechture.

>>106192343 (OP)
But that's correct. I don't get it.

>>106192929
I suggest you try it with a calculator then.

>>106192875
Protip: always ask an AI bot "are you sure?" any time you ask it to do a math question.

>>106192343 (OP)
phd intelligence in your pocket.

>>106192429
Just 3 more billion

>>106192875
Do AI this smurt? I just out of date and never used it.

>>106192429
This. Luddites coping and seething. You will lose your job.

>>106192890
It probably confused them with software version numbers.

>>106192875
>"wait...that is 5.9"
same energy as
>*record scratch*
>"Yep, that's me. You're probably wondering how I got here."

>>106192990
are you sure about that?

>>106192343 (OP)

>>106195184
>Want me to show you a quick trick so you never mess this up again?
they must have added a passive aggressive tone to make it seem smarter

>>106195219
They just created the fuckwit coworker. What a leap forward in technology.

>>106192990
>the first thing computers could automate
>the first thing automated computers couldn't
palpatine-ironic.tiff

>>106192509
Isn't that quite literally imposible or am I missing something?

>>106192429
"""GPT-5""" is just a fake frontend for all the models which defaults to the minis as often as possible to save costs while the user is hopefully none-the-wiser.

And yet they somehow didn't anticipate or care that would be obvious to anyone with a brain within a few minutes of using it. This is mind-boggling self-sabotage.

4o wasn't perfect but they still could position themselves as the market leaders of its back. Now they've destroyed their reputation and credibility with this release and made their name a joke.

>>106195900
*off its back

>>106195184
Gaslight it. See how it holds up to user input insisting on something wrong

>>106192509
x - 7 = 19 + x

19 - 7 = 13

x = 13 + x

13 + x = 13x

x = 13x

x divided by x cancels out

the answer is 13

>>106195184
I don't follow. He got the right answer

Restart everything and when he gives you the wrong answer say "are you sure?" And he will try again and admit he fucked up. If he's right (like your img) then he lands on the same answer

This hack will save your life one day

>>106195431
No it's not possible that's the point it's a meme

>>106196411
Pic for proof

>>106197004
Continue to ask it if it's sure it's the right answer. Don't insist one way or the other, just keep asking it if it's sure.

>>106192429
Two more weeks

>>106192343 (OP)
I had a massive brain fart parsing this equation. The single digit after the decimal from the number 5.9 tripped me up.

>>106195127
CuteGPT

>>106192343 (OP)
its fake. even a retard version of Claude can solve equations with 4 variables

>>106192429
>just 2 more weeks

>>106196973
You can show that no solution exists

>>106195184
>Want me to show you a quick trick so you never mess this up again?
That's comedy gold

>ask the chatbot to solve simple math
>gets the answer wrong
>blames you for getting it wrong
>refuses to elaborate
>leaves

Why is GPT5 lagging out firefox? Multiple 4k videos, fine. One text response, "this page is slowing down your browser [kill]". Is Sam saving costs by having GPT run locally?

>>106199711
God it's just like a human, we were supposed to create something better, not another retard like us!

>>106195431
the answer is x belongs to an empty set

>>106199740
I've been enjoying reading horror stories of it fucking up in business settings.

>chatgpt, why did you not fix all the TODOs I assigned you to fix
>oh yeah there was a lot so I just did 4 or 5 and then got lazy and marked all the rest as solved without actually doing anything :^)

At least it's honest about it, that may be slightly better than your average junior.

>>106199667
it's not fake, it's just that without reasoning gpt5 is sloppa

what datacenter size to make this right?

>>106192343 (OP)
GPT-5 is based (if you are retarded)

>>106192875
so its Actual Indian on the other side
altman scammed everyone again!

Even Copilot, in both the fast and the new GPT5 model and even telling it that it made a mistake.

I tried switching the . for , to see if maybe there was a bug there, but holy shit, it just keeps doubling down.

>>106199740
It copied most humans and therefore it is like most humans.

>>106192429
just 500 million more gallons of drinkable water

>>106192343 (OP)
Can it count the number of times a letter appears in a word now? Google can't.

AI is infallable, you are the one that is wrong

>>106201076
Better to be thought of as persistently operating in an alternative dimension of thought, rather than just fucking retarded.

lol

Bing Copilot's Quick Response

>>106195912
get up on the hydra's back

>>106201908
"""Smart""" (GPT-5)

>>106193164
See >>106199555
Pic rel is you.

>>106201740
>those tabs
I kneel

>>106199715
Claude does the same thing. It's just text on a page. How did they fuck up so badly?

>>106199715
You're probably mining crypto or doing some other computation for them lol

>>106199786
LLMs don't "reason" and never will. Hope that helps.

>>106192343 (OP)
> 5.9 - 5.11
> = 0.9 - 0.11
> = -(0.9+0.11)
> = -0.21
I fail to see the problem.
t. totally human

>>106202067
based and redpilled

>>106202067
you think humans do?

>>106202146
cringe and satanpilled

>>106202146
Most don't

>>106195968
so 13 - 7 = 19 + 13 ?
that's crazy

>>106202146
At least some subset of humans have the capability to reason

>>106199711
>elaborates falsely
>stays and pretends to be in a position of superiority against you
Man, I really hate LLMs at least how they're trained to behave as if they're always correct.
>Ask it to tell me a way to wrap a number without division, specifically tell it I cant use division in my hardware
>still proceeds to lecture me on how % is the correct way of doing things and how am doing things the wrong way
Why would you want to deal with a dunning krugger in your pocket?

>>106195219
it just can't distinguish between messages you wrote and messages it wrote. it interprets everything in its context window as something you provided. all llms do this

>>106192509

i was a happy user of 4-o and now theyve taken it
i had a feeling when i began using it and something was different and that this was a worse model
seems to be more censorship as well

>>106192343 (OP)
>Gazillion of dollars invested into this
AHAHAHAHA
When the bubble finally pops in 15 more percent, many will be left wondering what those dollars were even used for in the first place.

>using AI as an calculator

lets use a screwdriver to hammer a nail next

>>106195431
its impossible. Suptract x from both sides and you get a statement that isnt true.

I wonder what the original scene was about

first try baby

>>106205081

>>106205418
>heres your superintelligence
>just dont ask it to do math

>>106192990
Treat it like Demiurge from Overlord

>>106192343 (OP)

never trust chip
ti calc rand same result twice

>>106205603
So when is it going to start farming humans?

you now remember shills saying gpt5 was smarter than all mathematical geniuses combined kek

>>106205552
WRONG

>>106205552

>>106205803
4 boxes

>>106205860
fuck am retarded, but ur still wrong(although it could be valid if it werent a meme answer)

>>106192343 (OP)
>>106192875
Kek all it had to do was 5.9-5.11 on a python command line, has to be jeet backend.

>>106205746
It's an ugly 8 but still an 8.