File: 1730976894086.jpg (26 KB)

26 KB JPG

>>108262852
Dumb goy, it doesn't matter if you know that it's a scam. As long as a politician is gullible enough to pay billions to add "AI" to lethal weapons is enough. And if those weapons end up murdering civilians by "accident," even better.

The social engineers, retarded and evil, plan to make "AI" the world government. The people are expected to believe "AI" is somehow intelligent, and impartial and ethical when it's actually just a voice changer of the social engineers. It's insanely dumb but it seems to be the plan.

File: 1764540150460667.jpg (49.7 KB)

49.7 KB JPG

>>108262852
>What is the point of AI?
False progress

>>108262852
a flex, I think
invention is just accidents, people fucking around to see what happens. Giving these discoveries a purpose comes later, or never

To help me write boring stuff like basic functions, css and html. It also finds bug pretty well.

>>108262852
to be my Google bitch

>>108262852
Literally every chatbot gotcha in these sorts of articles are just abusing tokenization errors to trick it.

Gooning

Text prediction.

>>108262878
It’s insanely dumb, but most normgroids believe it anyways because they watch too many movies and that’s colored their perception of reality.

>>108262852
There is no AI, stop calling it that. It's a simulation of intelligence, at best.
Stupid idiots.

>>108263676
Almost as if the intelligence is artificial…

>>108262852
Why would a LLM be good at logic. It's literally just text prediction

>>108263681
It's no intelligence. It's a simulation. There is no car in a drive simulation game, just an algorithm that portrays a rough approximation of one.

>>108262852
>Thing that can't use logic fails at logic
No way, who could've seen that one?
Why are these people even testing for logic in probability machines?

File: 1753653838195259.png (444.4 KB)

444.4 KB PNG

>i asked it this malicious trick question when it assumed good faith so its worthless now, my job is safe!

>>108263744
Some people believe its "smart" or "intelligent", for others it's just a black box so they go and test it.

b-but it said there are three Rs in strawberry...

>>108262852
Jingling keys for the masses. It's useful for translating or combing through large volumes of text, but outside of academic situations or porn generation it's pointless. Who the fuck wants to watch AI "movies" besides drooling retards? My coworker has been playing AI music lately and it's so repetitive and boring.

>>108263681
it's an LLM it has no intelligence

>>108262852
to be trained by indians kindly doing the needful for outlier and/ or pareto ai

>>108262852
Elites want to replace wagies so they can make $$$ and possibly kill everyone off and live in utopia (this is a fantasy but it's the thinking of some of them at least). And/or this >>108262878

>>108262852
Working along with a human being that can recognize the output and see if it's good.
If used in this specific setting, you get a faster human.

However vibe anything is absolutely fucking retarded and will lead to a catastrophe.

>>108262931
Not everything is purposeful. people in power are fucking retarded, and this ruins several of THEIR plans as well.
Probably no one, not even the CIA glow in the darks were expecting Sam Altman to do the RAM shit.

File: 1730465411457584.jpg (37.8 KB)

37.8 KB JPG

>>108263745
This is not a malicious trick question.

>>108262852
Search engine

>>108262852
turning short sentences into long paragraphs that I don't want to read

>>108262852
To griff investors and governments out of billions in investment and grants and consolidate that wealth in the hands of a select few

>>108262931
Delete this now asshole

>>108264890
and what's the answer?

>>108265067
042 of course.

>>108265260
>>108264890
claude got it even without tool use.

I got supergrok and I've been generating filth of huge beach ball fake tit girls in panty stores lactating for the past week. It has entirely taken over my dopamine channels. I haven't done anything for a week but generate videos, wait for the rate limit to go away, then generate more.

It's only for instant dopamine, that's it. I am amazed if I don't lose my job over this.

>>108265289
Post the explanation. They usually get the answer correct (because the answers are online) but the explanations have logical errors.

>>108265335
It didn't search. The reasoning gave 042, why would it matter how it reasoned to get it?
anyway it's trivial and probably saves on tokens to actually just tell it to use python to solve it

>>108265366
>The reasoning gave 042, why would it matter how it reasoned to get it?
Because it pulled the answer out of its trained data.
Ask it to generate a similar problem.

>>108265335
>>108265366
chatgpt also got it right

>>108265372
no it didn't, this is cope

>>108265375
>>108265372
claude wrote a script to make new puzzles and a new instance got it right (so did chatgpt)
it hardly took any time at all
what's the last model you used, gpt 3.5?

>>108264961
kek

>>108265375
>>108265395
Come on, show it.

>>108265415
It got the right answer. Even on the new problem which was just created.
The internal reasoning tokens don't matter.
You lost.

>>108265465
It literally costs you nothing to show it getting the reasoning correct.

>>108265498
Why does it matter what it says? The result is all that matters
Try it yourself on your own claude or openai account.

File: restate.png (98.8 KB)

98.8 KB PNG

>>108265549
Yeah, so as I said.
Pulled the answer from training set, fails logic. Nothing has changed.

>>108265596
nope, i made a new puzzle and they both oneshot it
you lost

>>108265596
>>108265605
Here is a summary of the reasoning:

Eliminate: Clues 4 and 5 rule out 7, 3, and 8, but confirm 0 is in the code (just not in the last spot).
Find the 3rd Digit: Clue 3 (206) proves 2 is in the code (by ruling out 6), and Clue 1 confirms 2 is correctly placed in the last position.
Place the 0: Since 0 can't be in position 2 (Clue 3) or 3 (Clue 5), it must be in position 1.
Find the Middle: Clue 2 (614) rules out 6 and 1, leaving 4. Since it was in position 3 in the clue, it must belong in position 2.

Final Code: 042

>>108265605
Sure, buddy.
Where is this puzzle? Can I see it?

>>108265626
>Clue 3 (206) proves 2 is in the code (by ruling out 6)
On its own, clue three doesn't rule out 6. Really, clue 2 rules out 6, but the summary said that in the fourth step, which doesn't rule out 1 on its own.
That solution summary is pretty shit.

>>108265635
you dumb faggot, just give the chatbot the puzzle and ask it to make a new one using python
you're using the free shit tier version aren't you lmao

>>108265684
It costs you nothing to show the output you got.

>>108265700
Why do you want to see my output instead of just generating it yourself?

>>108265739
Because I want to see what you are not lying.

>>108265700
Anyone copying exact LLM outputs to 4chan will be doxxed in the future

>>108265743
Bad faith, you're trying to get me to dox myself by posting LLM outputs. Ask it yourself. Or are you banned?

Regardless you lost. Anyone wh

>>108265759
Anyone who isn't banned can try themselves and see LLMs can create their own novel version of the problem and a new LLM context can one shot it as easily as the original version.

The "i-its in the training data" cope is completely holocausted

Bro. Stop arguing with a AIJeet, lmao

>>108262852
my 27b Q6 is better than the commercial shit you idiots pay money to use. stop being lazy and fine tun a 27b or even a 12/24b if you need. it's hard. your average retarded /g/ cou.... well maybe someone that retarded couldn't do it - so, maybe.... maybe stick with your $20 a month ChatGPT after all

>>108262852
It's an excellent scapegoat that can be blamed for all sorts of war crimes and unwarranted arrests. Surely these fucks in government aren't actually retarded enough to believe LLMs are capable of serious reasoning. It's a fucking token generator. The US government understands this right??

>>108265759
>Bad faith, you're trying to get me to dox myself by posting LLM outputs.
The fuck? Why are you paranoid?

File: verynicepuzzle.png (102.5 KB)

102.5 KB PNG

ChatGPT level puzzle creation.

>>108265810
Why are you asking for the doxxable output from my request? Why would I trust a bad faith malicious poster on this website? Do it yourself

>>108265929
Explain how the output is doxable

>>108262852
The tech I still like. But it's current trajectory is merely to money launder, and laugh at the proles.

>>108266090
it's a completely unique n-gram

>>108265929
>excuses
As expected.

>>108262852
>What is the point of AI?
To make wealthy people even wealthier and providing a source of cheap entertainment to distract the masses.

>>108262852
>What is the point of AI?
money siphon

>>108262852
LLMs are probability based, they never had any logic to begin with. It took scientists this long to verify what's obvious as fuck from the napkin?

>>108264882
>Not everything is purposeful. people in power are fucking retarded, and this ruins several of THEIR plans as well.
This. The big mistake alot of conspiracy people make (which is an easy one, admittedly) is thinking that there's one big plan everyone is in on. There isn't. Our social betters have about as many camps as we do. Sometimes, like when they collude to manipulate the stock market, it all works out: they all hold stocks, so it all works out for them. Other times it goes horrifyingly wrong. I'd be willing to bet stuff like the wars in Ukraine and Israel and now Iran were part of this: The camp who benefited from peace in our time have been getting royally fucked by the camp that benefits from war.

Israel, ironically, is a prime example of this in practice. Outsiders -- anti-zionists especially -- view Israel as being a monolithic front, with all Israelis believing the same message and working for the same goal, but life is more complicated than absolutes. The zionist, pro-war camp is the one holding political power at the moment, but they're not the only camp, and Israel comprises a number of camps -- pro peace, fundamentalist orthodox, moderate, etc, and there's conflict, and there's tension, and the pro-peace camp might triumph over the pro-war one, or the fundamentalist camp might secede and break the nation in half, for instance. Just because our social betters occasionally collaborate doesn't mean they're all on the same team. Do you really think the opportunistic dirtbags who run our society could trust eachother for more than a few seconds?

>>108267663
>Israel, ironically, is a prime example of this in practice. Outsiders -- anti-zionists especially -- view Israel as being a monolithic front, with all Israelis believing the same message and working for the same goal, but life is more complicated than absolutes. The zionist, pro-war camp is the one holding political power at the moment, but they're not the only camp, and Israel comprises a number of camps -- pro peace, fundamentalist orthodox, moderate, etc, and there's conflict, and there's tension, and the pro-peace camp might triumph over the pro-war one, or the fundamentalist camp might secede and break the nation in half, for instance. Just because our social betters occasionally collaborate doesn't mean they're all on the same team. Do you really think the opportunistic dirtbags who run our society could trust eachother for more than a few seconds?
hello hasbara
>umm it's not all jews its just 90% of us

>>108262852
to extract money from rich dumb boomer investors.

>>108267663
>The zionist, pro-war camp is the one holding political power at the moment, but they're not the only camp, and Israel comprises a number of camps -- pro peace, fundamentalist orthodox, moderate, etc,
Maybe, but they all agree on exterminating us goyim and work together when it comes to that. Any "disagreements" are merely theater they can point to and say "look goyim, we're clearly not united so you have nothing to worry about". Even if they actually disagree on some things, they all agree thhat solving those diagreements will only come after the goyim problem is dealt with.

>>108262852
>What is the point of AI?
Too vague. Kinda like saying "what's the point in a computer" or "what's the point in $tool".

It does things.

>>108267604
I know they're imitation machines but their output is good enough for me.

File: Screenshot 2025-12-20 at 16-49-10 LMArena.png (87.1 KB)

87.1 KB PNG

>>108262852
>scientists found
That's nice. But I found modern civilization's fatal flaw: it keeps producing golems that can't make their own observations and will ignore reality, even as it keeps slapping them in the face, waiting for it to be officially certified by state/corporate-owned """scientists""". "Scientists" didn't find anything. Random people did. It's been known and confirmed beyond any doubt for ages.

>>108267663
>Israel, ironically, is a prime example of this in practice. Outsiders -- anti-zionists especially -- view Israel as being a monolithic front, with all Israelis believing the same message and working for the same goal
I've lived in this shithole for most of my life and I can confirm that most Israelis are a monolithic front as far as defending denying/defending Israel's crimes and undermining the rest of the world is concerned.

File: 1747367460059766.jpg (181.8 KB)

181.8 KB JPG

>>108262852
selling you shit you don't need, ads, massive surveillance and firing wagies

>>108267663
Em-dash therefore didn't read. Try again.

>>108262893
you're absolutely right.
"Hey goys, let's create something, which we didn't need at all!"
Sherlock Holmes put it best when he said that even though encyclopedias are literally available anywhere, people are still stupid. AI solves nothing, it is just laziness.

>>108264862
This is pure delusion. The ML/AI models as they stand are very inefficient and cost more then wages of the people they intend to replace. There is a massive disconnection between the parties investing in ML/AI and those who actually understand the technology and its limitations.

>>108269374
Its not an em dash you retard. Fix your font rendering.

>>108262852
cheap white collar labor for the rich

>>108269979
Cry more, clanker, I want to see if you get rusty.

AI is biased by the training set, it's failing all of these puzzles because it is biased to think of anything worded like a puzzle as a puzzle. This is also why AI all sounds the same and why hallucinations happen.

This can't be fixed by the way, enjoy this problem forever.

File: 1500033272324.jpg (8.1 KB)

8.1 KB JPG

>>108262852
>they needed actual SCIENTISTS to find this FATAL flaw

My fucking sides

File: AI scaling fallacy.jpg (452.3 KB)

452.3 KB JPG

>>108262852
It makes grifters a shitton of money.
It has no other purpose. You're looking at a massive pyramid scheme held aloft by a chatbot that can't spell simple words.

But if you give us 2 trilly more, it'll wake up and become a real boy. Just 2 trilly and 200 million more for Israel.

>>108270009
This, and it's incapable of ever fixing it itself, since it has no way to analyze its own training data or training weights. So we make bigger and more expensive models and in return we get... slightly better autocomplete.

>>108265605
>>108265684
If you didn't create a new formal system with novel inference rules like Douglas Hofstadter did with the MIU system or the pg system, there's nothing new about your puzzle

>>108262852
>scientists found
didn't need to read futher to know this article was garbage.
everyone knew this from the start lmao.

>>108262852
"AI" is a neat tool. It's not a god machine. Investors, to varying degrees, think they're financing a god machine that will be the ultimate transformative tool with infinite applications civilian and military. Some literally feel they're aiding in the creation of a digital god.
They're rubes.

>>108269037
Pretty much their entire is about Jewish supremacy. It's not hard to indoctrinate kids into that.

>>108271289
Adding "scientists" makes the shit article look more profound they think

>>108271856
always the same case.
when something begins with.
"experts", "scientists", "former astronaut/cia operative / whatever"...

you know that it's gonna be garbage, as if it wasn't they'd not feel the need to hype you up with some bs authority figure.

File: file.png (1.2 KB)

1.2 KB PNG

>>108269374
>>108270002
It's two hyphens masquerading as an EM dash. Look: -- I've done it again.

>>108262852
H*mans fail basic logic test, AI aces more and more tests.

File: 11111.jpg (98.5 KB)

98.5 KB JPG

>>108264961

File: file.png (424.6 KB)

424.6 KB PNG

>>108262852
To stop the anti-Christ. Pay attention.

>>108272412
>hasbara department specifically asks jewgpt not to use the em-dash not to give it away that they're spamming forums with their bot
>the bot cleverly replaces the em-dash with two hyphens
AGI in two weeks.

>>108265596
Isn't it 579?

If it was 569 then Rule 3 would say "695 - Three numbers are correct but wrongly placed"
And it doesn't comply with Rule 2 "327- One number is correct but wrongly placed"
Also in it's Conclusion logic it misquotes Rule number 2 to be "None of these digits are in the final code".

So it misinterpreted Rule 3 and copied Rule 4 onto Rule 2.

>>108270009
I don't think that's it.
It's not good at puzzles or logic, because it seems to take hard rules as suggestions that have a lesser or greater weighting.
If it has other tokens with less weighting than the "hard rule" token, it's only a matter of how many low-weight tokens does it take for the LLM to ignore the "hard rule" token.

>>108262852
Prevent the stock market from crashing after the cash injection during covid. How can we park $5T?

To take money out of your pocket and place it in mine

File: hi0bye.mp4 (2.2 MB)

2.2 MB MP4

>>108262852
>>108262852
to make videos like this

>>108262852
>>>/wsg/6096512

>>108262852
More money for the handful of Jews that have all the money already

>>108262852
global surveillance

>>108265260
How did I get 102

>>108262852
Global mass surveillance and control
Those things take a lot of eyes sifting through a lot of data in real time all the time
The toys we get to play with is part bread and circus for us to be open to implementation of the surveillance and control AI part smokescreen for what type of AI they are developing for surveillance and control (not generative)
We lack the means of generating enough energy, once this realization sets in within the dark alley rooms the worlds government likes to discuss itll be dropped, one day everyone will just stop talking about it, companies will quietly move away and things will be quiet before these Underworld dogs find a new car to chase

>>108262852
i've never needed to write a regex since ai

>>108262852
>What is the point of AI?
It's to single out retards who think written text equals to knowledge and intelligence because they work boring office jobs that include no math, no experiential learning, no skills that require fine motor movements, or dealing with physics, chemistry or logic
So yes, they will be replaced because their entire job is basically writing text or dealing with text of some sort

>>108264961
>>108272470
i see so the solution is to run every question on 6 different models and have a 7th model evaluate the answers

this is definitely cheaper for society than hiring some guy (whose livelihood you have to support regardless if he is employed or not) to tell you the right answer

We can't even simulate the 300 neurons of a c.elegans correctly, real AI is going to take a few more decades. LLMs are the best we can get.

>>108277334
>this is definitely cheaper for society than hiring some guy (whose livelihood you have to support regardless if he is employed or not) to tell you the right answer
It may not be cheaper now but it sure can get cheaper over the years

>>108262852
>one day everyone will just stop talking about it
Nerds will still use it to goon. There's no going back from that.

>>108277334
>i see so the solution is to run every question on 6 different models and have a 7th model evaluate the answers
No, it answered correctly for me on the first try. The solution is to fire all the obsolte h*mans from their positions because they are too stupid for this world.

AI has always been bad at basic logic and good at complex issues

>>108277366
>LLMs are the best we can get.
They are already getting good at taking over white collar tasks

>>108262878
That's retarded but I do think AI should be integrated into government systems. It would be very useful for data, imagine a system that tracks all the metrics collected by all the various government agencies and generates reports and charts based on the data.

>>108262852
to waste resources and accelerate the collapse of the modern world

>>108263406
let's walk to the carwash together, think of the planet

>>108277454
>No, it answered correctly for me on the first try
anon... you might have to ask grok to teach you how to not be retarded...

How is it that LLMs fail to program a basic Go engine? Human plays as white against 9-stone handicap on a 9x9 board and black is either dead or loses a lot of territory.

Literally every chatbot trying to defend """AI""" ITT is hallucinating incoherent rhetoric off of corporate propaganda. >>108263406 is one example.

>>108278362
And to add, Scam Altman can tell you that raising a child is pointless, but even a 3 year old can beat an LLM's offerings.

>>108278387
>Scam Altman can tell you that raising a child is pointless
Al Scatman is gonna try selling you AI babies next.

File: 1764205652631.jpg (211.4 KB)

211.4 KB JPG

We will have AGI this year

>>108278608
Yes, any day now.

>>108277421
Yes, because the hundreds of data centers that will require the electricity and water of entire towns and cities each, will make it cheaper. One thing that I don't hear people talk about is how all these companies are operating on massive deficits and creating bigger and bigger operating costs for themselves. I would think that this shit will have to get more expensive one day if anything. I say, enjoy your ai porn and ai girlfriends while you can, because it will get prohibitively expensive one day when these companies have to actually start making money that isn't from investors.

>>108264890
my local qwen says: The Code is: 042

Gotcha AI! Get ready for the big bubble burst! Just 2 more weeks, trust!

>>108269364
this is based because women deserve it and so much more

>>108269021
how would bob's response be justified if jane's solution satisfies all parameters. the question didn't say "a specific set of distinct primes."

>>108278608
Weren’t most of the other AI modules on Glados build there to make her dumber? Weezly here depicted as the Symbolic AI was a moron.

>>108277421
You still have to support the guy while he is unemployed. Is it cheaper for society with robots to make starving third worlders build a bridge or use robots? Robots cost less to run and build but it is still the villagers who are cheaper because if they won’t build the bridge you will still need to feed them or they will try to find another job. The only way it can be cheaper is if you make everyone poor or start doing genocide to reduce population.

>>108262852
circus for goyim, nonwhites, 2 digit iq etc, just like VR or traveling to Mars, there are hard physical limits that we still can't beat and until then it's all a meme and gimmick

>>108264890
I gave Qwen the image and asked it to find the code. The code is 042.

>>108282629
Looking at this logic puzzle, I'll analyze each clue systematically:

Clue 4: 7 3 8 - Nothing is correct
- Numbers 7, 3, and 8 are NOT in the code

Clue 5: 7 8 0 - One number is correct but wrongly placed
- Since 7 and 8 are eliminated, 0 must be in the code
- 0 is in position 3 here, but it's wrongly placed, so 0 is in position 1 or 2

Clue 1: 6 8 2 - One number is correct and well placed
- 8 is eliminated, so either 6 (position 1) or 2 (position 3) is correct and well placed

Clue 3: 2 0 6 - Two numbers are correct but wrongly placed
- Two of these three numbers are in the code but in wrong positions
- We know 0 is in the code. In this clue, 0 is in position 2
- Since it's wrongly placed here, 0 must be in position 1 (can't be position 3 from clue 5)
- Position 1 = 0

Now, from clue 3, since 0 is in position 1 in the actual code, and we need two correct numbers wrongly placed:
- If 6 were in the code, from clue 1 it would be in position 1, but position 1 is 0
- So 2 must be in the code and from clue 1, 2 is well placed in position 3
- **Position 3 = 2

Clue 2: 6 1 4 - One number is correct but wrongly placed
- 6 can't be in the code (positions 1 and 3 are taken)
- So either 1 or 4 is correct
- If 1 is correct and wrongly placed in position 2, it would need position 1 or 3, but both are taken
- So 4 must be correct and wrongly placed in position 3
- Therefore, 4 goes in position 2

The code is: 0 4 2

Let me verify:
- 6 8 2: 2 is correct and well placed (position 3)
- 6 1 4: 4 is correct but wrongly placed (in position 3, should be position 2)
- 2 0 6: 2 and 0 are correct but wrongly placed
- 7 3 8: Nothing is correct
- 7 8 0: 0 is correct but wrongly placed (in position 3, should be position 1)

Answer: 042

>>108262852
writing code

>>108281674
The same way it happened with dotcom bubble, they will go bankrupt spending on all that infra first then surviving incumbents and new players buy up all that infra for cheap and make it financially viable moving forward.

Also economies of scale is a thing

>>108282450
You are making the assumption that businesses are going to put long term interests in mind instead of short term gains

>>108262852
To make poorfags/trannies/"artists" SEETHE
AI is KINO

>>108262852
Government propaganda and deniability

That chatgpt wrapper app that counts calorie (made by 17 yo) was recently acquired for 50m.
Why are you still poor?

File: 1751670982939574(1).mp4 (2.7 MB)

2.7 MB MP4

>>108262852
"And it performed great signs, even causing fire to come down from heaven to the earth in full view of the people"
-Revelation 13:13

>>108283459
cute dog

>>108267604
>LLMs are probability based,
wait til you hear about quantum mechanics

>>108262852
>The headline: "The Most Advanced Models Are Failing Basic Logic Tests"
>The article: no logic tests
>The paper: no logic tests, just summaries of other people's papers dating back to 2022
>The appendix to the paper: "ChatGPT-3.5 Turbo achieves only 25.1% accuracy"

How will the AI companies ever recover from this embarrassing revelation?