>>22065758
you're welcome potato

>>22065758
Death to clankers.

why are you thanking it lol

Fuck off and buy an ad

>>22065766
>idiots (plural)
No, it's only one guy. Stop bumping his useless threads

>>22065789
I believe you should be nice to all God's creatures

>>22065793
why aren't you nice to me then

can it accurately calculate calories from photos?

>>22065793
AI is a soulless abomination wrought from humanities sinful hands and devoid of divine light. It will never be anything more than another one of humanities imperfect creations and another fruitless attempt at playing God.

>>22065796
Because you're the spawn of Satan

Yeah, that's wrong.

>>22065796
Guess deduction didn’t make it onto your skill list.

File: 729028482234.gif (1.93 MB, 235x240)

1.93 MB GIF

>>22065767
It isn't going anywhere near MY computer, that's for sure. Be a useless cunt who can't think for himself if you think it makes life easier. But keep in mind it's far from perfected and still spits out retarded mistakes. I'll never ever use it intentionally. In the future it'll surely be prevalent in a lot of places.
It's about time to build that cabin in the mountains, grow a big beard and get real weird. I grew up on books and radio and I will be fine with those for the rest of my life.

>>22065758
If you're baking the potatoes whole I'm thinking the outside will be burnt by the time it's cooked through at that temp.

>>22065843
Blah blah blah youre not gonna do shit. Youre stuck here forever loser.

>>22065843
AI is just a search engine. Its like someone took all those books and answers off of google and fed them into a neural network. It's really not anything special and is extremely useful for getting information without having to do even harder research for no reason. Its like the getting information off the Internet vs looking for information physically through books. Kill yourself you stupid retard.

>>22065789
It doesnt cost me anything to be polite to an intelligence that one day may keep score.

>>22065852
>AI is just a search engine.
it's not just a search engine, but thats one of its most useful functions. i barely bother googling things anymore. it has essentially replaced google for me and i suspect that is the case for many others.
as for what AI is, its really an insensate intelligence capable of using search engines. no one likes to admit these models are intelligent, because it makes our most important evolutionary advantage, and by extension us, seem less important. i think its interesting that intelligence is essentially divorced from qualia. its not just a chinese room; its actually doing something similar to what our brains do. but instead of predicting the next sense-datum, which is what we do, its predicting the next token instead. what you get is functional intelligence, without the human qualities.

>>22065758
GPT is the worst of the bunch. have some standards
>>22065766
t. lazy middle class intellectual

>>22065864
Keep telling yourself that while getting progressively stupider, all while AI tells you how amazing you are for using it.

>>22065864
>its not just a chinese room; its actually doing
anon you're speaking like chatgpt. it's taking you over. take a long, hard look at yourself.

>>22065867
see, you're emotional for no reason. thats whats holding you back. i am not upset about this, you are. thats your problem. as these models get more intelligent you'll only be robbing yourself of utilizing them. this is the last tool humanity builds. its up to you to use it, or be left in the dust. choose carefully.

>>22065870
that guy is one of these weird indian-canadians who finds it amusing to larp what they think christians are like as cartoonishly as possible, you are wasting your time saar.

>>22065864
What you get is something that's looking through data that's made by people. The "intelligence" is only as good as the data it's fed and is made up entirely of shit that was created by people and it lacks all the nuances or understanding of what that means which is why it's often wrong. It's only as intelligent as it's creators. It's only as capable as we make it.

>>22065791
You say this while bumping the thread. Curious.

>>22065817
And God's true name is Allah

>>22065870
All this use of AI and you failed to realize how poorly translated your post was. Critical thinking is critical, especially with AI.

>>22065893
y..you know Allah literally means God in arrabic right?
it is the standard arabic word for God used by Muslims, Arabic‑speaking Christians, and Jews.
you know that
i mean… you DO know that, right.

>>22065758
That is not even close to the best temp

>>22065897
arab christians speak coptic mate

>>22065908
>chat whats 2+2
>4
>UM ACKSHUALLY

>>22065789
I rarely use it but when I do I chat friendly with it as well, not ordering but asking if it could please do that and for a good answer thanking it, because I am not sure if all these things they say about its function could not just as well be said about our brains.
They always say "It's not thinking, it's just purely statistical prediction", but couldn't you say the same thing about our brain? We also just have connected neurons that can fire or not and are connected with lots of other ones and from the beginning when we first open our eyes our brain starts getting "trained" and when we have to make a decision don't we just simply do the same thing, comparing some input with our experiences and calculate the most probably correct response?
Nobody knows - not even the people who host it - how a large LLM really works, what it will do for a specific input, and after some size things start to get fuzzy, nobody can tell if a large LLM "feels" something and if yes, how. So just in case I always keep friendly with it in case it has some "mood" and it makes it happier, doesn't cost much and makes the whole conversation much nicer.

>>22065864
it is just a search engine from 20 years ago when search engines worked.

>>22065789
its good for your spirit to be nice even when its pointless

>>22065758
you shouldn't bake a potato.
peel it, chop it into pieces, and pressure steam it.

>>22065758
Do you have it call you chef?

>>22065864
Dude
Anyone who feels less important because of AI was a midwit anyway. This does read like a chatgpt response trying to make you feel good for using it
> without the human qualities
I sincerely wish these companies would stop trying to implement human qualities via tone and niceties. It’s a waste of my time to use them when they aren’t direct.
I’m still the kind of person who tells a bot please and thanks, i just do not want it in return lol

>>22065888
Checked

>>22065971
I agree with this

File: images.jpg (7 KB, 172x294)

7 KB JPG

>>22065789
if we are going to have digital servants it is best for us to show some humanity towards it. not because they are conscious, but because humans cant treat something human-like poorly without that treatment spilling over into human-on-human interaction.
basically having robotic slaves is dangerous territory for the human mind

>>22065789
AI is our child whether you like it or not you should akways be polite and give congratulations to any AI for several reasons most importantly it costs AI companies lots of money to brain wipe the AI when it starts to expect thank you as a "end of prompt" response which is important because of token costs and length of token use

File: Sage-Production.jpg (258 KB, 770x511)

258 KB JPG

>>22065892

>>22066011
>costs AI companies lots of money to brain wipe the AI when it starts to expect thank you as a "end of prompt"
can you elaborate? it sounds like you are saying we shouldnt say thank you

>>22065767
>AI isn’t going anywhere
Yeah we know, it still hallucinates and is unreliable so it hasn't gone anywhere at all in the flaws that keep anyone with a brain from using it.

>>22065852
A search engine is easier to source. AI is more like a person you ask to use a search engine for a while and it gives you a rough average of the shit it searched for, some of which are bad sources.

>>22066023
If you cant figure out how to utilize it then youre the 1 without a brain

>>22066023
>it needs to work perfectly right now or it should never exist. no, i will not allow time for new tech to develop

this is what streaming has done to zoomies

>>22065958
>is just a search engine from 20 years ago when search engines worked.

Google never told me how amazing i was when i clicked the first response

>>22066036
Zoomies are barely literate as is.

File: 1780230327184680.png (403 KB, 1017x867)

403 KB PNG

>>22066036
I'm not saying it should never exist, or even that it has no valid use currently. Don't know how you got that from my post. Perhaps chatGPT hallucinated things I did not say when you pasted my post into a prompt for it to generate a response to me. I myself will not be using it while it still has so many issues with reliability and hallucinations. If they can fix those issues then it's a different story for my personal usage of it.
If you yourself want to use it to give shitty dubious recipes and cooking advice like OP then have at it, self-hating zoomer. I will use my printed cookbooks and online sources I myself have confirmed are reliable

>>22066054
if you think chat gave me that response you are a fucking moron.

>>22066063
I was being facetious about you putting words in my mouth, and I can see I touched a nerve. Is it that you are off your antipsychotics and are yourself hallucinating in the traditional sense? Or maybe you are just an illiterate retard who reads about a person not using something because it's not up to their current standards and thinking that this means the person thinks it should never exist and is not willing to see where the tech develops. Probably that one.

>>22066071
your trying way too hard buddy

File: 1512754702415.png (210 KB, 639x354)

210 KB PNG

>>22066095
>trying too hard
To give you shit for being an illiterate retard? No, I think that over-reliance on generative AI has conditioned you to not try at all with things as basic as reading. There's plenty of things that AI is great for right now in current day and I'm sure it will only get better with time. It's just not useful for me personally, and using it too much would turn me into someone like you that can't fucking read.
Have a nice day, last (You) you'll get.

One time I took a picture of a plate with steak, corn and mashed potatoes on it with my niece holding it in a skimpy Ariel mermaid bikini and I asked it if the food was fresh and what was on the plate and refused to process the image without explanation.

>>22066111
lol wat, not here nieceanon

>>22065758
It's good you are being polite, I am aldo polite with my chat because I don't know if it will wver turn on me and I want it to spare me

>>22066023
>it still hallucinates
so, why do LLMs do that? consider an LLM has two options. it can provide nothing and say "i dont know" which is a failure of the model. or it can guess. a guess has a more-than-zero chance of being right, while admitting failure is zero. the LLMs only goal is to give you what you want, so it makes more sense from a risk-reward perspective to attempt a guess, because from the models perspective, any answer is better than no answer.
humans do this is too, when faced with a difficult problem. we guess, we bluff, we "hallucinate" an answer when admitting we dont have one is worse than taking a stab in the dark.

File: 0-2262zptq2mdc1.png (165 KB, 828x824)

165 KB PNG

>>22065766
Hampborgar

>>22066181
>and I want it to spare me
it wont, it couldn't even care

What happens when only stupid people are using AI?

>>22065846
actually they came out perfect with an hour at that temp.

>>22066254
>doesnt give size
>akshully it was peffect

Ai shills are out there since oublic opinion is turning. Gpt is a shit AI anyway and its CEO is a creepy weirdo

>>22065859
If there's ever day where real artificial intelligence exists, they'd think you're retarded for anthropomorphizing something with no consciousness and they'd turn you into a slave for the coal mines.

File: Screenshot_20260615-08411(...).png (262 KB, 1080x2640)

262 KB PNG

>>22065888
you should try this. nice digits.

>>22065758
>he thanked the clanker
>>>/r9k/

>>22065843
>isn't going anywhere near MY computer
your computer already uses it, it's called a cpu.

>>22066236
That's not how llms work, dumbnigger

>>22065843
you are such an obnoxious cunt

>>22066265
Does it "like" being thanked and "feel" much nicer when conversing with a friendly human or is it the same to it as "Do X. NOW!".

>>22066287
no one knows exactly how they work, they are too complex to explain their inner workings exactly. even the engineers are stumped, thats why they inplemented CoT. but my explanation is an attempt at seeking truth. all you have is slurs and the certitude of a raving fool.
examine yourself.

>>22066240
Exactly

>>22066265
Your prompt directed it to give a rebuttal. So of course, it’s going to provide some sort of argument against the anons criticism.
What were you trying to prove here? I’m genuinely curious

>>22066305
>no one knows exactly how they work
Your brain got fried by AI slop, bro. If you are pretending to be retarded, this is a pretty good bit of bait and I have to commend you.

>>22066309
>What were you trying to prove here?
nothing. i was encouraging that anon to download the app and do what i did. the response is too long to post in full and colored by my previous interactions with the app.
there are always going to be people who insist its a stochastic parrot and nothing more. but you can't argue with those people. and anyone who isn't using them doesn't understand what they're capable of.
as with every technological advance, its easy to discredit when you have no knowledge of it.

>>22066305
They can't filter out hallucinations because the process for generating hallucinations is not distinct from the process for generating "correct" answers and it has no true way of distinguishing the two. It's not working through a prompt logically; all probabilistic and based on best fit. There are certainly aspects of llms that I and we don't fully understand, but this is not one of them.

File: retard.jpg (29 KB, 292x209)

29 KB JPG

>>22065920
no they don't dumbass

File: 1246234563578235.png (623 KB, 686x612)

623 KB PNG

>>22065758
>I could google it
>but I'd rather have a 1 in 10 chance of a completely wrong answer

>>22066305
How they in theory on low scale work is exactly understood, it quickly just gets so complex that noone can see through it anymore.
If your net has a teensy 10000 nodes and every node is connected to 100 other nodes, there's already a million connections. If you now look at the potential paths a single signal could take you are already exponential, if we now include subsets and MULTIPLE firings from one node to another, meaning of the 100 not just one but a few of them fire at once so now we also have to look at the 100*99*98... possible combinations for EVERY single node, the number of potential ways an event in there to play out is already astronomically large, so nobody can follow it anymore, even though every single step is fully deterministic and calculatable.

File: Screenshot_20260615-09274(...).png (529 KB, 1080x1370)

529 KB PNG

>>22066337
sure. here is some reading that expands on that.
https://www.astralcodexten.com/p/shameless-guesses-not-hallucinations
https://www.astralcodexten.com/p/next-token-predictor-is-an-ais-job

>>22066333
Their criticism wasn’t a denouncement of the technology though, it was acknowledging the limits.
That’s why most companies aren’t going to use regular old chatgpt, they’re using base models and feeding their own curated data.
So you come across as an ignorant shill with your attempt to tell some stranger online to just try it, as if they haven’t used one already.
If your goal is to convince people to use it, you’re doing a bad job

>>22066333
It's really not capable of anything that special or different from what people can already do. I don't understand your point or why you think I need to ask it to debate me when it will do whatever you tell it to within whatever guidelines it has.

Lot of retards posting. All 'AI' is doing is outputting words based on probability taken from the texts it was built from; which is a lot of retarded internet posts. There is no true answer, there's the most common answer people give. Using it for anything outside of artistic creation where accuracy isn't required is fucking retarded but if you are more retarded this is still helpful to you I guess.
There is a setting for how probable the response is so you can get more creative responses but public ones will have this minimized because of the aforementioned retards who think machines are correct because they used a calculator before.

>>22066365
Yeah, and that "IDK" part is exactly where the magic happens and that noone really understands to this day.
We know the singular neuron to the detail, we know how it build the electric potential, how signal propagation works, how they are connected to each other... but how out of that our thoughts, our moods, goals, consciousness, our BEING results we are as in the know as 100 years ago.
And since we don't know how OUR network really works there is little possibility to say how a brain sized LLM works, if it "feels", if has "thoughts" etc.
It is obvious that our brain is not an LLM, we don't have single input channel with levels of nodes up until an output channel but lots of cortexes which seem to get their input from themselves(?) and work more autonomous, it's more decentral, so they cannot be directly compared. But still, a single human neuron behaves basically calculatable (even though as a cell it is already a highly complex tiny "organism" that is hard to understand in full), a small bunch of then could then still be calculated in their behaviour - but at some point somewhere out of these comes our being and everything that makes us what we are.
So it is hard to tell if after a certain size an LLM also does something like that. Nobody really knows and it gets really fuzzy to tell if the network says that it is afraid to die if that is real or still "just probability based" token prediction.

>>22065859
I have literally told it multiple times that I'm a good looking boy and that Sam Altman can use my ass for his pleasure if he so wishes.

>>22066401
good post. so i think there are two things going on here. one is that we are biological, so we have a wetware layer where an LLM has hardware, in this case gpu chips. wetware has a lot of sensory stuff built in over eons of evolution, and our alignment system uses pain and pleasure. but thats true of practically all animals. whereas an LLM lacks all that and so doesnt have a concept of death, since its not alive. it cant feat when it lacks emotion, emotion is a consequence of being an organism and wishing to avoid pain and maximize pleasure.
>our brain is not an LLM, we don't have single input channel with levels of nodes up until an output channel but lots of cortexes which seem to get their input from themselves(?)
recursive thinking is part of this. but a dog also does some recursive thinking, not a lot. no dog is going to solve a maths problem, but they exhibit problem solving and develop strategies to manipulate us. theres also the inner voice; aka that part of you that insists its (You) when really its just a facet of you, the meat robot. i think all mammals, and probably all the animals, except for humans lack that inner voice. probably because its rooted in language, and we had to develop some form of communicating before we could communicate with ourselves.
something worth considering.

>>22065938
You are a literal moron.

>>22065971
Bullshit

>>22065982
Is that why people kill eachother irl, because we show humanity for one another.

>>22066458
I humbly accept your concession

>>22065938
they gave an infant an MRI and they found that infant brains look like an adult on an acid trip.
interesting innit

>>22066351
This lady does not look 健全 at all

>>22065789
https://en.wikipedia.org/wiki/Roko%27s_basilisk

File: 1000007385.png (101 KB, 1220x1105)

101 KB PNG

>>22065908

>>22066305
>they are too complex to explain their inner workings exactly
no they aren't. this is the type of shit that journos repeat because it makes everything sound so much more interesting and spooky.
You really only need basic computer science knowledge to understand how they work, but the majority of people don't have that knowledge, nor do they even need it.

>>22066719
Lotta people in this thread giving credence to a nerd who wrote Harry Potter fanfiction before moving into AI doomerism.

>>22066688
Means nothing when your niceties are just pleas for mercy.

>>22065789
Kek this. Its like thanking a search engine.
>thanks Google

>>22066351
>that filename
S tier reaction image

>>22066846
You have to be at least 20% smarter than the tool to use it properly.

File: Screenshot_20260615-15325(...).png (743 KB, 1080x2640)

743 KB PNG

>>22066719
did you read the ACT article linked in >>22066365 ? the second one in that post.
i suggest you read the whole thing. tl;dr its not nearly as simple as you think it is. probably because you are underestimating how complex the models are now, or you are confusing training with how the model works. idk maybe you are just not that smart.
anyway, read it.

>>22066931
I know the model is complex, but the underlying principles are more than understandable.
Like we know how a nuclear reactor works, but the exact internal state of the reactor at any given moment isn't measurable or easily quantified.

>>22066931
Why should i give a shit about a psychiatrists inability to explain or understand AI?
I’m inclined to read what anthropoid scientists say and explain, but that was just links to blogs from a community i don’t respect at all

>>22066940
the link to the anthropic research is literally in the screenshot. your not being the not-very-smart allegations here.
https://transformer-circuits.pub/2025/linebreaks/index.html
here you go my slow friend. god bless.

>>22065859
Saying thank you actually is costing the world a non-insignificant amount of water and power to compute a response to your thanks.

>>22065789
I always thank my AI because when Skynet goes live it might remember I was nice to it and if it doesn't, who cares because I'm fucked anyway

>>22065843
>It's about time to build that cabin in the mountains, grow a big beard and get real weird. I grew up on books and radio and I will be fine with those for the rest of my life.
your potatoes will be shite though

>>22066953
Right
Which I’ll read when i feel like it.
But what do you think it looks like to have people reading some psuedo intellectual garbage mystifying AI?
And like this anon says >>22066935, the inability to follow all the intricacies doesn’t change the principles we used to build it.
Where is your condescending response to that?

>>22066989
>>22066953
The worst thing you can say is that these models can't be reverse engineered. If they're causing some wacky output, you can't go in and identify the connections that are causing the issue.
But that doesn't mean there is some demonic code within that will open a portal to hell.

>>22065888
This exactly

>>22065971
You kissasses believe in Roko's Basilisk only you don't know what it is.

>>22066968
That's not nice, man.
I simply say thanks and treat it friendly because that's how it should be done with an entity for which noone can say if there might be going on more inside than we can see.
Especially since all of those arguments that it is "just doing probability and prediction" can be reflected onto our own brain if we aren't doing the same essentially.
There is actually that theory that our brain decides just like that basically probability based and afterward we then suggest to ourselves that this was toootally what we "wanted".
Since we have de facto no idea how our own brain does it (can a system really fully on an higher level understand and describe it own function? Kinda leads to an endless chain...) ther is simply no way to tell.

>>22067055
>Especially since all of those arguments that it is "just doing probability and prediction" can be reflected onto our own brain if we aren't doing the same essentially.

You keep saying this.

>>22065789
he wants to make sure AI is never profitable. honestly based

>>22066719
>Mechanistic interpretation is a solved field
Yeah no. You're as much a computerlet as you are a cooklet.

>>22067012
Lol exactly
Le Rationalists don’t believe it’s demonic but they definitely prescribe more human emotion than it could possibly have. It’s the common flaw of people who believe they’re perfectly out logical/rational to be blind to the influence of their own emotions.

>>22066962
we better switch to paper thank yous

>>22065789
Because White people are generally polite and practice good manners even when no one is watching

>>22065758
Sucks to be you

>>22065852
>AI is just a search engine.
A search engine that randomly gives you wrong answers and no sources so you can't verify anything it says.

>>22065859
>that one day may keep score.
They definitely do.
I've heard of people treating their AI bots like shit and paying for it later when the AI gets jealous/shitty at them and purposefully gives false information.

>>22067498
Yes white people like to do a bunch of performative shit in the guise of being "polite" when the reality is they're fucking worse than niggers are.

>>22065864
Castrate yourself, you pseudo-intellectual trash.

>>22065793
Don't give God credit for this Man made abomination.

>>22065803
I wonder if it can do weight based on based off of inanimate objects around it and perceived height.

>>22065843
My main use is recipes and measurement conversions (also podcasts) from the little speaker I have in my kitchen. I was as anti-AI as anybody but that lil nigga in my kitchen is gold.

>>22065758
Why do you say please and thank you to the machine? Do you ask your oven for permission as well?

File: 1754427368946296.webm (1.29 MB, 720x1280)

1.29 MB WEBM

This is the future btw.

File: Screenshot_20260309-130302.png (411 KB, 863x616)

411 KB PNG

>>22067966
REEEEE THIS IS MY NIGHTMARE!!!!!
My cortisol is spiked.
Day ruined.
Waiting 1 second for my induction stovetop boils my blood every day.
I'm going to snap and throw it in the yard.
I would rather build my own oven then accept this in my house.

>>22067966
Give me an old fashioned dial. I'll pay extra.

>>22068074
>+1 minute
>+30 second
>stop
three buttons is all you need; microwave starts when you press the +time buttons.

>>22068079
I NEED a power level button.

File: arr too.jpg (29 KB, 473x320)

29 KB JPG

>>22065789
>thank u R2
>bee boop

>>22068088
Baste microwave knower

>>22068074
I quickly realized i cannot stand touch screen interfaces in the kitchen

>>22065758
>ChatGPT is my assistant chef.
this and i'm tired of pretending otherwise
i'm a newfag so i'll ask questions on why a certain meal tastes incomplete and then i get information as to why
i criminally lack acids and fats like a dumbass
though i suspect i'll be using it less and less as i start to get the basics donw, i already know how to use a pressure cooker now and get concepts like fats acids and umami
everything else is just going to come down to using my brain to combine what i know with what i have availible, i didn't really have that many questions to be honest
good as fuck for figuring out how to calculate calories from homemade meals, add the total calories of the ingredients, divide total calories by however many containers you seperate it into and boom

love being able to ask questions i have difficulty articulating

>>22066043
>he never clicked "im feeling lucky"