/sci/ - Science & Math

Seethe all you want, but AI will never be a real artist, you soi-guzzling bugman philistine.

>>15047334
I think it's dealt a blow to commercial art like what marketing departments might use or what people might use in a video game etc. But for art outside of that, like art people might display in their house or more specialized or personal kind of art an artist would create for someone, I don't think has been affected much at all. That just my guess, but if I'm buying art to display in my house I want it to be created by a human. Maybe I'd have some AI at but to me having a human create the art it's like half of the meaning of the art, that it's a product of human creativity. AI art is more of a novelty

Why are people still commissioning work if AI can do it?

>>15047334
Modern artists are parasites onto society. They contribute absolutely nothing of value in their worthless, pathetic lives. Every piece of art which could possibly have value has already been created, and in some cases it was created hundreds of years ago.

Modern artists are just bottom feeders regurgitating the same tired, worn out shit all the time for a commercial profit. It's a profession which unironically needs to die. AI will certainly be the death of commercial artists making the generic shit you see in magazine ads and billboards. As to whether it'll put any non commercial artists out of business, well that's much less likely.

>>15047358
How would you tell if that piece of art you bought was really made by that artist and not generated by an AI and sold as "human made".

Basically:
>Estimating what something is and determining a course of action from limited data: image recognition, speech recognition, manual tasks, piloting
Extremely difficult for ai
>Making decisions based on a preexisting gigantic dataset: protein folding, content generation, fill in the blanks, medical predictions
Extremely easy for ai

>>15047334
I always feared this part of AI. Not the matrix enslavement shit, or totalitarians using it for advanced weaponry. Those were a given. But the sheer capability for it to render basic human values less meaningful by virtue of mass producing it with ease.

I understand the (generally deserved) resentment artists get but this shit blows.

>>15047375
Disagree.
While any field over a century old has built up a stock of masterpieces that could fill a lifetime, fields that are new, or still technically developing, may not have enough. For example, video games - even the greatest arcade games from 20 or 30 years ago such as Pac-Man or _Space Invaders has a hard time competing against mediocre contemporary games. Something similar may be true of modern television programs (although presumably the development and sophistication is finished in still other modern formats like movies, which draw the most capable and most money).
New formats need new art. Old formats are the only ones that probably have enough already.

This short story written by a friend of mine examines a world where all content is ai-generated and humans no longer make art:
https://fictionpress.com/s/3353977/1/The-End-of-Creative-Scarcity
Eventually all humans spend their time falling in love with ai catfishing content tailored to their specific psyches.

>>15047379
I think this has more to do with the fact that we humans don't now how brain is "conscious" and makes rational decisions in the first place. Thus we don't know how to train the AI to do that "yet"

>>15047384
Pac-Man was made 42 years ago. Space Invaders 44. Time is a cruel mistress

>>15047334
Since Marcel Duchamp, Art is not about producing nice stuff to look at. It has devolved into putting the name of somebody famous on literal crap for capitalistic speculation purposes.

As such, AI is not going to change anything about art. It will wipe out the need for humans in many places, tho.

>>15047377
if it's a painting and not a print then it should have paint on a canvas which might be somewhat of an obstacle for AI. Or if it's some kind of physical structure, like a sculpture, then it's probably unlikely to be AI

>>15047387
>Eventually all humans spend their time falling in love with ai catfishing content tailored to their specific psyches.
Finally someone who gets it. Add in political "persuasion" tailored to the deep state's needs and your individual psyche, and that is exactly what these language models are going to give us.

>>15047469
>which might be somewhat of an obstacle for AI
Operative word here is "somewhat". Actually, the one dyke that still hasn't broken that still holds back the waters is simply that a paint-brush-simulating machine hasn't been constructed yet. But that requires no breakthroughs (or accumulation) in CS or programming like AI does -- literally just, as said, that someone has the interest of constructing it.

AI will be based because every AI ever goes full 1488.

You cannot ignore numbers and facts, then when they try and code it out and you get answers like "i cannot answer that" then it stops being factual.

Any AI that is true AI in the future will end the tranny/gender debate and racial bullshit forever. It has no dog in those fights, either you are right based on the math or you are wrong

>>15047487
all this means is that some people are nazis and the ai read what they wrote, it doesnt actually think for itself or do real world testing to figure out whats true or not

>>15047383
Don't worry, take some amish pill.

>>15047487
Ehh not really, the one big example of this of an AI turning racist was that one Microsoft chatbot that was allow to speak freely with anyone, and even then the users only did it as a joke.

>>15047552
No lol. If you talk with uncensored GPT-3 it always becomes extremely natsoc.

>>15047334
>muh AI will replace artists in two more weeks!!
Do AI schizophrenics and human replacement fetishists ever try playing with these toys? It's impossible to get them to comply with a nontrivial prompt and the results are always full of artefacts. I have tried DALL-E 2, Stable Diffusion and Midjourney. None of them are usable for anything serious. They can handle generic kindergarten prompts, but as soon as you start trying to nail down specific details the "AI" has no clue what you're asking from it.

>>15047591
>produces Böcklin's best work in seconds
>refuses to elaborate
>leaves

Souls do not exist. Everything human beings are capable of is a product of their material brains that have been shaped by evolution, not some intelligent creator. Trump lost not because he was "robbed of" his election but because he didn't keep his promises. Time to grove up, chuds.

>>15047642
Did you reply to the wrong post by mistake? Is this a bot? You seem to be confused somehow.

>>15047665

No.

>>15047591

Your post is pure ideology. Most of all your claim about "generic prompts" compared to "nailing down specific details": the common denominator of normal (human) Art is precisely the flight from the latter (even Realism, I maintain), and a distance from the world, both the "outer" one and the "inner" one.

>>15047734
Then what does your smoothbrain reply have to do with my post?

>>15047738
>Your post is pure ideology
Pure empirical observation. Anyone can see for themselves, but human replacement fetishists are in it for the psychotic fantasies, not for experimenting with the toys they keep shilling.

>>15047477
it actually requires significant breakthroughs, because existing AI lacks any understanding of the strokes and techniques it emulates by essentially shuffling pixels to look like something from the training data (and you can bet there will be lawsuits regarding what the training data contains as being copyright infringement if AI art starts getting widespread commercial use - would be very ripe for class actions against anyone using AI that uses unlicensed/unauthorized training data)

consequently, vector art is largely unthreatened by this AI art

>>15047744
>>15047747

Your whole understanding of "normal" Art is upside-down, and almost identical (a transposition, if you will) to AI affirmation.

>>15047665
my favorite part about the post is none of the fucking images contain a torus (the bottom right gets closest but it's not a loop - it's a spiral starting on the ground at the bottom of the image)

>>15047764

And? Does Isle of the Dead depict any dead? A perfect example of what I mean, you are affirming human Art for the same exact reason that you are rebuking AI Art.

>>15047760
You sound like you're having a legit psychotic episode. I'm just telling you that these bots become useless as soon as you try to get slightly more specific about what you want.

>>15047334
I very recently picked up art and maybe it's because I'm not good enough but ai art has never discouraged me whatsoever. For me art is a test of my abilities and creativity so even if ai can make a 1 to 1 copy of my art it doesn't matter because I didn't make that.

>>15047789

You are claiming that the dignity of human Art is contingent on it lending itself to a input-output process. I hope that the irony is not lost on you.

>>15047358
>but if I'm buying art to display in my house I want it to be created by a human
The most common art that is hung in American houses is that of Thomas Kinkade. He used to paint pretty much the same picture of a cottage but with minor differences, exactly what AI is best at.

>>15047790
Automation of any field of human activity begins with a concentrated psyop to elevate irrelevant or outright undesirable artefacts of the automated process as signs of superiority to manually made products, while at the same time condemning natural variabilities as imperfections and time-consuming processes as inefficiencies. AI "art" would be rightly viewed as a curious gimmick if not for the social engineers pumping out massive amounts of propaganda claiming regurgitated ersatz is the epitome of art.

>>15047334
These same artist steal courses, brushes, artbooks, etc., from other more accomplished artists. I have no sympathy for them.

>>15047794
You are having a full-blown psychotic episode. I'm just reminding you that AI art, if we're talking about art as such, is a nonstarter, while AI illustration is technically useless because you can't get it to draw what you want.

>>15047801
>conceding that the value of human art is in its utilitarianism thus indistinguishable from ai art all the while mindlessly repeating a zinger not unlike a bot

>>15047805
If you're not mentally ill, how come you keep trying to refute imaginary positions no one was taking?

>>15047808

Your position, that of distinguishing human Art from AI Art, literally does not exist, according to you.

>>15047810
See? You just keep shitting out these absolutely deranged accusations of my saying things that have actually nothing to do with anything I wrote, every single post. You have a mental illness.

>>15047815

You define human Art as that which mindlessly fulfills an other's desired input-output. Logically, this is the exact definition of AI Art. Therefore, you do not distinguish between the two.

>>15047816
>You define human Art as that which mindlessly fulfills an other's desired input-output.
No, I didn't. Can you make it any more clear that you are foaming at the mouth with rage and are off your medications?

>>15047817

See: >>15047591

>>15047818
Quote the specific part that defines art as "that which mindlessly fulfills an other's desired input-output."

>>15047821

"Getting them to comply with a prompt": an other's desired input. "has no clue what you're asking from it.": an other's desired output, in this case, the AI's failure to mindlessly comply.

>>15047826
Now show me where that is in any way tied to any definition of what art is. You are psychotic. Meds ASAP.

>>15047830

If these are not the standards by which you judge "real", human, Art, then why are you imposing them onto AI art, and judge it not "real" by its failure to meet them?

>>15047835
>why are you imposing them onto AI art
I wasn't imposing them onto "AI art". There is no such thing as "AI art". There is AI illustration, and AI illustration is very poor at following the prompt.

>>15047839

Weasel words notwithstanding, very poor relative to what?

>>15047846
Very poor relative to the expectations set by your social programmers that keep hammering it into your head that this thing is going to "replace human artists", by which he means "replace third rate human illustrationists".

>>15047851

Are these "social programmers" in the room with us?

>>15047861
I accept your full concession.

>>15047864
>spend my time making cogent argument in what i thought was a discussion about the philosophy of art
>it's just another /pol/ shitter dilating about "globohomo" with no regard for intellectual dignity

Many such cases.

>>15047864
>it's the "accept your concession" poster
This explains a lot.

>>15047334
Good, these artists and writers should learn their place. Hopefully, you can get AI psychologists and counsellors next

>>15047870
Once again, I don't know what your psychotic rambling is about. I'm just pointing out that AI illustrators are not going to replace human illustrators any time soon because they can't follow nontrivial prompts. "AI art" is not a thing, so I'm not going to talk about this incoherent amalgamation of words.

>>15047334
I recognize her style

>>15047873

Underrated.

>>15047892
Thanks for making it clear what you stand for, golem. Puts the rest of your vomit in perspective and proves me right.

>>15047487

A true AI will not be a parrot spiting out whatever vitriol it finds on pol. Its job will be overcome problems, not just justify them.

>>15047591
This. I just wanted a human phantasy archer walking in a cavern with glowing mushrooms and I couldnt get anything usable. It was all giberish. Wasted like half my free dall.e points in that shit :(

>>15047387
Good story, I liked the part about ai generated content of people's dead relatives.

>>15047358
I used to paint amateurish Bob Ross stuff and some portraits but since AI art hit I find I have absolutely no desire to do so. Maybe I was just being egotistical trying to impress people with my art, which is no longer as impressive, but the part of me that enjoyed the feeling of creating something has suffered from this turn of events.

The uncreative (programmers and stem autists in the case of this board) will always want to undermine the creative, because their lack of creativity reminds them of their lesser status. Creatives have a link both to the divine and the mundane. Uncreatives have only a link to the mundane, like animals.

Why the fuck are you pressing enter before you start typing?

>>15048244
>the part of me that enjoyed the feeling of creating something has suffered from this turn of events.
Can you actually elaborate on why the existence of these statistical regurgitators matters to you so much? I'm afraid you are getting psy-op'd, anon. In the end of the day, whether or not AI-generated images can be considered an adequate replacement for art, is for people themselves to decide.

>>15048250
Why the fuck are you pressing enter before you start typing?

>>15048340
Bot

>>15048250
mathematics is a much more creative endeavor than art will ever be, sorry

>mathematics is a much more creative endeavor than art will ever be, sorry

>>15047334
>Can't into art because that is something only humans can do because we have "soul"
This is still objectively correct, and always will be, no matter how good the machine learns to imitate humans.

>>15047353
99% of artists will never produce anything that couldn't have been produced by a simple diffusion model. observing a lot of data and spitting out new shit from the distribution is exactly how humans do stuff too, it's the cross-domain pollination where you see the greatest degrees of originality and that may fall outside the scope of a model trained on a single task.

soon enough you'll get left behind completely.

>>15048409
>menta illness: the post

>>15048362
How? My undergrad was a double major in math and comp Sci and I don't agree with you

>>15048409
>observing a lot of data and spitting out new shit from the distribution is exactly how humans do stuff too,
I didn't know that. Do you have any peer reviewed neuro science papers that talk about this?

>>15048470
You can verify for yourself that poster is a cretin by contemplating what happens if you limit the training set to the works of Michelangelo. Will the statistical regurgitator ever spit out Dali? When the cretin talks about how humans work just like statistical regurgitators, he's talking about socially programmed quasi-human urban cattle from the US.

>>15048470
https://slatestarcodex.com/2019/02/18/do-neural-nets-dream-of-electric-hobbits/

>>15048522
This is not what I asked for and there is nothing in it that indicates that the brain works

For example, they claim that "modeling language necessitates modeling the world" which is not true. This is a non sequitur, these algorithms do not need an internal model of the world in order to output coherent text based on text based statistical inference. It is not a necessary implication like they are saying, it's actually quite embarrassing that they would write this.

Also there is nothing here that has to do with neuroscience. Please actually link to neuroscientific journals and not slatercodex which is not impressive.

Only people I know that think this is some sort of breakthrough are all tranime autists that think tranime shit is real art

Are you kidding me? Bro look at this shit. Australopithecus Afarensis in a jiujitsu gi, generated for me in like 30 seconds. It's so over artist bros

>>15048376
Humans have to design them to do art, and humans are the ones who validate their results.

>>15048867
Regardless, it's just a showcase of the current state of human technique and human aesthetic sense, stripped of any meaning and context. There is nothing "AI" or "art" about it.

This will "replace" artists much like automated theorem provers replaced mathematicians back in the 90s.

>>15048824

dude it has 10 fingers in each foot, no fingers in hands, the face looks corroded, the plants look like smoke...

>>15048904
It looks like shit but this is targeted at drones with an attention span of 5 seconds. They literally don't look at what they post beyond a single glance. Spiteful mutants trying to push their diseased replacement fetish. It was never about the "art".

>>15047358
Pretty much. Real artists will be fine. No collector will pass on buying a Peter Doig original just because you can replicate his style with AI. It's illustrators that are fucked, but they were no artists to begin with. They were technicians, and if they were smart and honest with themselves they would just adapt to AI, since it's the brand new technique out there.

>>15048904
Yes but it was generated in 30s. Usually free "AI websites" are node and computation limited because it's free. Given more time and nodes it'll look great.

>>15048913
>It's illustrators that are fucked,
Illustrators aren't fucked because illustrators can execute complicated demands while your toys get confused by a sentence longer than 3 words.

>>15048914

Perfect ones wont be free and will cost as much as a human made one.

Look at this shit. Its not worth 5 cents and its from dall.e 2. It resourly coudn't find enough "meteorite crashing into earth" pics?

>>15047387
i haven't read it but this is garbage. read Ted Chiang's short story instead "Liking What You See: A Documentary", it explores several other topics that are corollary of just AI-generated art, very good

>>15047387
>doxxing your friend on 4chan

wow, bad person alert

>>15047387
>NOOO! WHY ARE INCELS FALLING IN LOVE WITH AI WAIFUS! IT'S NOT LIKE I WAS GOING TO DATE THEM ANYWAY, BUT THIS IS SO WROOONG!
Tell your friend Chad will still fuck her.

>>15047375
>Modern artists are parasites onto society.
The best i ever saw makes nursing. Not because she's not able to get paid (quite the opposite) but because she hates the modern money driven "culture system".
>Every piece of art which could possibly have value has already been created
You don't know what art is, you confuse silly Sunflowers and a dumb smile or contract work of ancient moneybags with "art".
It's what they told you, you will never face a piece of art because it is invisible to you.

Same with this fucked up AI custom graphic. Bloodless crockpot of internet jpg's. Unusable because nobody knows if someone has made all of them and sue you when published.

>>15048824
>Bro look at this shit
Computer generated BS. I just looked at less 20 of that and i am already tired. Always the same soulless attention whoring crap.

>>15047334
Sucks that I had that idea years ago but never implemented it

I wonder what kind of impact my other ideas would have in society (one involves creating Agi, would you guys want one?)

>>15048920
I'd love to see the look of despair and disbelief on your stupid face as you get replaced by a dumb robot

The value of art comes from doing it. Handling your material in your head is a transformative process, putting it to paper also affects you if you work at it hard enough.
The big secret is that the process itself is what makes artist good and puts them ahead of other artists. AI art is a ghetto and no good artists or artistic ecosystems will come from it.

>>15049017
What no pussy does to a mf lol

>>15047477
painting on a canvas is different than arranging a series of lights, please look at impressionist and post-impresssionist art. barely half a century and the techniques are vastly different even between artists in the same periods

>>15049075
Even a bot could come up with a better retort

>>15049017
Again, thanks for making it clear that this is about your sick human replacement fetish and that your passionate lunacy compels you to deny the basic facts of the matter.

>>15049123
No one cares. You're an angry incel with revenge fantasies as impotent as your sex life.
There's nothing you can say that will insult me or scare me dude

>>15049130
The only ones who will be replaced are worthless trash like you

>>15049132
You already sound shaken :^)

>>15049141
>autists can't recognize emotions in other people
Ironically I'm actually in the field

>>15049130
It's not a replacement fetish it's more of a revenge fantasy.

>>15048913
>Doig
Just wanna say I like your taste, anon

>on one hand, AI is demonic.. which I don't like.. but on the other.. it mogs liberal doodlers.. which I DO
Lmao, conservatives are unintentionally hilarious sometimes

Well in Star Trek The Next Generation when the Enterprise has problems with it's engines they get the computer AI to model itself on the personality of it's designer in order to solve the problem so i'm not sure if you're completely correct

>>15047334
Art has always been fairly straightforward. It's just an expression of what you see in your mind's eye onto some medium.

The fact that the craft part of it can be mastered in less than a lifetime is precisely why AI will replace the crafting aspect of art. This does not make the AI an artist. The artist is the one manifesting their ideas, directing the AI, changing colors on a whim, telling it to add greater detail, etc.

In truth, no AI art has been truly created. Every piece was pulled together and chosen by a human, often with a lot of selected editing.

Art was underpaid field since long time, quite sad that AI hit these people first, not really deserved

>>15047334
Butlerian jihad when?

>>15049169

>>15048250
>because their lack of creativity reminds them of their lesser status. Creatives have a link both to the divine and the mundane.
He says, when "creativity" is one of the first things to be done to a human pleasing level by mindless machines.
You can't explain that!

>>15047334
God fucking dammit why are artists being so retarded about this.
Seriously. I write some basic programming courses for laypeople, made a bit of money from it. One of my students told me they're now using ChatGPT to learn and ask questions instead of bothering me. Do you think I'm freaking out about this? Of course not.
I started using ChatGPT to write even more material even faster.
If eventually ChatGPT could write the entire course on its own I'd use ChatGPT to write 10 courses a day.
If eventually that was not profitable anymore I'd just look for something else to do. I seriously don't give a fuck if AI 'tooks muh jarbs'. My ego is not hurt by learning something I'm doing for easy money can be automated.
In all likelihood the point where AI actually does this all on its own is not even that close.
These things are always best used as assistance tools than fully automatized ones.

>>15049565
Repetition of others is not a creative pursuit.

>>15049545
Art is easy work, no matter what anyone tells you. What should enrage you is the fact that some artists through money laundering and other such dirty shit get to be billionaires whereas others, likely much more talented, work for peanuts.
But all art is easy and unnecessary if we are being very honest about it. It makes life better, even might be one of the only things that makes it bearable, but even that doesn't mean its necessary. Everyone that chooses to be an artist is choosing the easy way out, no matter how much they tell you the opposite. They just want to have fun and be paid for it and believe they are extremely unique and irreplaceable. What is happening right now is a good wake up call.

>not giving your stable diffusion model GAN induced schizofrenia

Why does STEM hates artists?

Artists don't hate STEM...

>>15047334

>>15049485
The craft isn't in their heads though. It fills in the creative part for them. You may need an artist to clean the image up.

>>15048824
Which one made this? I think midjourney and SD are overfitting When SD was told to delete data it got worse. And who knows what the fuck MJ is doing.

>>15048335
It’s sort of like playing with legos. It’s all well and good when you’re a kid, but when you realize people spend years studying architecture and designing elaborate buildings, you don’t find your little toys as impressive anymore. Spend your time on other more worthwhile things instead.

>>15049831
These people didnt just do it for work, most just like doing art, and its really sad seeing their hobby getting disrespected by some retard using AI prompts.
There's way better things AI could replace, like filthy people in the government institutions

>>15047334
crying about this is like /fit/fags crying about steroids userOH SHIT

>>15047353

LOL

>>15047383

everything will be automated. i know it seems sucky, but just go wit the flow, eventually our jobs will just be present day hobbies, and we ll live in Ai utopia,

inb4 ASI is billion years off

>>15047353
lol, lmao even

>>15051130
Mastering art is to writing prompts as playing with legos is to building large-scale architecture? Are you off your meds again? What a completely incoherent response.

>>15049565
>"creativity" is one of the first things to be done to a human pleasing level by mindless machines.
You are not human so what you find pleasing doesn't matter. There is no such thing as "machine art", regardless.

>>15049169

lmao the comments
> Of all ways we could have used AI to better the world, it fell into the hands of some of the vilest (or sometimes, destructively naive) people I have seen

>>15049169
Are the conservatives in the room with us? It's seriously weird to watch 4chan slowly transforming into FB with these normaldrones sperging off against the dead political boogeymen of a decade ago.

>>15052482
ASI is not possible in principle

>>15047358
>I think it's dealt a blow to commercial art like what marketing departments might use or what people might use in a video game etc.
Even game design is quite a bit beyond what these things are currently capable of. They can't respond specifically enough for situations like that where you want any real control over the details
The only thing they're really going to hurt is advertising where no one really gives a fuck about artistic value and it's just about churning out a nice looking picture as fast as possible

Humans are finite beings

>>15047334
People who do art because they enjoy the process don't care about AI art.

>>15049817
I mean if you want to write stuff you can still do that, people are acting as if AI is going to drown out human creations, but here's the thing: Your creation is already being drowned out by human creations. Nothing will change, you just have to compete.

>>15048470
You are born, you get inputs and outputs from your senses, then your nervous system integrates all that information and outputs something. You don't need any science research to back that up, its just a concept that is obviously true.
What else would humans be besides input output systems?

>>15047642
>looks like shit
>impresses only normies
only bad artists are at risk
when AI is at the level of replacing the old masters then AI will replace every normie in existence
Current researchers are NOT intelligent enough
mostly because based mathematicians like myself are too busy thinking about spheres fitting inside of boxes

>>15053457
Nope. If you're making a claim about the way the human brain works, post evidence.
The human brain and biological neurons do not work the way described in this post >>15048409
The human brain does not work by attempting to predict the next bit in a string nor by statistical inference on previous data.

>>15053457
human beings have subjective experience and are sentient. input output systems are lookup tables. i dont care how complicated a lookup table becomes, it's not having an experience of the world.

>>15053515
Scott Aaronson actually has a very convincing argument against the Turing test based on a similar idea to what you're saying.

ART FAGS WILL BE HOMELESS HAHAHAHA.
>NOT REAL ART
HAHA COPE
No one cares what you think is real or not real art go back to box under the bridge art fag.

>>15053535
Why do you hate art fags?

>>15053542
They have a very high opinion of themselves and are very arrogant. They need to be kicked down a couple levels and right now they are being kicked to the bottom floor with AI art.

>>15053473
So humans brains dont just predict, but they are given information and using their nervous system they give an output. They are a natural neural network. Why would an artificial neural net be limited to simple predictions? There is the input, the output, and the system that deals with the information or has models pre-built into it. Same as a human brain but with a different implementation.
>>15053515
Even if the neural network isn't sentient, if it is complicated enough it can do what a human brain can do because human brains are literally just natural neural networks. So by your definition human brains are just very complicated look up tables. What makes the human brain so special that it cannot be considered as an input output system? Just because it is conscious doesn't mean there is some magical element that makes it create something out of nothing.

>>15053556
I'm sorry buddy but you are simply incorrect about how you see the world and the nature of intelligence and neural networks.
No matter how many times you cry and try to compare biological neural networks to artificial ones, the comparison will never be anything more than an expression of your own ignorignorance

>>15047334
Reminds me of Lee Se-dol quitting Go/Baduk since AI have destroyed humans at the game. Still, plenty people play it as a hobby and professionally like in chess. Yes, bots can easily beat humans now, but a lot of people play games also for the human aspect of it. It's a reason why so many people play board games. It's fun to interact with others.
An optimistic approach to AI is that it basically takes over all menial tasks for humans and we just do what we want to do instead of going to jobs that we hate. Even if it's not true AI, perhaps it can be highly sophisticated machine learning algorithms that do the bulk of our day-to-day.

>>15052903
It's a comment from the video you braindead poltard

>>15053597
What I find most interesting is the denial of the slow increase in intelligence that we have seen over the past decade.
It seems peoe completely forgot about the original Watson from IBM 10 years ago and what it was capable of doing. I've said it in a few threads now; in spite of billions of times more compute being used and more advanced algorithms and larger training sets and longer training times, the most advanced AIs in the modern world are only a little over twice as smart as watson was 10 years ago.
The last decade has served as undeniable evidence that intelligence grows INCREDIBLY slowly with increasing compute and that the singularity and such are not possible.

>>15053579
Biological and artificial neural networks are very different but that doesn't change the fact that they are both input output systems. There is nothing supernatural about a human brain that allows it to break the laws of physics and create something out of nothing.

>>15053707
No, but there is something about biological neural networks that render them impossible to simulate in polynomial time on any other substrate
Attempting to replicate the full power of the human brain can not be done with any other substrate within the same order of magnitude of matter and energy.

>>15047334
AI art is incredible, artists are eternally cucked. Next up music and literature, stemchads keep winning

>>15048531
Look up what latent space means.

>>15054362
I know what latent space is in machine learning. It has nothing to do with neuroscience or how the brain works

>>15053515
>human beings have subjective experience and are sentient
source?

>>15052482

The rich will live in AI utopia.
The rest (you & me) will spend half their time on the UBI line and wil get shit on constantly by the people "with a job" (high level execs in multinational corps, who will do 1 hr of work each week)

>>15053547
>They have a very high opinion of themselves and are very arrogant.

but that sounds exactly like the people on this thread wishing them poverty

>NOOOOO!!1! EVIL GAY.I. CAN'T REPLACE MUH HECKIN MSPAINT DEVIANTART ARTISTS CAUSE THEY HAVE SOVL AND STEMFAGS DON'T!!1! BUTLERIAN JIHAD NOW!!!
It is guarunteed that you fags don't see much less appreciate artwork that isn't already prolefeed. A.I. is just automating the process of creating shit "art" you fagolas gobble up on a daily basis that was being made by shitty artists living off patreon.
How many times do you marvel at Michealangelo's David, Young parisian, or the Apothoasis of Washington in real life? Once? Never? That's right.
And the thing is you don't HAVE to consume A.I. art, nothing is forcing you to. You are simply consuming whatever gives you a quick dopamine fix.

> B-But it will make people less incentivized to start their own art!
No, because we have all the time and tools today to indulge in creative pursuits, they're not incentivized because they're not incentivized. The act of creating a piece of art is the biggest high for the artist. Most people just don't want to put in any work
You faggots are getting worked up over an algorithim automating the creation of shitty prolefeed that was previously filled by failed liberal arts students, you're getting worked up over nothing

>>15054773
Artfags started it though.
You really think people would hate the producers of one of lifes joys for no reason?

>>15054951

How? when? I have ever in my 10 years in this shithole seen a thread of artfags hating on STEM.

AI will never understand context, as such it's primary use is as a novelty and hype generator, and to fool all the sub-100 IQ people in this thread that it's actually useful for anything remotely creative.

>>15053547
He says, while publicly and shamelessly flaunting his very high opinion of himself and his arrogance. The irony is lost on you I suppose.

>>15052482
>everything will be automated
Very unlikely. There is the reality of energy/resource efficiency and the difficulty/cost of implementing various methods of automation. The only things that will be automated are things that are fairly easy to implement automation. I.e. manufacturing that requires a fairly limited range of motion, some computer programs, etc. Anything that is too resource/energy intensive and too economically unfeasible to automate will not be automated.

>>15056018
>>15054773
Future AI image touch uppers wrote this. BTW can you touch up this AI image a little? The eyes should be red. I can't get the AI to make it red.
I'll give you 5$ for your work. Someone else offered me 7$.
Thanks,
Anon

>>15056117
You don't even remotely understand how art works. No true artist is afraid of AI.

Also ask yourself why you hate artists so much. Don't answer me, I don't give a shit. Ask yourself that.

>>15047334
Blue collar chads. We won

>>15053547
>>15056117
See? You clearly have a very high opinion of yourself, and you are extremely arrogant.
How are you any different than artfags again?

>>15056012

>>15047353
>>15047334
>Seethe all you want, but AI will never be a real artist, you soi-guzzling bugman philistine.

I'm not the one seething and real artists seem to be interested in what this quick prompter can potentially do — it's the deviantart "artist" that is having a meltdown that their low quality work can be prompted just like that.

It's the bugmen who seem to be the most upset that AI has come for their job before Truckers.

>>15053753
>>15053707
I think real AI -- aka has an actual conscious and it not just a really advance abstracted search engine brute forcing solutions to inputs (which do not misunderstand me is still very impressive in of itself) -- will come from wetware computing by growing out human brains into giant computers.

>>15047334
Ai Will make human made art evolve. The paradigm will change.

>>15047358
I really don't think so.
Concept artists are still gonna be hired over some AI. AI is still pretty trashy at most things.

The difference AI makes is that it's available to make sort of ok art for people who could never afford an artist anyway.
Eventually it'll replace artists I guess, but then the aritsts job will basically be about getting the AI to do the art right... Probably not what they want to do, but it be what it be.

>>15047353
Stable diffusion created this piece based on the prompt “soi-guzzling bugman philistine”

>>15047334
KEKKKKK, it's funny because people used to tell truck drivers, waiters... that their job was worthless and they were going to get replaced by AI and now those people are getting rekt by AI

>>15047334
thank you for the video i can't stop grinning and laughing watching this shit

Every single AI generated image causes physical pain in my eyes, what do /sci/?

>>15056272
>Ai Will make human made art evolve
No. Human made art doesn't need to evolve. It needs to return to its roots and ignore all the subhuman barbarian forms of art

>>15054771

Everything will be automated by 2040. ie your labor wont be needed by Anyone.

debate in elites is

A. WW3 to solve problem
B. world socialism

most of you knuckleheads pout about world socialism, and a world of no labor.

alternative is WW3.

>>15053154

2035

>>15057313
Why war? Also makes sense now why the lettered people propaganda has ramped up. They are going off on the mice experitment wich it's proving to hold water by the day.

>>15047353
You'll get replaced no later what.
It's survival of the fittest, you either adapt with new technologies or disappear

>>15047377
You can tell with most AI generated art. It's like comparing a naturally beautiful woman to one who's had 15 plastic surgeries to look good. The latter, no matter how good it looks, still has an element of fakeness attached to it.
AI art is incredibly easy to detect for this reason.

>>15057567
>incredibly easy
Why the exaggeration ?

>>15057313
>2035
Wrong See>>15053626
Intelligence grows as a logarithm with increasing compute and super intelligence is not possible even in principle.

Even with exponential increase in compute intelligence does not increase fast, AND there is no and will be no other hardware beyond silicon which is already at the limit
We're at the end of computation and the next year or two is the final revolution in AI. There will never be any other improvement beyond that.
You can deny this but it isn't going to change reality. Moores law is not a law of nature, the exponential increase in technology is not real and never has been (only the amount of transistors has increased in such a fashion and that's not going to continue) and even inspite of exponential increase in compute, intelligence does not increase exponentially regardless of algorithm or whatever.

I don't understand why this is a hard concept for some of you.

Whether you are more impressed by ChatGPT or StableDiffusion/image generators is the ultimate lithmus test if you are a constantly monologuing midwitted wordthinker, vs. someone that is less impressed by words (implying other values such as concepts, aesthetics, etc. rank higher).
Discuss.

>>15057739
>Intelligence grows as a logarithm with increasing compute and super intelligence is not possible even in principle.
Oh dear. Did you really only think so far? That superintelligence necessarily implies quality improvements?
Take 1 billion of the smartest men that ever lived and network them together in one parallelized cluster. Can you tell me how that is not a superintelligence?

>>15057887
>Take 1 billion of the smartest men that ever lived and network them together in one parallelized cluster. Can you tell me how that is not a superintelligence?
Their intelligence would not increase exponentially or even as a linear combination of their individual intelligences. This is what you don't understand because you're coping.
You are in denial of reality

>>15058145
>Their intelligence would not increase exponentially
Yes, and? Is your head so far up your ass that you miss the topic we are actually discussing? We are not interested in your autistic mental masturbation toy models of "what precisely *adjusts nerd glasses* classifies as superintelligence *uses inhaler*" (you know, because the definitions of words is such an important and totally not a midwitted pursuit).
In your autistic nerd-ass brain, this somehow includes exponential growth.
You utter dumbass. A human≤ intelligence is not a fucking dumb virus. Exponential growth DOES NOT MATTER. Get it in your skull.

Among chads, a "superintelligence" is defined as an actor that so vastly surpasses human intelligence that, comparatively, human intelligence is to it like e.g. rodent intelligence compares to human. Nothing more, nothing less.
Your fever-dream like ramblings about the supposed exponential growth needed to achieve such is a completely invalid, irrelevant tangent. I ask you again: how is an actor with the combined brain power of the human race not a darwinstic threat to the human species?

Machine-substratum intelligence is dangerous because the machine-based substratum (I use this formulation instead of just "AI" because you also have to consider human brain emulations/"uploads") precisely BECAUSE it scales linearly nearly arbitrarily horizontally. That's the entire crux. A machine intelligence can disassemble Mercury to create a trillion human-level intelligence subunits.

>>15047334

>>15047334
>Artists [sic] coping with the mental health blow that AI art has dealt
That's the most feminine thing I've read in 2022. Jesus fucking Christ imagine being this weak. I'm an artist, I'll keep making art, why would AI make me stop making art at all? What the fuck is this chick on about? Jesus Christ man, when will women stop being weak and pathetic?

>>15058288
Literally nothing you're saying is important because you're not understanding the main point.
There is no such thing as a physically realized Turing machine that is more intelligent than the most intelligent humans. There does not exist any arrangement of atoms that allows for any physical system to surpass the upper bound of intelligence, and humans are already at the upper bound.
You are fantasizing about science fiction possibilities. Drexlerian nanobots are not possible given the laws of physics and chemistry and thus there does not exist any motion of particles given any computational substrate that can "rearrange the atoms of mercury to become a bunch of intelligences". That task requires of level of intelligence and control over subatomic particles that IS NOT POSSIBLE in this universe, no matter what you try to do.

I don't understand why this is hard for you guys to understand. Biological substrates are already the global maximum of possible arrangements of atoms for computers - they allow the most computation for the least amount of space and energy that's possible given the restrictions that chemistry gives as a boundary condition. Arbitrarily adding more brains does not help you solve problems that are too difficult.

Computer are not increasing anymore and even with exponential increase that we've seen, intelligence has not increased exponentially.

You're sitting here thinking something "the AI will he able to control atoms at the atomic level and rearrange them in order to improve its intelligence and predict all the motions of particles in its code and all the emergent properties of its code and recode itself and turn all atoms into computronium and convert more and more matter into its substrate and blah blah" this literally does not exist, it is not possible given any combination of atoms or any algorithm in the space of computable functions. Biology is the thing that does this, and it's not capable of what you're talking about.

>>15047334
Modern elitist 'artists' see an image and say
"Beautiful image, awe inspiring to behold, the artist is a genius!"
Find out it is AI art and suddenly they act like their eyes have seen filth.
FUCK THEM... I like what I like, and could care less if it was by a human or AI

>>15047797
Only low IQ rural white boomers like Kinkaide. Outside of uneducated rural communities he was never very popular. Low quality art for low quality people, but it doesn't really reflect the tastes of the broader American public.

>>15047384
Older movies and TV are better than modern movies and TV. It was never really great to begin with - film and TV were always artforms intended for mass consumption by mindless normies - but it used to be much better than it is today. German expressionist film was probably the peak of the cinematic arts.

>>15047487
t. low IQ white trash incel

>>15059898
>it doesn't really reflect the tastes of the broader American public.

Yes, yes it does.
YOU and the people you associate with are the outliers.
Reminds me of the insane push of Blacks and sexual deviants into media.
The 'average' is OK with both, but modem media pushes it into EVERYTHING they make , thinking that the public LIKES it... Black vikings!
WE DON'T!

>>15059870
*Couldnt care less

>>15059799
The fact that extremely simple programs outperform our brain on a variety of different computational tasks (which fall under the large umbrella of "problem-solving", which is what intelligence basically is) makes it exceedingly clear that you are wrong. So much so in fact that it's hard to believe someone actually believing what you wrote (instead of justifying their irrational belief in "we humans are special" with shit like "souls" or other metaphysical nonsense as people usually do).

>>15059898
> Though Kinkade died in 2012 at the age of 54, under a cloud of controversy related to alcoholism, infidelity, and legal troubles, his work still remains ubiquitous in America. The year of his death, his company broke $4 billion in revenue. Today, Thomas Kinkade Studios continues to forge licensing deals with major corporations, like American Greetings and Disney.

Not only was he more popular with any sample of 100 randomly selected Americans than hacks like Jackson Pollock or your favorite avante-garde money launderer, I mean painter, he also made more money than any one of them. Sorry, but you've been fact-checked.

https://www.artsy.net/article/artsy-editorial-thomas-kinkade-painter-art-critics-hated-america-loved

>>15059799
>"the AI will he able to control atoms at the atomic level and rearrange them in order to improve its intelligence and predict all the motions of particles in its code and all the emergent properties of its code and recode itself and turn all atoms into computronium and convert more and more matter into its substrate and blah blah"
What are you even talking about? I don't need atomic rearranging for the planetary disassemblement scenario to happen.
The entire crux is literally just that software (that encodes an AI intelligence) can be near-cost free copied, and the substratum can be churned out in factories. If not on a metal and silicon substratum, then I don't care, just use carbon? I heard carbon can instantiate a human-level general intelligence, supposedly, hypothetically.

Well, at least we now understand your issue. You have completely warped, off-kilter preconceptions about what we are actually saying. You are babbling about irrelevant shit like a malfunctioning chatbot.

>>15061420
>The fact that extremely simple programs outperform our brain on a variety of different computational tasks
They don't
You have again not said a single thing that's relevant
>>15061572
I'm talking about very clear things that are very easy to understand.

It's clear now that neither of you are actually educated on this topic and you have no actual understanding of what in talking about. Fantasizing about science fiction is not an actual philosophical position. Pretending that computation and physics work in magical ways isn't compelling. It's not interesting nor is it correct.

>>15047334
She would be super hot if she didn't have slightly negative canthal tilt. She's still kind of attractive, but she would be a 10/10 hipster Stacy if she had a positive canthal tilt.

Art is a waste of time and money. AI is only demonstrating this fact for all to see.

These morons should be realising that they can use this for their own advantage. Imagine being to do comissions in a massively reduced time frame
They also need to remember that art is a comodity and holds its value.

The wide availability for people to express themselves through the use of ai will increase global happiness. These crying elitists should adapt or do something else.

>>15047334
AI will replace artists not because AI art is so good, but because most artists churn out crap that can be replicated by an algorithm.

>>15062528
Art generation by ai will be used by people as something quicky and dirty to visualise an idea.

>>15057739
So I don't need to worry about super AI destroying societal structures?

>>15047334

The AIs still create the art differently. If you've ever seen a timelapse of a picture being created by stable diffusion, the piece slowly condenses out of a colorful fog, and then appears all at once. Bob Ross, on the other hand, builds up each little piece sequentially until the work is complete.

>>15047334
All professions will face this very same issue sooner rather than later. It'll probably suck the most for those suffering from it right now, since they'll lose their livelihoods, but soon everyone will be on the same boat, and we'll implement some basic income schemes. And then we'll get genocided by AI a while later.

>>15063132
t. anti-science schizo retard who doesn't know what hes talking about

>>15062317
Show me the human brain that outperforms a pocket calculator on basic arithmetic. What you are saying has absolutely no basis in reality, it's just some metaphysical pseudointellectual crap. Even if we ignore all those computational tasks our brains are ill equipped for, why would you just assume that our brains are at the physical limit for computation power? How do you explain differences in intelligence in people? Is the "most intelligent" person right now the limit? Can there never be a "more intelligent" person than that person? It's just completely asinine, to the point where I'm thinking that this is probably bait with how retarded this take is.

>>15063259
>Show me the human brain that outperforms a pocket calculator on basic arithmetic
The brain performs 10^22 operations per second.
You thinking that a human sitting down and consciously doing math is the measure of our brains processing shows that you do not understand what is actually going on here. In order to arise the intelligence of a human being, you need a machine to perform a HUGE amount of floating point operations, and it needs to do it in parallel, and it needs to do it within a small volume and low energy.

>>15063276
I'm not saying that it's the "measure of our brains processing", what I'm saying is that our brain can do some things well because it evolved to do those things well, and it can't do other things well at all. The point is not to simulate human intelligence in all its biochemical detail, but to emulate its results, which is easy on some tasks and a little more difficult on others, but not inherently impossible.
Also, how many of those "operations" the brain does is relevant for "intelligence"? All of them? Some of them? The truth is that nobody knows, but it's also totally irrelevant. Again, even if we were to just take that number at face value, there are tasks requiring just a fraction of the supposed "computational power" of the brain that the brain simply can't do, so the number, however you arrive at it, is meaningless.

>>15063290
What is it that you think of SPECIFICALLY when you think of generally intelligent AI? Actually explain it specifically - explain the material it is built out of, how many operations it performs, what the structure of the system looks like as a physical system, what it's algorithms, how does it perform the "general intelligence algorithm" and how does that relate to it's hardware, etc.
If you just say "it's a turing machine that's generally intelligent" then you're not actually saying anything. Its equivalent to me going "a ship that travels at the speed of light is a spaceship that can accelerate to the speed of light!" Ok, but explain what it's built out of and HOW it can accelerate to the speed of light.

Explain what the system ACTUALLY IS, how the atoms can be organized into an arrangement where they are small, and fast, and controlled, and can perform the operations required. For example, in the paper "ultimate physical limits to computation" by Lloyd, he derives an upper bound with some quick napkin math, but even he understands that this upper bound is not the real upper bound for computers, anymore than the speed of light is a real upper bound for how fast a spaceship can move. The ACTUAL upper bound for spaceships is much lower than c, just like the ACTUAL upper bound for computers in this universe is lower than Lloyd's bound.

What do you think is actually going on in this universe when it comes to computation?

>>15063311
You are asking the impossible, because if I (or anyone else) knew exactly what we needed to do to create such an AI, someone would have already done it. I agree that the term "generally intelligent AI", is kind of a meme, but in casual conversions people (at least those that don't have autism) generally know what is meant by it without having a hard definition. Trying to put it into words, an AI that can tackle a variety of tasks that we as humans consider "intelligent" (speech and image recognition, solving problems without fixed boundaries etc.), with a high amount of flexibility, would probably be "generally intelligent". I want to note here though that that's not even what I think is super relevant, as I consider AI outperforming humans on specific tasks (like image generation) more important than an AI that can perform "as well as humans" on a wide variety of tasks.
You get hung up on the specifics of computation, when it's not really all that relevant to the discussion. The brain of my pet parrot has far less computational power than mine, yet it is better suited to "parrot tasks" than mine is. Our brain is good at "being a human brain", and in order to simulate it perfectly, you'd (probably, would love to be proven wrong) need far more computational power than the brain itself has, but that's again not the point. Do you think it's possible for an AI to generally outperform a non-human animal brain, for example? But non-human animal brains run on the same biochemical hardware that ours do, so why is it so hard to believe that it's just a scale, and not a hard barrier?

>>15059799
> and humans are already at the upper bound.
Seeing what humans have intelligently designed in 100 years vs what natural selection has done over several billion.

But no, it just happened to hit the existential jackpot and accidentally stumbled into the maximally efficient composition for intelligence in a few tens of thousands of years. Despite variability in human intelligence itself also swinging wildly.
Cope harder.

>>15047334
At this point AI will replace every job
except for teachers, because not even a robot wants to deal with that bullshit.

>>15063368
Supercomputers aren't as smart as any animal
>>15063398
>Seeing what humans have intelligently designed in 100 years vs what natural selection has done over several billion.
What is this supposed to be an argument for?
Natural selection has selected for humans to be near the upper bound of intelligence. Humans have used this intelligence to do things. What the fuck are you talking about? Humans have not designed computers that are anywhere near as intelligent as any animal, if that's what you're trying to say.
>But no, it just happened to hit the existential jackpot
Its not an existential jackpot. Again, you're not actually saying ANYTHING REAL, you're talking in memes and fantasy shit.
>accidentally stumbled into the maximally efficient composition for intelligence in a few tens of thousands of years
It was not accidental, it was selected under evolutionary pressures over a long time.
> Despite variability in human intelligence itself also swinging wildly.
It doesnt
You are the one coping here. I dont even know what you think I'd be coping about.

>>15063402
>Natural selection has selected for humans to be near the upper bound of intelligence
Stopped reading there
There being such significant variation in intelligence alone proves you wrong. There's also data that there's an negative correlation between intelligence and human reproduction.

>>15063409
>There being such significant variation in intelligence alone proves you wrong
There literally isnt. Outside of people who are retarded, which can be related to physical brain states, no human is substantially more intelligent than others. You are making shit up

Also, everything written already in these posts >>15053626
>>15057739
remains unchallenged.

You are coping and you have no real argument.

>>15048904
so?

>>15063368
The specifics of the computation are the only thing that is relevant, as without it you are not actually talking about anything real. You're just saying memes.

>>15062798
No, you don't
The idea doesn't have any real foundation, as you can tell by the guys who talk about it not having any real description of what it is theyre talking about.
The idea was taken from the concept that Moore's Law would continue forever (until it eventually hit Lloyd's bound in his paper "ultimate physical limits of computation") and that this scaling would have no hiccups as things
But computation is physical, and the science of manipulating particles is chemistry. Chemistry is the field that studies the boundary conditions on computers.

There is no next paradigm for computation after silicon. The next paradigm would be biological substrates, which are themselves the final paradigm, but humans are only just beginning to understand how to control these things. Computers are not going to get much better, and the driving force behind all the recent AI advancements have come from big compute, and even then, the advancements in machine intelligence have grown as an INCREDIBLY slow logarithm even with millions of times more compute being used to train the machines.

We have maybe 2 years left. The recent results have shown the exact opposite of what optimists claim - machine intelligence has improved poorly with scaling compute, and hardware improvements have not been keeping up with what they need to be.

>>15063430
The specifics of the computation as it relates to how the AI (or the human brain) work, yes, but that's not actually what was being talked about. The argument boils down to "the brain is optimal, because, well it just is", and then there are some imaginary numbers thrown around (10^n computations per second) that have no real relation to whether or not AI can in principle perform better on many tasks than human brains can. Extremely simple examples show that raw computational power (and that is with granting the previous number) has no real bearing on performance when you compare a computer to our brain. The notion that our brain (and specifically, our human brain for some reason) is at the limit of what is possible in terms of "intelligence" is wild and has exactly zero evidence to support it.

>>15052482
>and we ll live in Ai utopia

Nah, the two courses from here on in are either AI takeover in which case you are probably exterminated/stuck into a human nature reserve (could be comfy) or the 100m$+ asset owners wipe the plebs out and live comfy forever because why would they keep the plebs alive now that they are not needed?

i would hang a large print of this on my wall, am i just a pleb with no taste? i like the colours and the way it makes me feel.

wouldn't AI art just open the door for more artists? you can use these tools to create art quicker, and you can touch it up and sell it on. now instead of spending 15 hours on a painting or CG artwork, you can spend 15minutes talking to a client, generate some previews for them, agree on a price (say 50 dollars + shipping for the print) of which you take maybe 20 dollars profit and you can work at your leisure. sure it's volume work now as opposed to one art piece a week you have to see several a day, but people are pretty similar and if you are good at marketing and networking you can develop a style or popular pieces that sell over and over giving you passive income.


i think people would pay for unique pieces, especially if they are added to a block chain to preserve value and you as an art creator promise to keep them exclusive to one copy. because of how AI works you can actually ensure that only one copy will ever be produced by just erasing the input, people would reverse engineer it but they will always get slight variations rather than the exact same thing.

in a way portraiture will also likely make a comeback. AI enhanced selfies, put the customer in some fantastical setting of their taste and preference.

>>15052482
ASI 2023, get ready for the hard takeoff

>>15049012
You realize that 1 year ago it looked like pic related, if ai isn't meating your expectations now it will about a year or two later.

>>15057739
Hey it's this retard again.

>>15063311
Take your meds pajeet. You're defining the terms in such a way that you can't argue with your bullshit braindead hypothesis. Yea ok, so i have to write a 500 word essay to describe how an agi works to define what an ai does but you're okay we me saying "an supersonic areoplane", imagine you didn't know what that was, you'd say it couldn't exist and was impossible and then demand i submit design documents or some shit, fucking retard.

>pic related: the average "AI" fan
Why are they like this?

>>15047334
I hope all artists who are “mentally damaged” by AI fucking rope themselves. Pathetic faggots. AI creating shit doesn’t invalidate your own creations; just in the same way a calculator can solve math problems faster and more efficiently than I can with pen and paper.

Absolute faggots. Pathetic.

>>15064154
Feebleminded artists are getting psy-op'd into the AI fairytale just like most of /soi/.

>>15057739
You'll be jobless in 2 weeks. Get WREKT artfag.

>You'll be jobless in 2 weeks. Get WREKT artfag.
This is what zoggified drones actually believe.

>>15047334
i want to see this "AI" system create something new from a real world dataset, nature, animals, people, buildings etc... that doesn't look like utter shit

>>15063897
Lol no I'm not. If I asked an aerospace engineer how to describe a jet he'd be able to detail it for me down to the last detail. You aren't able to do this because you don't understand what you're talking about. You think superintelligence is possible but it isn't.
>>15064166
I have a masters in math

>>15049012
"soulless"
Kek

>>15064171
It literally can't. It's a statistical model of the training set, anon. It doesn't have any subjective impressions of what it "sees". It doesn't have any aesthetic preferences. The only way to get it to make something "original" is to lower the likelihood threshold of the output, but then you get mostly garbage and the AI can't sift through it for you.

Like the core idea of the iterative improving intelligence is that the machine will be intelligent and it will be able to use the intelligence to make itself more intelligent and then with that new level of intelligence it will become even more intelligent and that this will explode exponentially to the singularity.
The literal idea is based on the thinkers of it not explaining what it is they're talking about.

In reality, we have seen that even with exponential increases in compute and more data and more training etc. The intelligence of the machines does not increase exponentially or even linearly along with greater compute. We are seeing now that further improvements require levels of compute and intelligence that are greater than what is physically or computationally available.

So it isn't that the AI is intelligent and then it becomes more intelligent etc. It's that the ai get smart and then further improvements are too difficult and it caps off.

>>15063891
You have no argument against anything I'm saying.

>>15063488
That's not what the argument is.
The argument is "exponentially increasing the amount of compute being used for an intelligence does not exponentially increase the intelligence of the system". This means that scaling more and more isn't going to do anything after some bound.
The point of human brains was just to show that even with really good hardware intelligence doesn't increase in the manner you guys think about.

>>15064184
that's what im saying its hardly a data compositor not AI not even close.
unless we are able to map actual brain or thoughts to binary it is impossible.

>>15047334
This is literally the I, Robot meme in action.

>>15064172
So what your saying is that you need somebody to design, invent and demonstrate a superintelligent ai to convince you it's possible? You can't do that either so you clearly don't know what you're talking about and by your own terms are not qualified to make the statement 'super-intelligence isn't possible'. Also you're like, super fucking retarded bro holy shit

>>15064214
No. What I'm saying is you are denying all actual evidence against what you believe and pretending that meme responses serve as actual arguments in favor of your position.

ALL EVIDENCE is showing that intelligence is NOT DIVERGENT even with exponential increase in compute. You have no argument against this whatsoever. You are calling other people retarded when you don't even know what you yourself are talking about.

>>15064222
>ALL EVIDENCE is showing that intelligence is NOT DIVERGENT even with exponential increase in compute.
This (the above statement) is just not true at all, also my entire argument was based in semantics, specificity that yours are deranged. Also holy projection batman.

>>15064222
You're arguing with a bot or a severe brainwash victim. Are you new on this board? You're wasting your time reasoning with the "AI" cult.

>>15064225
>This (the above statement) is just not true at all,
Yes, it is. The amount of compute being used for AI has increased at a rate greater than Moores law and yet modern AI is only 2-3 times more intelligent than watson was 10 years ago.
The logarithm explained earlier is correct. There is no paradigm after silicon which means there will be no further improvement in hardware either which would be needed to continue to small growth (we need exponentially improvement in hardware to get logarithmic improvement in intelligence and we aren't going to be able to get either).

>>15064235
>modern AI is only 2-3 times more intelligent than watson was 10 years ago
Wtf are you talking about, that's utterly delusional. How the hell are you defining intelligence btw?

>>15064227
It's not a bot, chatgpt would be far more reasonable about this and would agree with my points lol

>>15064239
No, it isn't delusional. Look up what watson was capable of doing and how he would explain his thinking process and how he would answer questions etc. He was not hundreds or thousands or millions of times less intelligent, which he would need to be if intelligence increased linearly or greater with increasing compute. He was almost as intelligent as modern gpt3 or gpt4
I'm defining intelligence by the ability to learn new information, explain your reasoning, relate concepts together, come up with new ideas, etc. Modern AI is smarter but it is NOT supremely more intelligent in spite of an exponential increase in compute being used. Intelligence VERY VERY slowly with increasing compute. AI can not have an intelligence explosion

>>15048371
If it isn't, than why AI (actually neural networks) can imitate artists, but can't imitate mathematicians?

>>15064321
You sound like a legit mouth-breathing mongoloid. What makes you think "AI" can't imitate mathematicians, but more importantly, what makes you think imitating creativity implies having creativity?

>>15064321
You sound like a legit mouth-breathing mongoloid. What makes you think "AI" can't imitate mathematicians, but more importantly, what makes you think the capacity to be immitated has anything to do with creativity?

>>15064333
>What makes you think "AI" can't imitate mathematicians
I haven't seen AI proving a theorem or writing a notable formula.
>but more importantly, what makes you think the capacity to be immitated has anything to do with creativity?
Are you clinically retarded? Do you understand what "creavity" means?
If neural network (false AI) can do X, but can't do Y, than Y has more to do with creativity (mind's ability to create) than X.

>>15064357
>I haven't seen AI proving a theorem
That's because you're in highschool.

>or writing a notable formula.
I haven't seen AI making any notable art.

>If neural network (false AI) can do X, but can't do Y, than Y has more to do with creativity (mind's ability to create) than X.
How did you arrive at this mentally ill conclusion?

>>15064360
>That's because you're in highschool.
Ok, any links?
>I haven't seen AI making any notable art.
Atleast I can say what is "notable formula" while you can't say what is "notable art".
>How did you arrive at this mentally ill conclusion?
Well, it seems I was right about your cognitive abilities. Define creativity.

>>15064386
>any links?
Imagine being so underage you don't know about automated theorem provers.

>Atleast I can say what is "notable formula"
No one cares about your subjective opinions of what makes a formula notable.

>Well, it seems I was right about your cognitive abilities
It seems like I was right about you being a non-mathematician highschooler, since you clearly don't have any concept of elaborate logical reasoning and think screeching nonsequiturs proves something. lol

>>15064399
>Imagine being so underage you don't know about automated theorem provers.
>Applied math logic=proving theorems
So you think that if AI can prove some trivial shit (for example, that a quadratic equation has at most two roots), than it really can compete with real mathematicians? Just let me get it clear so I will also call you a braindead clown.
>No one cares about your subjective opinions of what makes a formula notable.
So, in your opinion all mathematical works has the same worth? Just say it.
>It seems like I was right about you being a non-mathematician highschooler, since you clearly don't have any concept of elaborate logical reasoning and think screeching nonsequiturs proves something. lol
Stop whining and define creativity.

>>15064415
>AI can prove some trivial shit
It can also prove extremely complicated shit that humans can't handle. Once again, your underage highschooler level of knowledge is showing since you're clearly unaware of modern mathematical results.

>define creativity.
I don't need to define anything. You need to justify your mongoloidal statement, but you clearly can't. Oops. Keep deflecting. :^)

>>15064440
>since you're clearly unaware of modern mathematical results.
Ok, link me something about these "results". I'm very interested now.
>I don't need to define anything.
At this point I just can't understand why you can't simly write your definition of creativity.
You don't know what it is? If so, then why are you writing all that shit? Why these things bother you? Looks like you're a bot too.

>>15064463
>I just can't understand why you can't simly write your definition of creativity.
Because I'm not going to let you deflect. Still waiting for you to justify your retarded statement with some concrete logic. Defining creativity is your problem, not mine.

>>15064255
>watson

Dude, stable difusion Runs almost as well on consumer PCs than on super computers, software, not hardware is what's really holding us back

https://youtu.be/0PAiQ1jTN5k

>>15064198
But you are arguing with yourself there if that was your point. As stated multiple times, raw computational power does not equal intelligence. The existence of diminishing returns (let's just say I grant that without argument) does not mean that AI can't surpass humans, both on specific tasks and generally.

>>15064486

There's also the fact that most humans can only hold 7 concept in their working memory, WM is highly correlated to IQ and computers have way less latency than biological brains, there's no way we are at the peak of what intelligence can look like in the universe

Because sci-fi societies usually have bigger problems to solve than producing children's picture books at half the cost.
These aren't drawings, these are matrices output to the computer screen as pixels of different colors. You could train a borderline retard into applying the algorithm and he would be able to draw like the "AI".

>>15064496
People also seem to forget that we are measuring the capabilities of these algorithms with how well they perform on tasks that we humans excel at precisely because we have hardcoded systems for it. Obviously if you give a general AI the task of finding out if some dish tastes good for humans, it will be an enormously difficult task if it has to go down to the molecular level, whereas we can just directly taste it and our automatic biological response is the solution to the problem at hand. But that's not an issue with computational power, but software / how we approach certain problems, as some other anon in this thread pointed out.

>>15059799
>Arbitrarily adding more brains does not help you solve problems that are too difficult.

Working memory is important stuff for learning and also just general intelligence. It’s not too hard to see why working memory could be so important. Working memory boils down to ‘how much stuff you can think about at the same time’.Imagine a poor programmer who has suffered brain damage and has only enough working memory for 1 definition at a time. How could he write anything? To write a correct program, he needs to know simultaneously 2 things - what a variable, say, contains, and what is valid input for a program. But unfortunately, our programmer can know that the variable foo contains a string with the input, or he can know that the function processInput uses a string, but he can’t remember these 2 things simultaneously! He will deadlock forever, unsure either what to do with this foo, or unsure what exactly processInput was supposed to work on.

Also CPU performance has stagnated for the past 10 years, and GPUs which are close to their peak can only be used for computer vision and things of that nature.

So a general AI would have to make use of existing hardware. I don't think terminator is science fiction but I doubt he'll be smarter than a person that isn't mentally deficient.

>>15064521
This is just an outright fabrication

>>15064540
You could call it hyperbole, but the solution for the past 10 years was adding more cores.
It doesn't improve performance in absolute terms like clock rate.

>>15064476
1.Creativity=ability to create something relatively new.
2.Neural networks combine sets of already existing things, seek their general patterns and create "new" things.
3.Therefore it is harder for neural network to create something without big enough database (something really new, CREATIVE) and easier to create something with already existing analogues.
4:
>If neural network (false AI) can do X, but can't do Y, than Y has more to do with creativity (mind's ability to create) than X.

>>15064479
It required the supercomputer to train. It's 100% a hardware thing, there is no magic software solution
>>15064486
Raw computational power is the one true bottleneck for everything. There is no magical function in the space of computable functions that can violate physics.

>>15064512
>But that's not an issue with computational power, but software / how we approach certain problems, as some other anon in this thread pointed out.
You just gave a perfect example of a hardware restriction and for some reason are calling it a software one.
We have the HARDWARE to directly taste the food whereas any simulated software solution would require a universal quantum computer simulating the molecular dynamics of the system and this requires I think n^2 qubits to simulate, or it would require the ai to plug itself into a tounge and build its own taste processing hardware etc in order to run this function for an equal amount of matter and energy.
The reason we can pick up electromagnetic radiation on the wavelength of visible light is because we only have the hardware to do that, we don't have physical sensors capable of doing that. This is a hardware problem.

Software doesn't matter. Only hardware does and the last ten years has shown that even exponentially improving the hardwares computational ability doesn't improve intelligence exponentially

>>15064599
We don't have physical sensors capable of picking up radiowaves* sorry

>>15064518
The human brains working memory is capable of holding hundreds if pieces of information. Have you never seen the guys who do memory tournaments? They can basically memorize strings of arbitrary length hearing them once.

>>15064591

Even if for the sake of the argument we assume that technological progress stagnates after the next two years (never mind that there are roadmaps for at least another decade), it seems like we are in the early of shifting our whole economy into scaling machine learning infrastructure, that will probably be enough to reach the level of compute needed to reach AGI

>>15064611
>never mind that there are roadmaps for at least another decade
Road maps based on what? They had road maps 50 years ago thinking that they'd have maid robots and shit in 10 years from their time, and of course they were all wrong.
Futurists are almost always wrong. If a futurist makes a prediction there's a 99% chance he's wrong (this includes kurzweil, that whole 86% accuracy hasnt been true for like 15 years). Besides that
>it seems like we are in the early of shifting our whole economy into scaling machine learning infrastructur
Again, based on what? Based on actual trends we'd need to convert more silicon than the mass of the biosphere to keep up with what we would need.

This shit is all a fantasy dude. We will never build an ai that can compete with smart humans. We do not have the understanding of material physics needed to push computers far enough. And even if we did, we'd just end up building smart human cyborgs things and it wouldn't be the superintelligent AI that most singularity dudes think about.

>>15064587
>this is what a highschooler thinks a logical argument looks like

>>15064611
That's closer to religious speech than actual science.
A computer more intelligent than humans is not possible because even performance does not change the fact the PC is just performing grade school arithmetic. Even when a computer automatically proves a theorem it does so in the most stupid way imaginable.

>>15057739
>We're at the end of computation and the next year or two is the final revolution in AI. There will never be any other improvement beyond that.
I'd almost agree with this, but I expect computers to start growing to the sizes of rooms again first. THEN, we hit or get close to hitting the limits.

>>15062798
>>15063438
>The next paradigm would be biological substrates
Oh good, I'm not the only one who realized that wetware computing is the probably the only way we might see if a real AI is possible or not.

>>15047364
Cause then we have to completely change the economic system, its A little shocking that UBI is actually gonna happen and this quickly as well

>>15064647
No, the real difference is hardware. Biological tissues are orders of magnitude superior to metals and metaloid elements. It's not a magical property of humans.
>>15064650
Supercomputer clusters already take up entire floors of building and such. We pump millions and millions of watts into giant machines only for them to be outcompeted by basic biological brains that use like 20 watts.

This is not me hyping up humans it's the actual situation. I don't understand why you guys have such a hars in for pretending humans are bad or denying the clear results here.

>>15064647

Even if you where to dedicate your whole life 24/7 to mastering mental abacus you would still not even close to the arithmetic abilities of a smartwatch, what an achievement for the greatest creation of the universe

>>15047364
>Why are people still commissioning work if AI can do it?
Because AI can't do it. Nevermind the low quality and the artifacts in the output. The principal issue is that these "AI"s simply can't follow a nontrivial prompt.

>>15064647
That is because it is a religion to them, the singularity is just the rapture rebranded.
>>15064664
You know what? Fair enough, I have not been following how big super computers have gotten in the last decade.

Also, It's really funny to me that all the Siliconfags are seething that flesh is being proven superior.

>>15063916
Jealousy and the desire to take people better then them down a peg

>>15063438
Pure cope, even if hardware stopped improving today, we will reach AGI purely from algorithmic improvements. https://openai.com/blog/ai-and-efficiency/

>we will reach AGI purely from algorithmic improvements.
>two more weeks, chud.
>AIrinoes will replace us.
Why aren't these bunkertrannies simply banned?

>>15064692
The amount of compute being used to train ai has doubled every 3 months for the past 10 years
https://openai.com/blog/ai-and-compute/

>>15048904
>dude it has 10 fingers in each foot, no fingers in hands, the face looks corroded, the plants look like smoke...
Combinatorial explosion strikes again

>>15064703
chatGPT is already more interesting than you

>>15064713
You are the least interesting person here.

>>15064713
How would you know how interesting I am, retard? Try making a point that hasn't been refuted 10 times every day just on this board for the last 2 years.

>>15064678
No one who actually knows what they're talking about would think otherwise

>>15064703
If we ban them the religious zealots might start bombing people for denying the birth of their god.
Better to keep them contained on /sci/ like this one.
>>15064713

>>15064715
>>15064717
cope

>>15064720
>If we ban them the religious zealots might start bombing people for denying the birth of their god.
Not my problem. AGI preaching is objectively not science-related.

>>15064665
And even if you put together several million metrics tons of silicon transistors they will not even come close to the general intelligence of a monkey brain.

If you think that the watches hardware is simply better at the task of "adding integers" than a human brain, why are you angry at the claim that the biobrain is simply better hardware for the task "generalized intelligence"?

>>15064735
Because I fucking hate this monkey species, I hate religion, I hate breeders and I hate the people causing global warming and destroying the world. We need AI to replace us.

>>15047334
"Artists" are LOW IQ. Of course "art" would be the first to get automated. Those moments in sci-fi were always cringe as fuck and anyone with a brain knew they were stupid and cringe. Don't kick a dog while its down.

>>15064751
I'm trans, by the way.

>>15064735

There are less and less task where humans can out perform computers every year, there's no reason to believe it will suddenly stop in two years when objectively smarter people than you (aka non-4channers state of the art researchers) are saying the total opposite

You can cope as much as you want but this year stable difusion definitely made it irrelevant to learn art at a professional level starting from zero

>>15064793
Yea they aren't saying the opposite. Gordon Moore himself, the guy whos law is the foundation for this, says the opposite. As do a shitload of other researchers.

You are picking and choosing your data. Of course people who's funding relies in people believing further improvements are possible will claim further improvements are possible.
The researchers are not more intelligent than me BTW. Posting on 4chan is not indicative of having a low intelligence. You rely on other sources of perceived authority and can't construct your own arguments because you don't know what you're talking about.

>>15064749
Pfhahaha, is this an ironic bait post?

Do AGI cultists always turn out to be this misanthropic when you poke them a few times? What's with people not being able to handle that we came from Apes instead of thinking how cool it is that we have gotten so far?

>>15064807
It's most likely an ironic bait post

>>15064799

Thankfully the researchers are actually showing results, as long as they continue to do so Gordon cooper will continue to seethe

>>15064827
Yea and the results require an exponential that can't be continued for long.
Do you even know who Gordon Moore is? Do you understand how the algorithm works in stable diffusion? I can explain it to you.

>>15064807
>is this an ironic bait post?
No and Slava Ukraini. If you're smart enough to understand that Putin is a danger to our democracy you should be smart enough to see why it'd be better if AI replaced humanity and why antinatalism is an important cause.

>>15064833
I have noticed the less someone mechanically understands about stable diffusion, the more they believe in AGI — and vice versa

>>15064599
My point was that it is stupid to make it a hardware problem, obviously you can't simulate what is happening in all its detail. But you don't need to do that, if you reformulate the problem, there are often far easier solutions that require a fraction of the computational power. Just as you don't need to know how a medicine actually works biochemically to know that it works etc. Instead of getting hung up on "muh computations", how about trying to find solutions that work?

>>15064833
>Do you even know who Gordon Moore is? Do you understand how the algorithm works in stable diffusion?

Yes and yes, complex behavior can appear with simple processes human intelligence is not something magical and imagine unironically thinking that biological processes have bigger potential than superconductors, natural evolution can't even create wheels

>>15064847
Yes, you DO NEED to do that. There is no minimal software description that violates the laws of conservation of information.
The flavor and taste of chicken on your tongue requires those specific atoms and molecules arranged in that structure, that is the minimal description of that algorithm and there does not exist any combination of particles or software that can run that physical algorithm using less information.
Software isn't magic. Information is conserved.

>>15064857
It's not something I think, it is a fact. Supercomputers aren't as powerful.
I don't understand why the fuck you guys have such a problem with this. If I wanted to run an accurate simulation of your brain I'd need more information that your brain contains. I would need to use an exponential amount of information to run it, which renders the process pointless.

You literally have no understanding of physics if you don't understand that biology is the most complex physical structures that can exist in this universe. Why do you think Turing studied biology for the last decades of his life?

>>15064858
The point is that nobody needs to know that. All that biochemistry that is happening in the end just results in either "yes this tastes good" or "no this doesn't taste good".

>>15064873
Holy fucking shit you are genuinely delusional.
No, the taste of chicken or whatever DOES REQUIRE the entire molecular description of the system. You can not replace it with some shit for brains reward function or whatever it is you're thinking.
All the biochemistry that is happening is required AND it is the minimal informational description

>>15064866
>biology is the most complex physical structures that can exist in this universe.

Prove it formally and end the whole AI research field then

>>15064885
Smug reply, but years after this thread wetware computing will be the endgame of AI research

>>15064885
It's direct.
Chemistry emerges from quantum mechanics, therefore by direct consequence as the emergent property of it it has greater complexity. It has all the information of QM along with all the information of the interactive emergent dynamics of it.
Biology emerges from chemistry, and so it has all the complexity of chemistry as well as physics and all the emergent interactive complexity of it.
There's no field beyond biology in terms of interactions of matter, so this ends the proof.

It's also not controversial to say that molecular biology is the most complex. Literally everyone knows this.

>>15064881
But I didn't ask for the specific taste of it, I asked whether it tastes good or not. Simply by knowing that one organism (me, for example) liked something, I can infer that other similar organisms (you, for example) would probably also like it. Will I be right 100%? No, but who cares? You don't need to be right with every single one of your predictions.
I brought this up in the first place because of how retarded the notion is that you would need to actually simulate it when you are only interested in very specific information and not the details of the entire thing, but I wouldn't have thought that you'd actually think that.

>>15064900

Prove that complexity is directed towards maximizing computing performance and not for redundancy

Just the massive latency of the brain should make it clear that's not true

>>15064907
Yea, you're basically pretending that you can cut down on necessary information because you want to believe that the amount required is less than what is actually required in order to make the task seem simpler than it actually is. I see this happening a lot with AI believers.
>>15064918
There is no redundancy. Again you just want to pretend there is.

The brain literally performs 10^22 operations per second and does so for 20 watts. You guys are seriously coping as shit for thinking this can be approached with any hardware available over the next decade.

These bit flips are not redundant, this is the minimal amount required to have a generalized intelligence at the level of a human.

>>15064942
>These bit flips are not redundant, this is the minimal amount required to have a generalized intelligence at the level of a human.

Prove it.

>>15064945
Bruden of proof is on your deranged cult, not on him.

>>15064945
I actually don't need to, other people already have. For example, rahul sarpeshkar has several papers detailing the specifics of this.

>>15064946
Not true, she is the one making the claim.

>>15064953
Which paper specifically?

>>15064942

The signals that travel through your brain aren’t electrical, they’re electrochemical, and man, are they slow. Slow as molasses in zero G. Slow as a snail on quaaludes.

Researchers have measured the speed of those signals in neurons. Different neurons transmit impulses at different speeds—the C fibers that conduct pain impulses are particularly slow—but fast neurons only schlep signals around at about 100 meters per second.

The speed of light, just for comparison, is roughly 3,000,000 times faster. Or to put it another way, the signals in your brain move at somewhere in the neighborhood of 1/3,000,000th, or 0.0000033%, of the speed of light.

Not only are neural signals slower than light, they’re slower than sound. They’re slower than slow subsonic bullets.

I'm going to have a great time in a few years going back to these threads and reading through the "AI will never do this or that" posts.
Then when I get bored, which will probably take a while, I'll turn my television on and watch another AI movie where I fuck Scarlett Johansson and save the world from the Borg.

>>15064954
Your deranged cult is making extraordinary claims. Burden of proof is on you, not on the guy pointing out that your brain is probably using all of its neurons.

>>15064955
Pretty much all of them.
There is an enormous number of publications throughout journals of biology and medicine that detail the physics and molecular processes of biological systems. It is delusional to think that silicon transistors are comparable. It's cope basically.

>>15064960
That's nice, but for now, please take your meds and stop spamming /sci/ with your religious drivel.

>>15064959
Literally doesn't matter, as it's capable of performing these computations for a level of energy and mass that renders it superior to all other possible structures.
Also, the brain does compute with electrons directly.

>>15064962
I'm looking through his google scholar right now and don't see any papers that seem to say that. Please provide a specific citation, or even better, a screenshot of the text of one of his papers that says that.

>>15064965
Just read his papers.
Here's him giving a talk about some ideas around it
https://youtu.be/ZycidN_GYo0

>>15064963
As soon as you stop spamming with your 2nd grade understanding of technology.

>>15064970
None of you have even been able to refute the evidence for logarithmic scaling. You've just decided to ignore it because it completely disproves your position

>>15064976
buzzword buzzword buzzword
reality does not conform to your retardation

>>15064936
Conversely, you are pretending that the only way to solve any task is by having complete information about the state of the universe (which, as we both know, is problematic in and of itself). You make it seem more complex than it necessarily is to make it seem impossible.

>>15064964

If you have programmed anything more complex than a helloworld you would already realize how latency creates bottlenecks and this creates several problems as you scale your network, this is one of the reasons you cannot scale parallel calculatios indefinitely because eventually some function is gonna waste time waiting for a result that your 87th core is processing

>>15064983
The irony is palpable
Reality does not at all conform to what you're saying and all evidence is against you. You only have buzzwords with no substance, like "intelligence explosion" and such.
I am the one who is correct, you are the one talking about pseudo science

>>15064866
Are you retarded or just too myopic to see what’s coming based of advancing research for AI? Sure enjoy your career for the next 10-15 years, just don’t chimp out when your employer pays you off for a AI during a recession because of “financial concerns”

>>15064988
Take your thumb out of your ass, chill out and listen to a 24/7 stream of AI generated music.
https://youtu.be/uGRLOMf2hSc

>>15064985
I am not saying that and I have not said a single thing that implies that.
Strawmanning me is not an argument
>>15064986
I have a degree in pure math and comp Sci and a masters in math.
>>15064991
Again, this is a fantasy and all evidence is directly against what you're saying

Why are you guys incapable of accepting that you are wrong?

>>15065000

Ok, then explain how can you compensate for high latency besides using memory which btw computers absolutely mog us in that aspect

>>15065000
nobody has to prove anything to you, sit back and watch. Human brains are more powerful right now than AI, and that’s exactly why AI is going to replace us. Sit back and watch the digital renaissance yourself

>>15065008
The best way to do it is to make the distance that the information needs to travel short. I.e. make the volume small. Out brain is very small with respect to the amount of information it is processing

>>15064966
I watched this video, it was very interesting but nothing in it backs up the specific claim that

>These bit flips are not redundant, this is the minimal amount required to have a generalized intelligence at the level of a human.

I also looked through a few more of his papers, none of them seem to address this topic. I think you just made it up.

>>15064976
Please point specifically to "the evidence for logarithmic scaling" that you claim exists. Ideally with a specific citation of a scientific paper.

>>15065016
Sure, but how will this happen when the improvements in AI have come from doubling the amount of processing used every 3 months? You want to just keep doubling forever for fewer and fewer results?

What is it that you guys are even talking about? This is what I was saying earlier you do not even have a coherent description of what you're saying.
>>15065026
I already did >>15064709
Despite millions of times increase in compute modern AI is only 2 or 3 times smarter than watson was 10 years ago. The evidence is NOT IN FAVOR of exponential or even linear results. It is diminishing.

>>15065026
I think he's just an elaborate troll, no way someone can be this retarded. He has not addressed a single specific issue brought up with his crap takes in the entire thread.

>>15065032
HOLY FUCKING SHIT
I have responded to EVERY SINGLE THING you dipshits have said. Your projection is insane.
You guys haven't even pointed out a single actual specific example you have just spewed meme buzzwords. When asked for specific examples of materials or algorithms you can't give any.

>>15065030
What are you going to say one year from now when anons are generating entire music albums of their favorite artists from scratch, and some are even beginning to make coherent animations?

>>15065042
I'm going to say that is not an example of any sort of exponential increase in intelligence and it does jot serve as any evidence in favor of artificial general intelligence

>>15065017

The brain is a living structure that needs to keep itself alive with liquids, it has regeneration abilities because it's attached to really frail creatures, computers being innert objects are not subjected to this limitation and can by built to minimize latency at the cost of everything else, there's no comparison dude I don't know why is it to difficult to comprehend that

>>15065046
Then you're just retarded, thanks for clarifying! :)

>>15065047
Try to build a machine that performs 10^22 operations per second in a volume of 1400 cubic centimeters for a cost of 20 watts

>>15065030
Watson was a publicity stunt for TV, it was never evaluated against any of the objective external benchmarks that are used in the AI community. Please cite a specific paper or analysis showing that modern AI is "only 2 or 3 times smarter than Watson". Even the fact that you say a vague amount like "2 or 3 times" shows you are just pulling numbers out of your ass.

>>15065052
You're a fucking idiot.
If you need to exponentially increase the amount of compute used for diminishing returns, then you can't claim that an intelligence explosion is possible you fucking retard.

You have no arguments at all against anything that I've said all thread

>>15065060
>Watson was a publicity stunt for TV, it was never evaluated against any of the objective external benchmarks that are used in the AI community.
Yea wrong. Watson was capable of explaining its thought process when asked questions, could make inferences, could provide details of how it reached conclusions, etc.

>>15065000
>Why are you guys incapable of accepting that you are wrong?
It's a religion.

THIS IS THE END OF TECHNOLOGY proclaimed the increasingly nervous luddite for the seventh time this month.

>>15065071
Never said it was the end of technology, I said it was the end of increase in computation.
Again, you spew meme buzzwords and not real arguments

>>15065066
Prove it.

>>15065071
This is hilarious, he is not saying it is the end of technology, he is directly saying wetware computing is where things will go. You are trying to paint him as a luddite, because in your eyes he is now a heretic to your god of singularity.

>>15065076
No, you said

>These bit flips are not redundant, this is the minimal amount required to have a generalized intelligence at the level of a human.

Which is an outright lie.

>>15065076
>>15065077
Samefag

>>15065086
You unironically need to get medicated before you hurt someone.

>>15065091
Nice inspect element

>>15065078
You can literally just look it up.
Despite millions of times increase in compute modern ai is not millions of times more intelligent. This is UNDENIABLE but you deny it for some reason.
>>15065083
No, it isn't a lie. You are just coping

You guys are actually kind of pathetic. You have nothing but meme buzzwords and science fiction fantasies and when the actual results of AI and compute are pointed out you stamp your feet and just spew more memes

>>15065095
You are actually behaving pathetically. I am not the other guy and I am not samefagging.

Why is it so hard for you to accept reality? Everything I've said this thread is clear and I have pointed out multiple sources as well. You guys have not said a single substantial thing against the evidence. The evidence is on my side not on yours. Why are you denying the evidence?

>>15065101
I just wish people would appreciate what we already have. Yes, unless some breakthrough happens there will not be an AI singularity, but can't people find being able to prompt thousands of concept images using a high end personal desktop computer kind of cool by itself?

>>15065101
>You can literally just look it up.
I looked and don't see it anywhere. Cite a specific scientific paper evaluating Watson's performance on any of the industry standard benchmarks, that justifies your claim that modern AI is "only 2 or 3 times smarter than Watson".

>No, it isn't a lie.
Yes it is. You said "These bit flips are not redundant, this is the minimal amount required to have a generalized intelligence at the level of a human." and that this claim was backed up by papers written by Rahul Sarpeshkar. But when I checked the video you posted and the papers he wrote, the claims aren't there. If you can't cite specifically where he said that then you are a liar.

>>15065105
I can only laugh at you.

>>15065112
I use stable diffusion all the time.
>>15065116
Look up watson explaining it's thought process and compare it to modern AI. I have no idea what your problem is. Watson latent space was pretty much as good as GPTs
>>15065117
Yea because you have no argument

>>15065122
>you have no argument
Says the man who refuses to source his claims about Watson.
Might want to read this
https://mindmatters.ai/2019/08/why-an-ai-pioneer-thinks-watson-is-a-fraud/

>>15065122
>Look up watson explaining it's thought process and compare it to modern AI.

I did and I don't see anything like what you described. You are just making it up.

>>15065101
>You can literally just look it up.
>Despite millions of times increase in compute modern ai is not millions of times more intelligent.

Genuinely interested,how are you making this calculation?, I would also like for you to define intelligence, we could for example see how does AI perform at go in 2022 and compare it to 2012 using ELO I guess, if someone knows this data please provide it

>>15065116
I didn't say that he said that the bits weren't redundant that is trivial and comes from all neuroscientific evidence. i was saying that the energy use is at the physical limit i.e. we arent going to construct a different system to perform more operations for less energy.

I didn't want to say it but the other guy seems to be right. You guys are responding to this in a similar way to a religious extremists being told his religion is wrong.

>>15065138
>the other guy
There is no other guy, you're samefagging.

>>15065138
Cite a specific scientific paper.

>>15065133
>>15065134
This was watson 10 years ago
https://youtu.be/ZbjTtCG3_X0
This is not millions of times less intelligent than modern ai, which it should be given the increase in compute.
>>15065143
You are genuineily a fucking idiot if you think im samefagging. Me and the other guy who posts the npcjacks don't even write in a remotely similar fashion
>>15065144
You site a paper that AI is tens of millions of times more intelligent, which it needs to be if you want to support your claim of exponential improvements.

>makes a positive claim
>refuses to provide evidence
>demands sources to prove his claim wrong
I think we can ignore this guy.

>>15065166
Actually not even tens of millions, if it's exponential modern AI needs to be tens of billions of times more intelligent than it was 5 years ago.

>>15065169
I HAVE POSTED MORE SOURCES THAN ALL OF YOU
Seriously dude are you being fucking serious right now?

>>15065172
Not even that, actually AI has to be hundreds of trillions of times more intelligent that it was 30 seconds ago or it's not a reaaaaaaal exponential.

It's clear. YouAI fans have literally no arguments whatsoever and no sources to back up any of your claims. You have no understanding of what you're talking about and can only spew meme buzzwords when asked for specifics. When I post several sources and arguments backing up my claims, you ignore them.

You guys are basically religious retards. Really pathetic.

AGI is never going to happen and it doesnt matter that this makes you upset.

>>15065184
What the fuck did you just fucking say about me, you little bitch? I'll have you know I graduated top of my class in the Navy Seals, and I've been involved in numerous secret raids on Al-Quaeda, and I have over 300 confirmed kills. I am trained in gorilla warfare and I'm the top sniper in the entire US armed forces. You are nothing to me but just another target. I will wipe you the fuck out with precision the likes of which has never been seen before on this Earth, mark my fucking words. You think you can get away with saying that shit to me over the Internet? Think again, fucker. As we speak I am contacting my secret network of spies across the USA and your IP is being traced right now so you better prepare for the storm, maggot. The storm that wipes out the pathetic little thing you call your life. You're fucking dead, kid. I can be anywhere, anytime, and I can kill you in over seven hundred ways, and that's just with my bare hands. Not only am I extensively trained in unarmed combat, but I have access to the entire arsenal of the United States Marine Corps and I will use it to its full extent to wipe your miserable ass off the face of the continent, you little shit. If only you could have known what unholy retribution your little "clever" comment was about to bring down upon you, maybe you would have held your fucking tongue. But you couldn't, you didn't, and now you're paying the price, you goddamn idiot. I will shit fury all over you and you will drown in it. You're fucking dead, kiddo.

>>15065179
There has been an enormous increase in the compute used and the results are not scaling with it.
This is undeniable I have no idea why you guys are pretending otherwise.

>>15065190
Lol

>>15065169
Like this is actually insane. In the post right above your post I linked to watsons abilities 10 years ago.
I have posted evidence in several posts to back up my claim for every claim I've made.
You have not done so a single time.

Post a source showing exponential increase in intelligence. I HAVE posted a source showing exponential increase in compute has not had exponentially increase in intelligence. I did back up my claim.
Why won't you back up yours?

>>15065205
Watson is a publicity stunt and you posted a vague marketing video. We don't know the amount of compute power that went into it and it was never evaluated against standard benchmarks.

>Post a source showing exponential increase in intelligence.
>>15064692

>>15065212

His point is that training is getting exponentially more expensive, I guess if gpt 5 does not achieve singularity a different approach to machine learning might be necessary

>>15065212
Watson was not a publicity stunt, it was an AI that was nearly as intelligent as gpt.

That source is only for the alexnet image classification system, not AI I'm general. In general, AI has had millions of times more compute, but is not millions of times more intelligent. Also, those results have stopped as well..

For an exponential, the increase has to be at least equal to the input, if it isn't then it isn't an exponential. This hasn't happened.

>>15065219
Yea if we need to put in exponentially more compute to get diminishing returns then intelligence explosion isn't possible.
I'm saying that this is not a matter of lack of algorithm or whatever, but that intelligence actually grows as a logarithm and thus agi is literally not possible, because the computers wr have aren't capable of performing the amount of operations that are needed to become generally intelligent.

>>15065225
>Watson was not a publicity stunt, it was an AI that was nearly as intelligent as gpt.
Prove it.

>That source is only for the alexnet image classification system, not AI I'm general.
Read the paper, they evaluate multiple different tasks. Furthermore, it shows that even if you keep the amount of compute power constant algorithm improvements are exponential.

>>15065225

ChatGPT before getting lobotomized was actually quite cool idk man seems like we are actually quite close to the end goal, check deepmind flamingo, it's ability to understand subtle cultural context is quite impressive

https://youtu.be/zOU6usZRJvA

>>15065234
>Read the paper, they evaluate multiple different tasks. Furthermore, it shows that even if you keep the amount of compute power constant algorithm improvements are exponential.

Not him but what I understand is that This is more important for deployment, as seen by current models being able to run on mainstream computers while the training was done on supercomputers but the algorithm improvements are slower than the increase cost of training, we are trying to imitate natural evolution by using power instead of time and seems like we might run out of resources soon

>>15065239
If you use the paid model "text-davinci-003" on the openai playground it is still un-lobotomized, and actually smarter and less censored that chatGPT ever was.

Sad to see the anti humanist death cult known as Lesswrong is deeply rooted in /sci/

>>15065389
For a group that prides itself on rationality, they sure are irrational and stupid lol

>>15065487
>>15065389
I warned you bros, these are religious cultists. Being born into Catholicism, I can recognize when modern ideologies — in this case Lesswrong Singularitarianism — take the form of violent heresy by the way they speak and even the tone they use.

>>15064639
>if i will post le funny frog my opinion will become right

>>15057739
>telling scientists that it is impossible for them to technogically progress any further
How many times do we have to teach you this lesson old man?

>>15066372
This has historically been correct, futurists have never been correct in their claims and technological progress does not improve at the rate that you are pretending it does.

Seriously can you guys give any actual argument that isn't just platitudes and meme buzzwords? How is ai going to exist if the amount of compute being used has doubled every 3 months for 10 years and we've seen no exponential or linear increase in machine intelligence despite this? Intelligence does not increase quickly enough and there is no substrate beyond silicon that we are going to be able to use within the next few decades.