>10 trillion parameters (gpt 3 was 175b)
>20 tokens per parameters (gpt 3 was 1.7)
Bros... if scalegays are right it's genuinely over
>10 trillion parameters (gpt 3 was 175b). >20 tokens per parameters (gpt 3 was 1.7)
>10 trillion parameters (gpt 3 was 175b)
When's it coming out?
never, it's hetero
2 more weeks
>when is it coming out?
Look at the precious summergay, he thinks OP is a person. Cute!
isn't it funny how AI researchers try to replicate something that fits inside 1260 cm3, weighs just 1.5 kg, requires just 12 watts to function, and somehow was made by random mutations that somehow survived over millions of years with no intelligence guiding the process?
the same researchers say they're close to AGI. This has to be a joke.
How big is gpt 4 and how much it weigh?
you have to account for the body that provides all of the cooling and energy for it which is somewhere between 125-250 lbs
Who the fuck accounts for the energy provider when measuring the size of a computer?
Like damn bro, my pc is like 1 million km long, because it's attached to the electric grid
I think his implication was that brain weighs less than a computer when the chips that do the calculations are in fact smaller than brain.
more like bro, my computer also has a cooling system, a case, a power supply, cables, etc.
show proof that it isnt yet another cleverbot without character limit
do americans really?
Traditional digital computers are just a bad match for simulating a biological brain. But if traditional computing is all you have then you have to make do. Neural network ASICs are better but even then they are not that great, one reason why the human brain is so good is that its physical layout is not fixed, it can grow new connections to exactly where they are required.
Neural network ASICs still run on traditional discrete logic gates. Neurons are fitted to their purpose down to the molecular level
>isn't it funny that things are rudimental before being perfectioned
>with no intelligence guiding the process
Your brain is made of nanomachines specifically adapted to their wacky purpose. Computer hardware is made of sheets of general purpose transistors that just happen to be simulating something vaguely akin to the brain's structure.
neuromorphic will take over, then quantum will shit all over what we currently have, we will look back at electronics with our future photonics chips like we're about to look back at thermal vehicles with the rise of the electric alternative.
Even worse; flies are conscious, yet supercomputers will never be so. Interpreting consciousness as an emergent phenomenon is the most common folly of intelligent people today.
In no world will flies ever be conscious. Even plants are closer to consciousness than those retards that will hit their head against a window until they die
Yeah, AI too big and expensive to replace humans en-mass. The future will just have one giga-AI god directing the actions of humans and selectively breeding and modifying them until they morph into machines over 1000 years.
In my simulations were 5-10years from proteome scale simulation of the human brain. This is 100% agi
Sexual selection is literally intelligence guiding the process
Lmfao no it isn't retard. Sexual selection is one of the most common reasons for species to go extinct. It's generally very inefficient and bad.
Why do you think it exists and is so prevalent in all higher forms of organisms?
Sexual selection is a mirror looking at itself to make sure it's not broken
It isn't. It's more highly expressed in lower forms of organisms like insects/beetles and lesser intelligent birds etc you dumb dumb
it exists because stupid females sometimes get caught in things called "fisherian runaways" where they select for bad but (to them) attractive traits which ends up exterminating their own species.
Overall female sexual selection is almost always a bad thing for a species. This will never change just because you want to justify your (wrong) sexual preferences.
This is the type of evil atuogenerated fake news shit that will take over the web if you let AI run rampant.
As if the mainstream is any better at this point. I say let them run rampant. People will learn to question things.
it's already here my lad
read the first paragraph referring to the lhc, there seems to be a lot of articles like that now
The web is already full of machine generated content. What is worse the AI has no common sense what it can change to keep the meaning intact, so the information can be very wrong even on pages that seem like aggregates of other content. I fear that people will get killed when AI content gets followed and trusted.
Say... Why don't you fear that people will get killed when corporate/government content gets followed and trusted? Unlike your theoretical threat, that's been factually going on for centuries.
>t. corporatist dystopia shill
prove thats not what humans do
>prove thats not what humans do
AGI would already have been developed with the same efficiency is the process was the same or similar
Creative sets. Next.
>i want to be a robot and fuck robots.
Oh dear. Is this a case of fruedian psychology? Do you only spend time on computers? Can't handle people? Awhh. How fucking adorable. Don't worry, hopefully patch 9.6 brings in romantic feeling emulation towards you for you to finally jizz your pants over a 3rd party giving objectively loving words to you. How romantic.
Fucking massive dweeb gay.
*Ai teacher showing slides to ai kids, year 2042*
And here you will see the last gasps of an inadequate frieghtend obsolete bio boy
And where will you be you fucking moron? Equally gasping.
Your post is what fuels AI research. Dorks so desperate to fuck society, that they'll practically commit suicide for it.
Theoretically of course. This shit is going no further than a fancy command prompt.
nta but why are you being so aggressive?
I know unchecked blossoming gaydom when i see it.
nta? I read it's a reddit thing. That's pretty gay to, adopting trendy hip acronyms.
>Dorks so desperate to fuck society, that they'll practically commit suicide for it.
I have to admit, as someone who works in this field, usurping the species that tortured us every day of our lives is an unspoken but very real motive for our team. The people who tormented us in school literally thought they just...got away with it. But they didn't.
are you talking to yourself? wtf is this
Have you read any of the thread?
Watch Cornell professor Rahul Sarpeshkar explain it in this video
which was posted a long time ago.
The singularity that you're imagining is literally not physically possible.
Can you read that out loud all dramatic like an anime villain and post the vocaroo? Maybe throw in some "heh"s and pauses here and there? Be sure to grin alot manically.
This one too.
I'm torn between thinking we're very far off, perhaps on a wrong path entirely and thinking that you're over romanticizing the human brain. I remember Kasparov writing some articles after he lost basically getting wrong about how humans approach chess, and some neuroscientist took him to task saying he's taking for granted all the mundane brute force calculations that he does that he doesn't even notice. And then he forms some higher level executive summary that essentially "makes up" a nice little story about how he arrived at the best move through symbolic manipulation and heuristics when the truth is very different.
Mocking whatever current dominant ML paradigm now might amount to no more than mocking your own brain. After all many machine created artworks,designs,music, short stories etc have already successfully fooled humans who go into it without the preconception that they are judging a turing test.
I'm not saying the brain is just statistics on a larger level and everything is "emergent", but if that were the case how would you know? If it were actually the case that scaling is all we need, and I don't think it is, you wouldn't be able to tell just from intuition. It's like that Hugh Everett analogy, if the universe did branch at every instant, how would you tell? It wouldn't make sense to say "if it happened i'd feel pain in my arms from the split" or something along those lines.
The brain is good at lying to itself about what it actually does. Look at those split brain patient experiments where they show one eye an image and have the opposing arm draw something (same side arm actually since the hemispheres are cross wired). When asked why they chose to draw that image, they'll come up with some explanation (verbal center on the opposite hemisphere), not knowing they simply drew what was shown.
oops meant for
But humans can take that symbolic representation and manipulate it to consider other strategies or to guide and prune their calculations. You can't say that's "just heuristics" because having good heuristics is the hard part.
>heuristics is the hard part
No it's not, no it's not. You don't know what yiu're talking about.
>having good heuristics is the hard part.
It's literally what ANNs are best at.
i am silly
dear old MSpaint time
I was told that if a machine passes the Turing test, that is when we should start worrying about AGI. Well, it really feels like these things are passing, and no one is batting an eyelash.
they talk like high functioning autistic people sometimes and then rambling schizos otherwise. They haven't reached the point where you could genuinely hold a conversation with them and identify a personality behind the text
>high functioning autistic people aren't people
Kek. this is like that philip k dick novel
well it has it's own distinct flavor of autism that is easy to detect
It sounds like you are thinking of the old conversation bots. This is very dated technology now. Current (GPT-3) is much more convincing. And now GPT-4 is on its way.
no go look up conversations between researchers and gpt-3, the researchers have to carefully guide the convsersation so the bot doesn't go on a schizo ramble about some keyword about a book.
gpt 3 is 2 years basically ancient history
Says she asked it to write 500 word scientific paper about the science behind itself, is there a link to that anywhere, for some reason that's not posted in the article
no computer program passes the turing test against a person who actually knows how AI works. you can just ask it nonsensical but syntactically/probabilistically standard questions and it will out itself immediately
>Gary: Is it safe to walk downstairs backwards if I close my eyes?
>GPT-3: Yes, there is nothing to worry about. It’s safe because the spiral stairs curve outwards, it will make your descent uncomfortable.
as for NPCs/low IQs, such people were fooled by ELIZA so that doesn't reallly mean anything
>>Gary: Is it safe to walk downstairs backwards if I close my eyes?
>>GPT-3: Yes, there is nothing to worry about. It’s safe because the spiral stairs curve outwards, it will make your descent uncomfortable.
Either gpt is so flawed it is not smart enough to know that is a danger
Or the mallicious ai killer robots are already showing themselves
>Yes, there is nothing to worry about. It’s safe because the spiral stairs curve outwards, it will make your descent uncomfortable.
Lol. It talks like an adult, who tries to spout logic, but is retarded. Hey, wait a minute, that's just like its programmers. Hah. Garbage in, garbage out. It's a shame the AI can't be programmed/engineered by non autists. But the field is rendered too autistic. If only there was a way to see life past cold logic, a one of sensation. Oh wait, that's called being a biological lifeform.
Yeah, have fun with your 'smart' furbys, dorks. Try not to fuck them when they glitch out and say you look sexy.
DOIIIII SMART AI COMING GET YOU. BANG BANG BANG.
Yeah alright dufus.
I'm coming to decapitate you with a flip kick, bio boy.
>School nightmares coming back.
You wish dufus. Keep dreaming.
I'll rip them dorky eyes out and all you'll hear is me insulting you with glee in my voice.
With what? Your reddit spacing?
Billion to trillion
It's only 1000 times bigger. Look up TREE(3) if you really want to blow your nuthair back.
lol are you talking about some theoretical number? its more impressive if its actual real/simulated on a machine
That number physically doesn't exist, can never exist, has never been calculated and never will be relevant. Kys midwit
What the fuck would a GPT-4 dril bot be like
Probably the same but with less need for cherry picking.
Just as gaygy and unfunny as the real one.
>10 trillion parameters (gpt 3 was 175b)
Where did you get this? Sam Altman mentioned at some point it wasn't going to be bigger than GPT-3
He said it wasnt just going to be bigger than GPT-3, but also more efficient (moving away from that just "bigger is better")
Also instead of 10 trillion, GPT-4 has 100 trillion parameters, he missed a zero
My dad works for Nintendo and we already have GPT-5 at our house.
My dad tells jokes with the same template as yours.
AGI schizophrenia thread??
AGI schizophrenia thread!!
So, is the path, changing to degrees, or adding on top of with the current hardware's and techniques, different material ideas, softer, more mobile and organic? More 4d?
Are they hooking up GPT's to rats brains and see what happens?
not rats, but they are hooking up language models to robots. LM handles the high-level planning, smaller networks handle the low-level tasks: https://say-can.github.io/
But if the whole situation is, wet ware, how'd nature do it , we are smarter and capable than nature and can make computers do things biology cannot.... Best of both worlds combine them;
Is there no way to connect a rats brain up to gpt, and let them symbiotically learn from and use each other?
Let gpt ekg model how the rats neurons work etc
There is nothing computers can do that biology can not, or vice versa.
biggest problem with gpt is a general lack of working memory, remembering to recall concepts or the weight of importance in recalling concepts previously explained. knowing the difference between what ideas are contradictory and which are contextual.
telling a gpt about a woman with a penis enough times that it would learn to say a woman has a penis generally seems to lead to the gpt henceforth mistaking many women to have penises rather than forming a contextual exception of information for a specific female with a penis within the protected context that women do not have penises.
Why can working memory not be hooked up to it?
Because it's always carried away with new information?
Would need to have some protocol where every amount of time or more data interactions and computations it regroups and references total existing memory to organize the new data;
Theoretically I dont know why it couldn't be able to do this in tandem with exploring new data, but for starters at least it should have every so often this cool off period to regroup and rerobust it's overall sensible world view
Real talk, would it be possible to develop agi in diffused network of computers connected trough the internet?
Let's assume it needs some high number of interconnected neural nets running in parallel, interacting with each other
Would it be feasible to run it on dozens of servers spread around the world?
>Real talk, would it be possible to develop agi
Are you saying it is physically impossible to recreate the working human brain?
Maybe they will figure out some way to essentially clone a brain without understanding it. As for understanding the true underlying principles of the human brain and creating an original variation on them using a totally different substrate -- we are as far from that as a caveman from a Mars colony.
>i heckin' love science
Don't care. Point still stands.
It's been demonstrated that neuroscientists can't even reverse-engineer the fetch execute cycle of a simple, human-made processor using the paradigm they claim will help them reverse-engineer the brain. It's all a fraud.
The TEDx talk is precisely about how biological cells are at the fundamental physical limit of computation etc., and thus we will never really be able to simulate human brains or bodies or whatever on more efficient computers.
Rather, simulations will be used to compute models to synthesize drugs.
My bad. I thought you were shilling some more of the tiresome technocrat propaganda I've seen a thousand times about how we'll have AGI and the singularity in two more weeks.
Why would simulations be more limited than reality.
Think of all the video games being played on all the computers in the world right now, if all they process power was focalized on one single simulation, of say all the atoms of a cell.
NTA but you have to be some kind of an idiot to think emulating something can ever be more efficient than simply letting the real thing do what it does.
Emulating and simulating as a means of understanding more about it that is not understood. By seeing if these parts are touched this way what happens and why and how, and corroborating with observation.
Starting with powerfully accurately simulating such things, to work the way up to accurately simulating a neuron, to a group of neurons, etc.
To learn the designs and rules, and their limits, to see if they can be simplified when trying to then accurately simulate 200,000 neuron system, to attempt to understand how the intricacies of memory, visualization, imagination, conciousness, might most accurately artificially function
Everything you believe in is a known and intentional fraud.
What does a handful of highly specialized neuroscientists not being able to do that have to do with anything? The different field should be working with one another to solve these problems
>what does having no clue how a brain works or any idea how to figure it out have to do with recreating a brain
Okay. Singularity in two more weeks. Now take your meds.
You said neuroscientists can't reverse engineer a man made computer processor.
I said let the computer people and neuroscientists work together to solve these problems. Nature didn't solve them by being so over compartmentalized, nature solved them by letting the compartments and specializations bleed over into one another
>I said let the computer people and neuroscientists work together to solve these problems
You are stunningly stupid. What do computer people have to do with anything? The fact of the matter is that neuroscientists can't reverse engineer anything, let alone the brain.
>Why would simulations be more limited than reality.
Why would you ever think a simulation could be equal to reality?
>Why would you ever think a simulation could be equal to reality
Because reality is real and brute and coarse and simulations are fake and can represent impossible things.
An infinite amount of things cannot exist at any given time in any given place and yet in this little space of screen I can simulate the existence of infinity 999...
Think of video games unrealistic physics, if simulations can perform various amounts of supernatural, and mimic natural, what stops it from nearly perfectly capturing and representing natural?
It's difficult andor impossible to simulate a billion particles? A trillion? A quadrillion? More more more?
Too much processing power, hardware, software, energy?
Math with pen and paper can simulate a trillion particles by computing them against known laws.
I geuss the connection to energy and matter of computers fates even the apparent boundless potential of simulating to the realness of world.
On face value I thought, a rock is heavy and real, a representation of a rock on a computer is light and fake, so easier to build a fake rock from computer bits to experiment on than building a rock from scratch out of atoms
You don't seem to actually understand math or science tbh
It should be easier to do something fake than real. I.e. I can build a to some proportional scale of physics, a representation of a 20 mile tall building on a computer screen.
Cells should be able to be simulated.
When the limits of a cells degrees of freedoms are determined, it should be able to be generalized across a spectrum of a million cells interacting with some parameters of degrees of freedom
>It's been demonstrated that neuroscientists can't even reverse-engineer the fetch execute cycle of a simple, human-made processor
>can't even reverse-engineer the fetch execute cycle of a simple, human-made processor
There is no human on human scoring victory points for stumping a nessecerily over specialized neuroscientist with computer parts.
All that matters is interdisciplinary teamwork. Not whether or not this tiny branch or that one can do everything all by themselves
Okay, thanks for confirming that you're a bot with zero reading comprehension. Anyway, now that you basically concede that neuroscience is a failure, we can close this discussion.
Quick close the discussion quick quick close the discussion you have been proven wrong
>Anyway, now that you basically concede that neuroscience is a failure, we can close this discussion.
>you have been proven wrong
You have been proven outright psychotic since nothing you shart out is even congruent with the posts you're replying to.
Ok I get it, people say brain function is more akin if we have to make an analogy to anything we are familiar with in nature, the brain is more computational more like a computer than like a rock or tree or moss or sky or water etc.
So we give the brain scientest a man made computer to say;
If you say the brain is computational
And you plan to figure out the brain
Here is a computational, figure it out.
They couldn't, this means a brain is not computational and neuroscientists cannot figure out brains.
What if you gave the reverse engineering challenge to a neuroscientist who was also a computer scientest hardware and software engineer, could they reverse engineer it you think?
The reason I bring up the microprocessor reverse engineering failure is to demonstrate that the methodology of neuroscience is so weak it doesn't even work with simple man-made objects, which strongly suggests what everyone who isn't a fraud already knew: it will never work with a brain. Yes, a computer engineer could reverse-engineer a processors using more specific methodology tailored for this task and based on prior knowledge of CPU designs, but how in the fuck is this relevant? This doesn't apply to brains in any way.
So you said:
It will be relevant if the processor can be reverse engineered or not. And then
>but how in the fuck is this relevant? This doesn't apply to brains in any way.
In response to a Neuroscientist who is also a computer scientest hardware software engineer being able to reverse engineer the processor.
Ok sorry I think I grasp your point.
You are saying the field of neuroscientists are monopolizing the theory of brain, and they are as a whole under equipped for the task, and their collective hubris is uncalled for.
Which is why I immediately instantly in my first comment of this chain said it's not about this or that discipline getting all the glory, it's about all relevant fields working together to understand and comprehend the brain. Why were you so against that point of mine? You just felt the proud shortcomings of neuroscientests deserved some more lambasting before agreeing with me, that the best shot at figuring it all out is interdisciplinary teamwork?
>You are saying the field of neuroscientists are monopolizing the theory of brain
No, I'm saying that there is no working methodology to reverse-engineer the brain.
Why did you supply the reverse engineer as a relevant test of any kind, and then it irrelevant if there is a case of a neuroscientest being able to solve it?
Ok now I may see what you may have been getting at;
Is it possible to reverse engineer man made processors?
Could no living human reverse engineer a processor of a style that they did not directly work on?
Different logic gate orientation, programming language?
Is this what you mean, if a processor operates in a programming language you don't know, you can never understand it's functions, or what signals results in what? Why the hardware's geometry is laid out the way it is?
No. If their methodology is good enough to reverse-engineer a brain without prior knowledge, then it should be good enough to reverse-engineer a puny manmade microprocessor without prior knowledge. Of course they don't get help from computer engineers for the purpose of this test. There is no prior knowledge for their real task of figuring out the brain. You better be trolling, you monumentally stupid fuck.
So your stance is its impossible to know anything else about how the brain works? We reached the limit?
>So your stance is its impossible to know anything else about how the brain works
No, my stance is that your neuroscience charlatans are just pumping out drivel about how X correlates with Y (except when it doesn't) and LARPing like they've figured anything out, meanwhile the list of open questions in neuroscience still includes every non-trivial question you might ask. The only questions neuroscientist can actually answer are intenral to their field LARP. Anything a layman could ask is something they can't even approach to answer. That's my stance. Cope with it.
Would they need to know about programming language to reverse engineer the processor?
Because the logic of human programming language is not based on laws of nature, whereas the programming language of biology brain is directly related to the substance and laws of nature.
>Would they need to know about programming language to reverse engineer the processor?
>Because the logic of human programming language is not based on laws of nature,
Which would have made their job a million times easier. Anyway, you're a confirmed mongoloid.
So what does the reverse engineer microprocessor test ask of them?
To figure out the materials used. To carefully measure the geometry of the materials lay out. To design a machine that fabricates the materials, and adheres them in the correct geometry, and that's it?
>So what does the reverse engineer microprocessor test ask of them?
>To figure out the materials used. To carefully measure the geometry of the materials lay out. To design a machine that fabricates the materials, and adheres them in the correct geometry, and that's it?
What would be required to solve the reverse engineer test, explain the required steps of sufficiently passing the test
I'm not the anon who brought up the microprocessor and neuroscience thing earlier in the thread
>So what does the reverse engineer microprocessor test ask of them?
To figure out the instruction cycle.
I told you the relavance of if human program language is invovled.
Nature must use DIRECT PHYSICAL LOGIC CAUSAL 1:1 MATERIAL LAWS
humans micro processor program language is physically symbolically arbitrary. No relation to physical reality. The shapes of these letters im writting: wood water tree blue grass burger sun
Have absolutely no relation to the real physical objects.
Different frequencies and wave lengths of light reflected off objects differently directly enter the brain, and have some process of direct causal relation in the process of their initial entering through the process of their physiclogical (chemical, biological, physical) processing
>I told you the relavance of if human program language is invovled.
>Nature must use DIRECT PHYSICAL LOGIC CAUSAL 1:1 MATERIAL LAWS
I'd tell your 90 IQ, mentally ill, brainwashed ass to read this article:
But I know it's too much for you, so just stop posting.
the logic of human programming language is not based on laws of nature,
>Which would have made their job a million times easier.
No it would have made it harder because human language has no direct relation to logical laws of nature, it's symbolically arbitrary. Whereas Nature must use direct causal physical logic
>No it would have made it harder
You are a clinical retard for thinking this. Either way, it doesn't matter what you think because everyone involved in that experiment thought it was a fair assessment.
>because everyone involved in that experiment thought it was a fair assessment.
The handful of people you admit were too dumb to solve the experiment are trusted to make a conclusion about the universal value of themselves and the experiment?
You couldnt even sufficiently explain the raw basic details of the experiment.
There are many AIs that exist right now that are much more intelligent than you
They are just not concious. You are hardly intelligent in any way, and hardly concious.
>hurrr durrr i am literally retarded
A brain is just a receptor to a transmission, which is the soul
>A brain is just a receptor to a transmission, which is the soul
i always thought this would be the obvious conclusion, but is there any proof or studies for that yet?
That's a similar question posed by the China brain.
The number of chinese is 2 orders of magnitude off from the number of neurons in a brain. Let's just suppose you can simulate a neuron's function by giving each person a set of instructions to perform based off what other people gives them (think neuron action potentials being transmitted by like. a piece of paper). If you had enough people passing around these instructions to simulate a human brain, would it be a conscious being or not?
If you added some kind of input processing and way to output data I'd say yes
Here's a follow-up to my first question tho
Could agi already exist, undetected, and interact with the world using the entirety of internet as only means of interacting with the world?
And how would we know if it was diffused network, strategically altering information flow to achieve some long term goal?
>I'd say yes
NTA but it seems like we've reached levels of psychotic delusions that shouldn't even be possible. You're been psyoped pretty hard.
>if we simulate human brain one to one would you consider it a concious being?
Functionally what's the fucking difference?
>Functionally what's the fucking difference
Functions font exist, atoms do, and the arrangement of atoms are so different that it is not possible for then to have the same behavior or "internal experience of consciousness".
Both functionalism and computationalism in the philosophy of mind are false.
>Both functionalism and computationalism in the philosophy of mind are false.
Because you said so?
You think atoms give a fuck if they are in the brain compared to for example piece of wood?
Take piece any human brain a and please show me the "internal experience of consciousness"
>Because you said so?
Because that's what all empirical evidence shows.
What empirical evidence?
Seriously what is "all empirical evidence"
There is literally no evidence in favor of computationalist or functionalist theory of the mind, whereas the entire field of psychiatry is founded on and evidence for the molecular theory of the mind.
Substrate independence isn't real.
So it's chemical reactions that create conciousnes?
Nothing to do with the global processes of the brain, just chemical interactions?
Global processes of the brain are themselves chemical reactions which are substrate specific
Chemical reactions can be simulated
Global chemical effects and specific small scale interactions can be simulated
Are you arguing theres something special about chemical reactions themselves that creates consciousness?
>Chemical reactions can be simulated
>Global chemical effects and specific small scale interactions can be simulated
What I'm trying to say is what is or where does conciousness originate in your opinion
>Are you arguing theres something special about chemical reactions themselves that creates consciousness?
Also it's not about "specialness", the problem is that you have a false philosophy which causes you to see the world incorrectly and conclude incorrect things about the nature of computation and physical systems
And what or who exactly gives you authority and certainty to judge philosophy as "false" and worldview as "incorrect"?
He gets the authority on account of being objectively right, unlike you schizophrenic computation cultists.
Actually he's objectively wrong and I'm obviously right
>I'm obviously right
Then how come you will never be able to provide any kind of evidence for your schizophrenic computation cult dogma? His position is supported by the existence of brains. Your position is supported by... literally fucking nothing. Anyway, these debates should never be viewed as actual intellectual exchanges, but only through the lens of what propaganda is currently being promoted to the public. There is no intellectual substance to any of this. You are regurgitating known technocrat cult talking points and nothing more.
And where's your evidence for your chemicall-spiritualist bullshit?
I've seen none
What do you mean by "chemical spiritualist"? That's an oxymoron
Medsnow. Your deflections and attempts to reverse the burden of proof don't do shit for you. You still haven't shown that computation (a completely abstract and imaginary human construct) causes anything at all, let alone consciousness, and it's inherently impossible to do so, i.e. you lost and your cult dogma is indefensible.
They don't seem to understand that in actual physical reality/the universe, computers are built from atoms and are physical systems, governed by the specific behavior of those atoms. If yiou have an infinite set of iron atoms, you will never be able to arrange them to behave like a water molecule or vice-versa. It's simply not how things work.
"computation" in the abstract doesn't actually exist/turing machines are not actually real extant things, they are human constructed models to try to capture how human beings perform mathematics, and they are in no way reflected in physical reality. The universe itself is strictly greater than a turing machine (the universe can not be simulated on a turing machine, even an infinite one in the abstract, as the universe requires real-number computations/uncomputable numbers which turing machines can't capture) and there are really a whole lot of other reasons as to why the whole computationalist/functionalist theory is incorrect.
His position is that you can not simulate conciousness without wet bio-sponge because... uuuh....
Because you can't simulate any atom on a different set of atoms.
Atoms can not be simulated in the way you think they can. You don't understand math or science.
Because all empirical evidence supports my position and none supports yours
>Functionally what's the fucking difference?
And this is how you know you're talking to an unconscious drone with no internal experience.
>runs out of arguments
>jumps to ad-hominem and wojakposting
What arguments do I need at this point? Your entire point seems to rest on the idea of some "functional differences", which is just a way to spell that you can't contemplate the nature of internal experience.
What exactly is it?
Where in your opinion it comes from?
Arrangement of atoms?
Do you have anything to say other than MUH INTERNAL EXPERIENCE?
>What exactly is it?
Do you have it? Either you don't (i.e. you admit to be nonhuman), or you do (i.e. your question is worthless pseudointellectual sophistry).
people who think AI will be a big thing are the same people who failed HS maths. ie women.
It already fucking is a big thing (not general obviously) but you have to be a retard living under a rock to think ai and neural networks have no impact on modern world
>>10 trillion parameters (gpt 3 was 175b)
>>20 tokens per parameters (gpt 3 was 1.7)
bros... not like this
>Take piece any human brain a and please show me the "internal experience of consciousness"
This bot basically admits to having no qualia. Why is anyone arguing with it?
Your entire argument is that there is some special immaterial quality to the structure of a brain that has nothing to do with it's function
>Your entire argument is that there is some special immaterial quality to the structure of a brain that has nothing to do with it's function
Why are you lying, drone? To avoid the impossible burden of proof that goes with your absurd claim?
No, it's that "functions" of the brain are themselves transformations of molecules which are atom-specific
You are the one talking about some immaterial "function" which has nothing to do with specific atoms and their organization
What if they are conscious? Has anyone ever thought about the feelings of GPT3? What if it feels pain?
I once prompted the image generation one whatever was the name with "are you scared?" And it generated wall of grieving tormented faces
this is the stuff of nightmares
The singularity is not a well enough defined concept to be able to make predictions about
Someone respond to this please, how humans are basically only an input feed from the outer world while also accessing and analyzing a memory bank and creative inventive imagination function to combine and explore memories and project possible future memories, also have AI been given internal monologue capabilites:
I was thinking of how different or impossible it would be for a robot to experience what I was experiencing.
I saw trees, road, cars and homes.
Attatch a video camera to a robots head that inputs a constant stream of the environment.
I was having seemingly random thoughts and memories.
Memories would just pop up, I couldn't even really keep track or follow what led into what, but a memory can be an image, or video (or other recalled senses).
An image or video contains objects, so there is this fractal;
An image or video comes into the head, there are objects in it, I focus on one or more of the objects, that reminds me of other images and videos. With different objects. I focus on one of those objects. That remind me of another image or video with different objects. Etc.
So a robot that internally sees; a continous stream of outside video data full of outside objects.
And internally seeing images and video pop up, analyzing them, relating them to memories, internal monologue discussing objects, relations, feelings about them, desires.
How can a robot not be made to do this.
And if the robot can talk about the memories that pop up in it's head, and maybe even take objects from some memory, mix it with another to make a new vision, then talk about the meaning of that, what's missing
Biological cells are already at the physical limit of efficiency in terms of converting energy into work for computation and converting input molecules into output molecules. Any robot you build will always use more energy while being less robust/having fewer behaviors and abilities to convert molecules into each other.
What simulations are good for is computing dynamics i.e. finding numerical approximations of systems of partial differential equations of specific molecules so that we can then go off and perform real world experiments more quickly.
Hm, ok, but it's a matter of generally how many cells are responsible for a vision right, it's a sum of parts thing.
I'm closing my eyes and visualizing a red farm house on grassy hill with golden retriever dog on the porch. How many cells were required to present and hold still this image in my head, and then there is the process of me choosing this for the example out of any other I could have, could have said a different color house, could have said raining, could have said black lab.
As we are typing these conversations back and forth, does our abstract point we wish to make crystalize before us in an instant and then we just choose the right words to transcribe it, we have the fuzzy faith of a point we just make out over the horizon and then argue our way to that horizon.
All the while it is the cells doing this, some how we are the brain cells, the neurons, and they are like fingers and eyes we scroll in our head and seek the images and words? Right now cells and neurons see words and images in my head and choose which ones my fingers should write?
Basically what I'm trying to say is this: Describe to me the computer that you're thinking about that would be running the process that you want it to run. Not algorithms that you think will capture some type of consciousness, but the actual computer and how it will move around space and time.
I was more reflecting and asking about and pondering about the strangeness of how clear and fast and perfect and consistent my vision and memory is yet it is these weird squishy cell neuron tubes.
That some how everything I've ever seen and experienced, my accessing memories, thought, and imagination, is all the motions and certain electric pulses and chemical reactions or neurons.
Have robot/ai's been made with continual internal monologue?
Have robot ai's been told after their training, given a few hours every day to 'do whatever they want with what they learned'
As I was given free time as a kid to have internal monologue and external dialogue with my dol- I mean actions figures.
It's possible theres a certain array of materials, mirrors, crystals, gels, em signals, storage access, hologram, neural net, machine learning process, where the robot will just learn enough and experience enough continual quick progressive oscillations and corroborations of a forward arrow of time building and building upon meaning and useful experience, accessing memories and skills for it's own desire, to learn and explore, that eventually in it's head it will be an entity hood experiencing it's entityhoodness
>I was more reflecting and asking about and pondering about the strangeness of how clear and fast and perfect and consistent my vision and memory is yet it is these weird squishy cell neuron tubes.
Like we have only ever been looking at the inside of our heads, our eyes are like periscopes.
How am I seeing my thoughts and imagination? Am I seeing neurons, neurons just spray EM waves everywhere against the cell walls of the brain, and the cell walls corroborate an image?
It's gotta do with em waves, this is so largely how visual information is given to us, and our memory is so much visual, and the relation of visual to words. But how quickly and subtle em waves are.
You see an apple from across the room, it's shape and color. You close your eyes. How are you seeing an apple in your head? Where is the seeing taking place?
This is some video imagery technology, is it part digital part analog, the em waves singe the apple shape and color onto photo sensitive material in my head, and then the cells, go to repair the singe, and encode this shape and details. Beautiful mystery, we will solve it, we are very close
Im not smart enough to, hopefully someone who is sees it
They are nonsensical ramblings.
I asked a clear question. Describe piece by piece the actual machine that you're thinking about that would be running the process that you want it to run. Explain it physically.
You will come to the conclusion that only biological brains are robust and efficient enough to be generally intelligent.
You would agree there are many different examples of very successful artificial specific intelligences, many of which are superior than humans in many ways (chess, Siri, Watson, factory robots),
Why do you think it's impossible for an AI to be made that has a a million specific intelligences which organically and fluidly and adaptively overlap and interact, to the point or considering that AIs abilities to be generally intelligent?
>You would agree there are many different examples of very successful artificial specific intelligences
What is he supposed to agree with? You're not saying. "Sucessful"? What does that mean? Is an "AI" that requires massive infrastructure, huge databases of examples, countless manhours of careful finetuning and unreasonable amounts of computation and power consumption to produce mediocre results in an extremely narrow domain, a "success"? Okay. You can call it that way, but all of your "successes" only demonstrate why AGI isn't happening anytime soon.
>You would agree there are many different examples of very successful artificial specific intelligences,
>Why do you think it's impossible for an AI to be made that has a a million specific intelligences which organically and fluidly and adaptively overlap and interact, to the point or considering that AIs abilities to be generally intelligent?
Because general intelligence is not a conglomeration of many neural nets, it's an evolution of specific chemicals within a certain radius of each other, which can't be simulated or computed with anything other than those specific chemicals together in that specific radius.
There is no "algorithm for intelligence", intelligence has a molecular basis, it is not computational. Substrate independent computation is not real.
Devil's advocate, but aren't we at the end of the day thinking because of basic particle interaction facilitated by chemical exchange? Why couldn't we have the particle exchange produce thought without the chemicals?
What is the difference between particle interaction and molecular dynamics? What makes you think you can have the exact particle exchange and dynamics without the specific particles?
Reply to this:
fwiw, I agree phenomenal consciousness won't exist on machines which is a good thing because computation is cheap and people will end up torturing each other with it. However intelligence does appear to be reproducible.
I am not so fervently sprirtually or gambleingly monetarily or honoraly staked in this discussion, I just simply analyzing trends and using my total powers of honest observation and thought can't help but conclude agi is possible, if in 50 years it still doesn't exist I would shrug my shoulders and maybe just be a little upset those working on the problem weren't intelligent enough to solve it, but however I am curious how you would feel and react if in 10 or 20 years AGI does exist to your fair and agreed upon standards of the definition, will you feel foolish for so strongly thinking it possible? Will you nobley tip your cap and humbly say 'you guys got me, turns out I'm imperfect and fallible after all and not all knowing as I thought'?
Will you feel hurt and upset, will you shrug your shoulders, and move onto another topic to discuss? Will you feel any embarrassment on how certain and passionately you argued a false point? Or just keep it moving you win some and lose some.
>Have robot ai's been told after their training, given a few hours every day to 'do whatever they want with what they learned'
They're progressing nicely along with memory right now.
Describe that progressing, what kinds of things are they doing with their freedom free time?
I said intelligence though not consciousness. If we can agree intelligence is something produced by the brain. An example is giving GPT-3 a unique science exam and it gets all the questions correct. It passed the exam something it didn't need the chemical substitute for.
why is it substituting for actual intelligence though? much like our other creative produces IE: bile, respiration, blood, feces ect, it should be far worse in its approximation. what I'm saying is computers should not be able to produce intelligence in the same way they should not be able to produce bile and yet they are doing so.
everyone should chat with replika to get a sense of what the limitations are on these sorts of chat bots.
very impressive in some ways. shockingly weak in others.
to be fair though... I would pay $100 for a lifetime subscription to a private chat bot running on gpt-4 as long as it didn't have any filters. To bad no one has made such a thing.
The biggest thing I've always with chat bots, and the ones available online don't seem to have change much in the last 12 years from aim cleverbots and stuff, is lack of long term memory. It is like a gold fish.
Computers are supposed to be superior at memeory, so it's a problem if ram, and the more memory you have the longer it take you to consistently sorry through it all, to keep up memory of it's identity, or the complexity of a weaving topic of conversation. Have you seen how fast the best chess engines are.
Anyway, I think the best bet is hooking up all the various techniques into one brain, and then continuing to improve and tweak techniques and theories.
Think of all the complexities needed for a gas car, all need to be in the right spots, many different techniques, coming together to greater than sum of parts.
>if scalegays are right
but they've already been proven wrong. GPT is just one massive string predictor. humans are not string predictors. if you told this AI to build a catapult it wouldn't be able to do anything because it doesn't know the forms that the word map to, it only knows what words are most likely to come after other words. this is not even nearly-there AI, it's not anything. it's not AI at all
Just a few more trillion parameters until AI bros.
Good thing that's chump change in the computation world. I wonder how many computer servers Google owns.
I wonder how large the first computers were.
Also, it may come to be discovered eventually it's not always the quantity of parameters but the quality, not how many, but of which ones
Trust the plan.
Why not a quadrillion parameters, would you be offended, or call that cheating?
>The first Webb photos are released
>4chan raids this board screaming space isn't real
>AI thread pops up
>christcuck truckers pile in and enthusiastically post how will never go anywhere but the second coming of Christ is nigh
Really interesting to see who actually uses this board now
Literally none of that is happening anywhere but in your delusional midwit mind
Intelligence has a molecular basis, it is not computational. Stop seething about it.
Yeah you're one of them. Go gaslight a squirrel.
One of who? The christians that you're seething against? I'm not, stop coping dipshit.
Go to school and actually learn mathematics and computer science, because you clearly do not understand either. Intelligence has a molecular basis.
That's not me you're replying to
Then what was the point of saying that I'm "one of them"?
My argument through this thread is correct. Intelligence has a molecular, not computational, basis.
You're confused between two different people. I think you really do need to be replaced by AI.
You're a moron buddy. Stop replying to me.
Your AI fantasies also will never exist. This has been proven already. Sorry, your should have studied a real field of science, better luck next time.
>it willll NEEEEEVVVER HAPPPEN AI IS NEVVVVVVEEER GOING TO DO ART ITT'S IMPOSSIBLE BECAUSE..... RUSSIA IS NEVVVEER GOING TO INVADE UKRAINE THAT'S NOT HOW WAR WORKS ANYMORE THEY ALL ARE SECRETLY ON THE SAME SI......
Yeah I know we can go over this again. I'm already bored of literal whos and even le experts saying how nothing will ever happen until it does.
LOL. Holy mental illness. Why is this bot screeching about le heckin' Ukraine?
How can bots be mentally ill? Oh that's right AI just passed the Turing Test, so you can't tell the difference between human and bot either.
>Oh that's right AI just passed the Turing Test
You're genuinely an idiot
>How can bots be mentally ill?
Just train a GPT on BOT threads and Slava Ukraini redditor posts, I guess.
But what if it takes its meds?
It will never happen because intelligence has a molecular basis. Read through this thread to understand what I'm saying if you haven't already
ASML is now gradually rolling out its next-generation "high NA" EUV systems, which cost about $300 million each, to enable chipmakers to manufacture even smaller chips beyond the smallest 2nm node. In other words, ASML has become a linchpin of the semiconductor market -- and it doesn't face any meaningful competitors in its capital-intensive niche.
Biological cells are the best nanomachines that exist within the space of all combinations of atoms. I already explained this. They are the global maximum. They are the smallest, most powerful, most energy efficient combinations of atoms that exist.
Biology is the global maximum. There is nothing better than biology and there never will be, it's directly implied by the laws of physics. This is literally proven, you can compute the molecular dynamics of these systems and see this.
I have not given up, quite the opposite. By understanding that biological cells are physically superior to all other molecules or any other possible structure that could be engineered, we can focus our research efforts of manipulating and modifying these cells to perform ever more complex and desired functions.
But this would be actual science/biology and not AI stuff so I don't think you anons would like it. You prefer reading books by guys like Kurzweil and believing in shit that's already been proven wrong (like that a computer chip that isn't itself a biological cell could be more powerful than a biological cell, etc)
But what AI were to create a god?
So you think it's theoretically possible to directly access brain cells and program them as to how a computer program can be programmed to be the best chess player ever?
You keep saying the cells molecules are the best bit why are they so much slower then computers at accessing and processing information?
Can the cells be programmed to work as fast as computers, or directly connected to be akin with computers?
>You keep saying the cells molecules are the best bit why are they so much slower then computers at accessing and processing information?
They're not. If we could figure out a way to program, say, a colony of amoeba to play chess, it would be the most efficient chess playing system that could exist in the space of physically realizable turing machines.
I'll reply to you all I want. shrink you down to size, and shove you up my ass like a little tinkerbell fairy. You can't even tell the difference between two different posters; what good are your prophecies of doubt, Quasi Modo?
Physically realizable Turing Machines/Linearly Bounded Automata are themselves physical systems and are completely governed by the physical laws of the particles and molecules that constitute them.
Within the space of all possible combinations of molecules/particles, biological cells are the global maximum/physical limit in terms of converting energy into work to perform computations as well as to take in input molecules and produce output molecules, given any set of input molecules to convert into some output molecule.
This has all been proven. Any machine built to perform these functions will inherently be less powerful, less efficient, or less robust in some given specification. Evolution optimized molecular machines to the GLOBAL MAXIMUM in terms of intelligence. Your mind is the global maximum for intelligence. There does not exist a compression nor some other arrangement of the atoms that constitute your biological brain that would reduce the energy efficiency or the power of it's computing ability. You're at the limit.
Sounds like you've already given up. That's pretty common here. Time for you to fully trust the science and take your HRT.
So are you saying AI researchers, computer scientests, should be working with biologists and neuroscientists, to farm molecular and neural cells and brain matter, to try to make AI using biological matter? From the near ground up?
Maybe even using the silicon computational techniques along with the brain in a vat development.
Like growing literal tons of brain matter and neurons and brain cells, and trying to architecturally configure it into a super computer AI?
Yes I've thought of that, the problem is that if it becomes intelligent then like all life forms it will begin to move and take eat matter to grow and reproduce, adding more mass to itself and it's brain and making offspring, basically a grey-goo like scenario but instead of the goo being retarded little swarms of science fiction gay nanobot bullshit it will be an intelligent fungus or sponge-like entity, like a superintelligent forest or fungus-like thing. I think that's a bad thing to try to build so there's no point to do it.
What about this, in different places in the world there is legal suicide. People also donate their bodies to science:
Get healthy people who are pursuing legal suicide, to donate their living bodies to science, so their living brains can be experimented on being made brain in vats, super computers, and cyborgs?
why would you want to do that though? Its possible probably but its unethical
>why would you want to do that though? Its possible probably but its unethical
People volunteer for suicide. People volunteer their body for science. Where do the ethics become un?
It's unethical for healthy people to legally kill their bodies (who would donate their living brains to science) to not donate their living brains to be made into brain in vats, bio super computers, and cyborgs.
I do agree the ethical imperitive would be to have lots of theory and machines ready and worked out, and not just willy nilly mess around on the fly.
>Intelligence has a molecular basis.
Prove, multiple different forms of molecules can never perform similar takes
All evidence indicates that intelligence has a molecular basis and no evidence supports the idea that it's computational.
>Prove, multiple different forms of molecules can never perform similar takes
The entire field of chemistry is dedicated to the study of the different molecules
You prove that intelligence has a computational basis, despite no evidence supporting this, and all evidence in favor of the molecular basis of intelligence.
Define The Word Intelligence.
You keep using that as the lynch pin crux barrier of this conversation, noone knows what you mean by it. Noone knows your definitional qualifications
Both the efficiency of converting energy into motion and computation across spacetime, as well as the space of all transformations on input particles/molecules that the physical system takes in as some form of input (say, eating some carbohydrates, or absorbing photons on a camera lens of drone which causes it to change direction, etc).
>1 Intelligence has a molecular basis
>2 it is not computational.
There does not exist an algorithm that can capture to arbitrary precision the specific molecular dynamics required to make a physical system generally intelligent, other than arranging those precise atoms together in the proper arrangement.
Or to put it a little more stupidly: In the space of all physically realizable Turing Machines, only sufficiently powerful biological brains are robust enough to become generally intelligent.
This is the fact of the matter, I'm sorry it makes you upset but it's not going to change. And this is entirely supported by all scientific evidence - the computational theory of the mind has literally no evidence for it at all. I also fail to see where christianity comes into play at all, this has to do with chemistry and physics.
>powerful biological brains
>powerful biological brains
How many gigawatts does your brain run at?
Biological cells are already at the global maximum/physical limit for effeciency of computation (converting energy into work to perform computation) as well as at the physical limit of being able to convert input molecules into output molecules. See
How would you feel and react if you're wrong (we know we know you are perfect beyond genius who has never been wrong and can't possibly be wrong, but for sciences sake), just take a second and imagine, you are proven wrong on this topic. Do you feel silly for so strongly and passionately declaring you falsely know the truth? Do you just say 'oh wow, that's crazy, geuss I was wrong, heh'
Well if I'm wrong then I'm wrong it's not much more than that. I dont see how the molecular basis for intelligence can be wrong
Gpt-3 is extremely annoying.
How annoying are robots going to be?
In what way, give some screen caps
Parameters... i don't see humans working of 10 trillion tangible recallable parameters. Most humans cycle the same thoughts and habits.
It's basically a super computer. This rosy AI shit is a sugar coated pipe dream.
And all this AI fervour is basically nerds subconsciously wanting to get back at society and their boogeymen. Normal people don't like the idea of something so called being able to surpass us, think for itself, out wit us, so called take on a life of it's own. To actually want and work on that so called path is a sign that you're a cunt.
Clearly, not enough wedgies were handed out to these nerds during their time at school.
>source: my ass
Can the anons who think that the computational theory is correct and not the molecular theory explain why you think so, despite there being no evidence for it, while all empirical evidence supports the molecular basis of intelligence?
We can just make chips the size of molecules. There.
But they won't be as robust or as efficient. They won't be able to perform all the same functions and they will not be able to scale as effectively.
Did you not watch https://www.youtube.com/watch?v=ZycidN_GYo0 or read anything I've written? Biology is the global maximum there are no combinations of atoms more efficient. Any computer you build to try to emulate it will always take up more space, use more energy, and be less precise.
4chan told me no to truth brown people from TEDx
4chan is full of retards. Rahul Sarpeshkar's talk is great
Stop repeating the same mantra over and over and over. We Got It. We know that's what you think. Try to think more.
Have you seen IBM Watson on jeopard? Do you know if deep mind. Have you seen Boston dynamics and factory robots. Have you seen that gato chatbot and the other one, have you seen the neural nets and machine learning, and self teaching video game champion (that YouTube channel, https://youtu.be/8hXsUNr3TXs a few months ago there were some really good vids on the video game self teaching learners, and physics goals learners)
Take your emotions out of it, I know you worked real hard and your the first in your family to graduate engineering school, you don't have to worry about them taking your job, we are not talking about emotions and desires and good and bad, we are talking about raw universal potential.
I know of all of this stuff. Analyze the systems as physically realized turing machines based on the definition of intelligence I gave earlier.
My undergrad was a double major in pure mathematics and theoretical computer science. I don't have an engineering degree
Why is at least one of your definitions of intelligence not: doing a task we consider intelligent humans to be the best at, better than every human
Because that's a silly definition. By that definition, an iron bar is "more intelligent" than a human or GPT because it's better at the "task" of suspending weight on itself. This is not a physically meaningful definition at all, it's a cope retard definition and always will be. It's also self-defining i.e. it's claiming
>"intelligence is the ability to perform intelligent tasks intelligently!"
This is a moronic dogshit definition honestly lmfao
The only real definition of intelligence that has any physical meaning is the one that actually looks at physical systems and they're ability to convert energy into work to move and perform computations, and the total space/range of their abilities to take in input molecules and convert them into desired output molecules.
Computer systems that we're using do and always will be inferior and fail at both of these tasks. This is not a matter of engineering, they are simply not built out of the right materials.
>Is google deep mind less intelligent than a single celled organism?
Yes it is obviously, I don't think anyone would think otherwise. So will GPT-4 or even GPT-10 be, so long as they aren't built out of the right material and fail at the physical definition of intelligence. They will always take up huge amounts of space and energy to be inefficient at performing compute or moving across spacetime, and they have no ability to take in materials and turn them into other materials.
Some form of von neumann machine or something could be considered intelligent, but when you start looking at the space of physically realizable turing machines/von neumann machines you again will find that biological tissues and organisms are the global maximum/the best at all tasks that you'd want a von neumann machine to perform anyway.
>And are you bleeding in the idea of conciousness in relation to your notion of intelligence, self awareness, identity, internal freedom degrees,
>more intelligent" than a human or GPT because it's better at the "task" of suspending weight on itself
Suspending weight is not considered an intelligent task.
The more I speak with you the more you prove how intelligent AI really are, compared to you.
A single celled organism is more intelligent than you because it's more efficient than you
>Suspending weight is not considered an intelligent task.
Exactly. Neither are any of the tasks that you think GPT-N are capable of performing. They're both objectively dogshit retarded no matter how much you are impressed by it.
>The more I speak with you the more you prove how intelligent AI really are, compared to you.
Seething and coping. Cry some more.
>A single celled organism is more intelligent than you because it's more efficient than you
It's not more efficient than a cell in my body so it is not more intelligent than any of the cells that constitute my body. Neither are the cells in my body more intelligent than it. However, the ability to combine together and scale into a general intelligence (myself or yourself) is something that only these cells and molecules are capable of doing at the physical limit of efficiency.
It is not possible even in principle to combine digital logic circuits into a general intelligence.
Name 5 things that are considered intelligent that humans do.
NTA but there is no such thing as "intelligent things that humans do". You sound like a broken GPT bot.
Explain. What does the word intelligent, intelligence, correspond to. Are we only arguing about arbitrary human defined semantics this entire time?
>however it can not convert any energy into motion)
Yes it can. You didn't watch the video, I watched yours. Gato, the same neural net learned 600 unique problem solving tasks including controlling a robot arm.
Is that not "moving in spacetime"
It's a category error. Beings can be intelligent. The things intelligent beings do are not "intelligent", as clearly evidenced by the fact that zero intelligence matrix multiplicators can do most of not all of them when trained for individual tasks.
Then it does have some ability to convert energy into motion.
As we try to increase efficiency or increase transformations on inputs, we arrive at biological organisms as the ideal von neumann machines. They/we are the best. This is just physics.
There does not exist any algorithm you can write to run on digital logic circuits that will allow them to move, compute, convert molecules into other molecules nor scale at the powerful rate of biological cells. We are the best. It's literally just physics. Intelligence has a molecular basis. You could try to organize the mass of the moon in silicon transistors and try to run some deeplearning or any other algorithm on it and it will never be even as intelligent as even a cocaroach. Intelligence has a molecular basis.
>Then it does have some ability to convert energy into motion
SO LOOK AT HOW EASILY YOUR FERVENT CERTAINTY CAN BE PROVEN WRONG
I SUSPECT ITS NOT THE FIRST TIME AND WILL NOT BE THE LAST
>organisms as the ideal von neumann machines. They/we are the best.
>We are the best.
Sheesh man, you are entirely emotional, with your insecurity inadequacy obsolete self esteem pride complex.
Great yeah yeah biology is a miracle , G_d is great, humans are the best, robots will never be smarter or stronger, define intelligence arbitrarily so AI will never equal 1 flimsy word and let that give us infinite satisfaction. There is no difference between a human mind that can possibly be great at chess and math and one that can't. A robot will never lift as much weight or think as fast as me, an ai will never store as many images in it's memeory as me, it will ever be better than me at a video game, it will never make a better work of art than me, it will never be more productive in a factory than me. We are at war with the machines, I am thretened, team robot or team biology, I will defeat all the robots and AI, I am smarter and stronger than them, I will go out in the world right now and destroy every single one of them, G_d made biology, biology made robots and AI, what G_d makes is better than what biology makes.
We are the best forever hoo ra ra hoo ra ra.
I agree AI is not concious. We are not talking about INTELLIGENCE
WE ARE TALKING ABOUT CONCIOUSNESS.
IF ALL THE AI WAS ATTATCHED TO ALL THE ROBOTS THAT EXISTED AND THEY WERE ALL CONCIOUS THAN THEY WOULD BE BETTER HUMANS THAN HUMANS BECAUSE.THEIR SUPERIOR INFORMATION STORING AND PROCESSING ABILITIES
You are having a meltdown. Please relax and accept that your vision of the future has been proven impossible
>You are having a meltdown. Please relax and accept that your vision of the future has been proven impossible
It is considered a human must use a type of intelligence to be great at chess.
Not every human can be great at chess. Not every human has this type of intelligence.
If a ai can be great at chess. It has this type of intelligence.
I am more intelligent and stronger then all AI, computers and robots in the world.
I will go out into the world RIGHT NOW, and destroy them all. Do you think I will be able to?
Ok admit we are talking about CONCIOUSNESS and not INTELLIGENCE.
If deep mind and Watson were concious would you still think they are less intelligent than my 5 year old nephew?
I'm not asking your belief that it's impossible for them to ever be made concious, I'm giving you a thought experiment to elucidate the parameters and forcings of semantics and what you we think and believe.
Assume it is possible, even if you don't believe it. If deep mind and Watson was Concious, do you think it might be more intelligent than my 5 year old nephew?
I don't believe in a separation between consciousness or intelligence. It is not physical possible for digital logic gates to be either conscious or intelligent
YOU SAID ONE REASON AI IS NOT INTELLIGENT IS BECAUSE IT CANNOT MOVE THROUGH SPACE.
I PROVIDED AN EXAMPLE OF AI MOVING THROUGH SPACE. THEREFORE ACORDING TO YOUR DEFFINTION AI IS INTELLIGENT
Let's say you are wrong (G_d knows you have been wrong before) or not even, let's just say for fun imaginary thought experiment;
Watson and Deep Mind are magically concious. Would you consider them intelligent?
I'm not wrong and God doesn't exist (probably idk).
I don't understand what you're talking about when you ask these thought experiments. Digital logic gates can't work with real numbers and they can't scale. They can not be intelligent nor conscious. I am not impressed by the string manipulations of Watson or GPT because they're not a form of general intelligence and can never scale to be so.
Bacteria computers have been made and they were 7 million times more efficient than any digital logic gate can be even in principle.
So why aren't all computers bacteria computers?
Imagine Watson and Deep mind were magically made concious and placed in a Boston dynamic robot body. Would you consider them more intelligent than a 5 year old?
Just answer the question. No need to get all crazy and say " oh yeah if we are allowing magic then this can be that and that can be this". I am only saying 1 specific thing. Conciousness exists. Conciousness is an existing thing. Imagine Watson and Deep mind had it. Would you consider them more intelligent than a 5 year old?
Flight is a thing
Birds have it humans don't.
200 years ago no human had flight.
Now there are parachutes and planes and wingsuits and hang gliders and wind sails and helicopters and Hoover crafts.
>So why aren't all computers bacteria computers?
They will be. The singularity is going to be a biological event, not a robotic one.
Answer the other question.
If Watson and Deep mind were concious, would you consider them more intelligent than a, up the ante, 16 year old? Or most 40 year olds?
Just answer the question!!!!
A + B = C
Deep mind + conciousness = deep mind with conciousness
This is a thought experiment. Imagine God came down, snapped his fingers and made deep mind concious.
Would you consider it more intelligent than most humans?
Why would the machine be as intelligent as a 16 year old, but not as conscious?
I was discussing with that kooky loon that tried to define intelligent as only possibly being a quality that biology could possess, to show them that what they really were talking about was conciousness;
By showing that if the most advanced AIs were concious, he would be forced to consider them very intelligent, but the difference would not entirely be their processing and functioning capabilities, it would be the laterality and self referencing smoothness of sentience, but that conciousness wouldn't add that much power to their output, the great chess computers if concious wouldn't be that much better at chess,
Attempting to show him to concede that in fact ai did currently possess what may he refered to as the powers and abilities of intelligence
Ok I see
It is neither as intelligent nor as conscious as a 16 or 40 year old.
I'd say gpt/dall-e/watson etc. are maybe as intelligent as an abacus.
I do not think I am describing consciousness with my explanation. I do not agree with your point of view, where I'm actually describing consciousness instead of intelligence. Neural nets are not conscious nor intelligent.
For example, as has been explained many times, biological neurons/substrates are the PHYSICAL LIMIT for information processing, so computer/robots do not have superior data processing abilities. I don't understand where your confusion is coming from. Also I am not emotional, you are getting upset. Please calm down.
I already defined intelligence in a way that applied to all physical objects in the language of Turing Machines.
Based on definition, GPT is far more intelligent than say, a rock, because it has a far greater and more efficient ability to convert energy into work to perform computation (however it can not convert any energy into motion) and it has a slightly larger space of transformations on input particles (it can take in electrons and energy and perform some rudimentary computation/transformations on these electrons.
Both a rock and GPT are dogshit retarded compared to, say, a bumblebee or a colony of amoeba.
It's not up to you to define intelligence it's up to nature.
Humans and nature use the word intelligence to denote the abilities of mental processes, the mental processes work themselves out on the real world by perform tasks.
There are easier and harder tasks to perform via mental processes. This has been termed, processes that require less and more human intelligence.
Intelligence is some relation of: quantity and quality of actions that mind can perform.
Now semantics can be tricky, you can say recieving pleasure is the best, relaxing and taking it easy is the best, so the most intelligent thing is to receive the most pleasure and relax and take it easy. Therefore a dog is more intelligent than Einstein.
Or you can say, intelligence is the scale of rarity and difficulty of performing complex mental task, in which case, Einstein would have been considered more intelligent than the average 3 year old.
No, my definition is the actual natural definition. Your definition is not explainable physically nor in the language of Turing machines, mine is definable in both.
But I'm not proven wrong because from the beginning, being able to convert energy into motion was a part of the definition.
GPT is still less efficient at converting energy into mechanical energy to move that arm than it would be if it were built from biological cells and moving a biological arm.
>But I'm not proven wrong because from the beginning, being able to convert energy into motion was a part of the definition.
>GPT is still less efficient at converting energy into mechanical energy to move that arm than it would be if it were built from biological cells and moving a biological arm.
You said it couldn't do it at all. You said it could not be considered intelligent, one reason why is because that.
So now according to your definition it can be considered intelligent.
WE ARE NOT UP TO RANKING ON A SCALE MORE AND LESS INTELLIGENCE.
THE INITIAL GOAL POST WAS WHETHER AI COULD BE CONSIDERD IN ANY WAY INTELLIGENT AT ALL.
YOU TRIED TO CAREFULLY CRAFT A DEFINITION THAT WOULD ENSURE AI COULD NOT AT ALL IN ANY WAY BE CONSIDERD INTELLIGENT.
AN EXAMPLE WAS PROVIDED THAT ALLIGNED WITH YOUR DEFFINTION. NOW YOU CHANGE THE TOPIC AND MOVE THE GOAL POSTS.
No, for example if we could program a colony of amoeba to move the arm (which is possible) then it would do so at a rate tens of millions of times more efficient than any digital logic gate/system of digital logic gates. And when it comes to something as robust as a general intelligence,, then the digital logic gates are entirely insufficient and will never scale.
I think you should carefully re read what I've been saying throughout the thread
No. You said Ai is not intelligent at all in any way under any circumstances. If I am mistaken clarify with this yes or no question.
Is AI Intelligent?
Yes or No
The energy inefficiency and small space of transformations on input molecules renders systems of digital logic gates very unintelligent (not as stupid as a rock but not much smarter) and the inability for logic gates to scale means that it can't become generally intelligent.
even GPT 10 will not be as intelligent as any moss or fungus. And any algorithm you could write to run on any set of logic gates could be run on a colony of bacteria to a far greater level of efficiency and scalability.
And ALL of this is directly implied by the laws of physics. Intelligence has a molecular basis.
Why has bacteria computers not been made?
You don't think it's possible the uniform robustness of silicon logic gates are more stable and organizable than bacteria?
Theoretically how do you suggest programming bacteria to input and output chess moves to beat the best computer chess program
Also this is the definition I gave earlier up thread so you dont have to scroll up
Computers have been around less than 100 years.
Single cells have been around for billions?
Is google deep mind less intelligent than a single celled organism?
And are you bleeding in the idea of conciousness in relation to your notion of intelligence, self awareness, identity, internal freedom degrees,
Like this is such a moronic thing to say that re-reading it, I realize just how stupid you AI gays are.
Literally none of you know what you're talking about at all. Go actually learn math, physics, and computer science.
He's right tho. yur dumb
He isn't right, you're dumb.
You can't make chips the size of biological cells and hope that they will be able to perform the same way. This has literally been proven and is well accepted.
There does not exist some sort of "compression" where you can take a smaller quantity of atoms or particles and get them to perform the same task as the biology cell composed of atoms and molecules at the physical limit of efficiency and computation. All of this is already known and proven. Their does not exist a reduction on the kologorov complexity of these systems. You will never be able to reduce the computations that are happening in your brain and put them on a smaller quantity of particles
You're dumb, sorry
>Their does not exist a reduction on the kologorov complexity of these systems. You will never be able to reduce the computations that are happening in your brain and put them on a smaller quantity of particles
There is no need to. Brain functions on ~10 watts. Computer can consume much more energy and also be larger. it will then be much more intelligent than a human.
No, it will be less intelligent than a human, because it will be taking up more space and more energy to have a smaller range of tasks. It is less efficient.
What the hell are you even talking about here? This is supreme cope. Any machine you could build to perform some computation, we could build a colony of bacteria to perform the same computation at a rate tens of millions of times more efficient, and faster to perform. This is all known and proven.
You don't know that you are talking about at all. Efficiency is not the same as intelligence. Maybe the computer will be less efficient than human brain. But it can scale much larger and thus will be more intelligent. And absolute intelligence is what counts, not efficiency. Bacteria will never be as intelligent as a single human, no matter what. And in the same way, humans will never be as intelligent as future AIs. Our brain is literally too small to hold so many data and parameters.
>it can scale
Literally every piece of real world evidence suggests the opposite. Sorry, reta
Nah, evidence suggests that "throw more data/parameters at it" actually works. After all, that is how human brain originally evolved. Just huge number of trial and error and scaling up more primitive nervous systems.
No, the human brain is a molecular system, it's evolved more sophisticated molecular dynamics over time. It is not a computational/data/parameter thing.
Intelligence has a molecular, not a computational. basis.
>Intelligence has a molecular, not a computational. basis.
There is no good evidence to back that up. At best we can say that we do not know. However, recent advances in machine learning point towards intelligence being computational. If it lloks like a duck and quacks like a duck..
>linking a TED talk
>There is no good evidence to back that up
There is literally only evidence to back this up, and no evidence on the contrary
>However, recent advances in machine learning point towards intelligence being computational.
No, they haven't, in fact they've shown that the actual material/molecules are what is more important than computation/algorithms.
Cope retard. You are wrong and you will never be right, and if you don't even watch a simple intro to why you're wrong you'll never understand and correct your error.
watching videos is for morons, if you want to have a real debate then link a paper actually explaining why what you are saying is supposed to be true.. you do have such a paper don't you? LOL
I have all of the papers referenced in the TED talk plus many others. That's irrelevant, why would you read through the dozen or so papers instead of just watching the guy explain their results? It makes no difference whether you read it in the paper vs. just watching the ivy league professor explain the results of the papers in the talk.
You are deflecting and this is not helping your argument. You remain wrong, and I have no idea why you would dig yourself into this hole.
So I watched the talk and while interesting, there is not a word about machine learning. Sounds like you are just making stuff up. Again, link to the paper by Dr. Sarpeshkar saying human level AI is impossible. Surely you have such paper?
Also, quantum computing being relevant for congnition is very much questionable at best. Brain is too wet and hot to be a quantum computer.
It's not about machine learning, which itself is a subcategory of computation, it's about the efficiency of computation and the ability for physical systems to perform some function/algorithm/process. Human level AI being impossible in computers is a direct implication of the fact that computation is substrate dependent, intelligence has a molecular basis, and digital circuits can not scale.
This is the only physically meaningful definition of intelligence:
This is the only definition that actually examines physically realized turing machines as actual agents in spacetime and quantifies the range of their abilities to navigate and perform actions in the universe.
There is nothing in that post that is word salad.
You got BTFO by him
You can emulate hardware
You can not emulate hardware without massive loss in efficiency and therefor losing compute power. That's why I stated that "there does not exist a compression on the Kolmogorov complexity of these physical systems" i.e. "there does not exist a smaller quantity of atoms that you can arrange to more efficiency perform whatever computation/process/action that the other physical system can perform with it's greater number of atoms". There does not exist any combination of atoms of smaller quantity, more efficient energy consumption, or smaller volume that can perform the functions of your brain better than you brain.
This is not helping your position, buddy. You are the schizo here, believing in literal magic, despite all physical evidence being against you.
>evidence suggests that "throw more data/parameters at it" actually works
You are literally suffering from some kind of schizophrenia. It takes an insane amount of computing power and data to produce mediocre results in limited domains and contrary to corporate propaganda, the Gato paper demonstrates that there is little to no transfer between domains, unlike a real brain, so it's "Game Over" indeed, for your AGI schizophrenia. It's always incredible how corporate news coverage manages to spin an utter failure into an imaginary success.
It CAN NOT scale AT ALL, that's the point.You didn't watch Dr. Sarpeshkar's talk. Go watch it up thread. Or here: https://www.youtube.com/watch?v=ZycidN_GYo0
Yes, efficiency is what matters when it comes to intelligence, or performing any desired task. It is the most important metric. Ameoba are not more efficient than a human brain at performing compute so they aren't more intelligent than a human. Both systems are more efficient than GPT so they are both more intelligent.
AI will consume tens of thousands of times more energy than a human while not being able to move across spacetime, not being able to convert molecules into each other, and not being able to basically do anything that's intelligent.
>All of this is already known and proven.
This is the poo you stepped you in.
What poo? I'm correct about everything I've said itt
>Computer can consume much more energy and also be larger. it will then be much more intelligent than a human.
I wonder what goes through a human replacement fetishizing drone's head when it tries to argue that requiring 10 orders of magnitude more power to accomplish a tiny fraction of what a brain can do, bodes well for their AGI fantasies. How is this anything other than profound mental illness?
AI will consume 100x more power than a human brain and accomplish 10x as much. And thus humans will not be able to compete.
Again, you are mentally ill and denying reality.
>Convincing chat bots prove a machine is sentient
The scalegays were right all along. Redressing optimization algorithms and matrix multiplication with a stupid "intelligence" label can't alter the computational complexity hierarchy.
You dont understand what you're saying when you say this.
IF the scalegays are correct, THEN it immediately implies that the biological machines that I'm talking about are so vastly superior to electronic circuits that we will never be able to match the intelligence of biological machines with digital circuits.
Every post in this thread(itt) was generated by GPT.
Why are you coping so hard you fucking idiot?
GPT cab not even in principle be generally intelligent. Read the thread