AI is just bruteforce calculations with a large dataset and is not the future. True AI cannot exist

AI is just bruteforce calculations with a large dataset and is not the future. True AI cannot exist

Ptove me wrong

Attached: artificial_intelligence_benefits_risk.jpg (1400x1050, 757K)

Other urls found in this thread:

podomatic.com/podcasts/independz/episodes/2018-06-07T19_18_04-07_00
youtube.com/watch?v=fPqbe75_kwM
twitter.com/SFWRedditGifs

If a configuration of atoms can produce consciousness in humans then it can do the same in computers.
Unless ofcourse we are not "true AI" either.

We are a living proof

But you're correct.

Attached: AI.png (697x534, 262K)

are calculators ai

>it hasn't happened yet so it will never happen

Attached: 1521904674691.png (600x600, 122K)

Guys. Guys, guys... What if ...
What if-- what if WE'RE ... the AI?

Attached: 1513542515508.png (312x268, 72K)

>we're totally going to do something that we don't even know what it is

Consciousness is just a collective of analogue algorithms mostly set in stone by your genetics and environmental factors. All AI needs to reach human level consciousness (ie feeling and reacting to disasters, neurosis) is to find those algorithms and implement them.

However I postulate human level consciousness is fucking garbage and nothing AI should be slowed down by.

Attached: 15d.gif (500x382, 125K)

I don't believe your assertion that we are made of algorithms is true.

Reality isn't made of numbers. Numbers are defined in a way which matches reality.

Then explain why a single change in genetic or environmental factor completely changes the consciousness of someone. Exposure to drugs like meth causes permanent damage to it but exposure to healthy diet and exercise also causes beneficial long term side effects. Hell one extra chromosome in your DNA permanently makes you tarded.

True AI can't exist like how we want it to exist.
No matter how many datacenters you toss at it you'll never have anything more than some data and math. If you want something more than math we need to stop chasing AI with the bullshit commodity hardware we have

Cause and effect is not the same as mathematics being reality. You are drawing a false parallel.

I agree with this, raw analogue computing will probably be better at emulating a conscience rather than a crude digital computing working a million times harder to emulate analogue computing.

I never said mathematics to be reality, it is after all just a construct to describe observable events and things. However genetic and environmental factors greatly affect the consciousness giving merit to my postulation that it all works around algorithms in much way solving a rubix cube does.

>If a configuration of atoms can produce consciousness in humans then it can do the same in computers.
No according to Godel.

I under stand that you are trying to demonstrate a parallel between environment/genes and computation. I am saying that you cannot make such a parallel without such intense approximations that your entire argument falls apart.

I also don't reject the conclusion that AI is possible. I do reject your argument that AI is possible for our ability to compute it though.
I think that if AI is possible, then we will not be able to write an algorithm for a computer to be conscious. I find it much more likely that we could build a physical device rather than a computational device to be conscious.

I think you could build a brain, but not program a brain.

so is regular intelligence

Why are CS kiddies so convinced that physics is computable?

>True AI
why do you think that True AI is not bruteforce calculations with a large dataset?

To a guy with a hammer, everything looks like nails.

Ha, this is actually an insightful analogy. Thanks user.

>I think you could build a brain, but not program a brain.
You got me there lad, that's a real doozy in and of itself. Beyond postulations we know nothing of how a consciousness works and even half accurate simulations of it will probably take god knows how long.

Attached: 1520551508256.jpg (512x498, 30K)

>a.i. thread on the consumer electronics board
this is going to be funny

Attached: 3_qt_Popcorn_Bowl_1.jpg (511x405, 100K)

>Ptove me wrong
>Ptove
That's all the proof we need.

>2018
>typoshaming

True AI, fake AI, as long it does the job it does the job.
And in both cases we're talking about a "black box" we don't know how it works exactly.
Nobody knows how the waifu2x scale your art, or how the youtube videos are recommended to you, or how Ryzen predict branches.

Exactly. This is my point. I find it totally possible that we can make conscious machines. Those machines will not be computers in the way we know them today though.

I think you mean
>ptoof
ftfy

Do you have to see 100 million pictures of gorrilas and black people to know the difference?

Even being optimistic, it'll take at least another half century before we can hope to actually have a proper highly detailed understanding of how human intelligence and thought works. Hoping to actually have "True AI" on machines is inevitably even more distant.

that's your genetic code though

No.
AI is a glorified calculator.

A machine can be trained to tell the difference in less than a day with hardware that's 100 times worse than a human brain. A baby would need years.

In this case everything is a nail but people who dont own hammers are salty

>consumer electronics
Sad but true, how can we take en/g/iniggers back?

>namefag
ew

Nope, I'm a tripfag which makes is worse but I respectfully ask that you just add me to your filter as pointing something this obvious brings no merit to the discussion at hand.

>Consciousness is just a collective of analogue algorithms mostly set in stone by your genetics and environmental factors
That's just your guess dude.

>to find those algorithms
Good luck, there are an infinite number of algorithms to sort through.

You should read about the computational physics and numerical analysis. Our best theories for physical simulations involve computing using some sort of finite approximation of reality. The approximations are inherently flawed in that they cannot beat the accuracy which is given by the word-length of a register in memory.

Reality (at all levels. yes even quantum.) is governed (to anyones' best knowledge) by smooth structures. These don't exist in the machines we have today. There is also no indication that we will be able to do this using the modern model of computing in the future.

>a baby would need years
A baby can tell the difference between gorillas and niggers instantly. They literally born with a keen hability to diferenciate faces.
Do you remember those headlines saying "is your baby racist?" those were made because it was discovered that babies can discriminate faces of members of different races as early as two days of being born. And they always feel easy with members of their race since sixth month.

AI is data, algorithms, and computation.
The algorithm that allows us to recognize patterns extremely quickly is still an algorithm, though it may require powerful AI to figure out what that algo is.
"If humans were smart enough to completely understand the mind, then we would be too dumb to understand it"

So two days compared to one day of machine, and with a hardware that's at least 100 times better? Hmmmmm.....

well yes, except that out brains have a chemical component as well. were not sure if we can electrically simulate that the same way a transistor can easily simulate a neuron.

we live in a simulation with a discrete minimum step size.

quantum phenomena happens as a result of running out of significant digits for floating-point computations on the simulation server, as we can see with IEEE 754 on our machines.

if thats true, then its possible that computers will eventually actually do better physics calculations than real life, if they dont already. we might not even think about it that way because we care more about real, practical results, which right now simply means making models that mimic the physical world as 1:1 as possible

>Ptove me wrong
A general AI won't make typos.

Can a machine be organic? can programming be done in a dna'esque way? etc etc

till eventually your computer is a human brain and you are forced to debate what a computer and programming means and no, you dont win that debate by quoting some dude (be it turing, church, whatever)

Peter Norvig is my hero.
And he uses python.

We're made of algorithms. Reality is modeled by numbers because they work. If they didn't then we wouldn't be using them.
Everything about us is algorithmic. DNA transcription, protein expression, cellular communication, respiration, sexual selection, immune response, all the chemical processes that run our body. Our brain is algorithmic too, and we've modeled the shit out of it. Advertising, social engineering, cognitive behavioral therapy, psychological warfare, all take advantage of the quirks in our algorithmic mind.
We're an incredible stack of complex algorithms produced by the algorithmic process of evolution. Algorithms make sense to us because they are us.

A turing complete machine can solve any computable problem. To say AI can't exist is to either say our brains aren't as powerful as what we invision AI to be, or our brains can compute problems which cannot be computed by a turing machine. The thing is, we can't compute problems which are uncomputable by a turing machine. Many mathematicians have tried to do so. Problems like the halting problem or proving a system consistent within itself are simply undecidable. We are not hypercomputers, and if it turned out we were, we'd just end up figuring out how the brain did it, then use the same technique in technology.
We can't though. Hypercomputers cannot exist because they require the execution of infinite steps in finite time.

Our brain isn't special. It's simply not completely understood. What we do know is it's not breaking any physical laws.

Analogue computation has no advantage on digital computation. You're always limited by noise. See: The Shannon-Hartley theorem.

Where did Gödel say this? In fact, his incompleteness theorems prove the above: There are problems which are, and problems which are not computable. Since a turing machine is proven to be able to compute all computable problems, we are turing machines.

You're smoking crack. Reality is not smooth. movements,energy, mass, etc. are bounded by the Plank units derived from the physical constants.

What exactly do you think us humans are?
We're just machines. Organic ones, that have had billions of years of trial and error to reach our current state, but machines nonetheless.

We don't know what causes or constitutes consciousness, much less sentience or sapience. Bacteria aren't conscious, they're basically organic robots. Insects are conscious, but most would say they're not sentient. Reptiles are somewhat sentient, but they're not self aware to the degree that a cat is, and the cat isn't sapient like a human is.

It certainly is possible to achieve some sort of sentience in a computer, but without us knowing what exactly us self-awareness, how are we going to teach computers to feel?

>this has totally never happened before in history
always turned out pretty good right

That noise could be part of how the consciousness works. Perhaps crude numbers are not enough, maybe randomness and inaccuracy is required. It would explain lossy memory storage that tends to break and suddenly work spontaneously, emotions in general, and why despite its performance being unable to do complex arithmetic like division between floating point numbers.

You're totally out of your depth, tripfag, if you don't know that fuzzy logic and monte carlo algorithms exist, and are often used in AI research.

My dude you should do some reading on the general operation of the brain, everything that you said has been narrowed down to a certain mechanism of action.

We don't work at all like a CPU. Our brains are massively parallel, and for example, doing conscious arithmetic is basically like computing a program on a billion layers of abstraction. All kinds of calculus and shit is performed in the vision center, but it's hardwired in and was "programmed" via millions of years of evolution.

The problem with "real" AI is that we don't know enough about how our brains work to have a full understanding as to what brings about the thing we know as "consciousness". Brains weren't manufactured or made to be taken apart, so it's very difficult to figure out how they work.

>Exposure to drugs like meth causes permanent damage
Considering you have a wrong uneducated opinion in the first half of your post, why would I bother to finish ?

I fail to see what it is about organic life that's so unique that it can't be replicated with a computer.

>AI can't exist! I read the intro to an old textbook once, that makes me an expert!!!!
The author of that book, Stuart Russell, has even said that superintelligent AI is possible and even a serious threat.

Attached: 2018-06-09-011217_566x819_scrot.png (566x819, 128K)

To simulate an atom in real time you need the processing power of your average high end desktop. How much does a desktop core weight?
Now multiply the weight of your desktop core by the number of atoms in the Earth. The mass of the silicone we are talking about is already many times more than our solar system. That is just the silicon, we still need matter for the wiring, circuitry, power supply etc. Using quantum computer memery to run the simulation doesn't do much better.

I still can't believe retards think Qcomp's are fast in all regards. The opposite is true, they are very fast at finding one solution in a stack of problems, but completely inept in dealing with anything else. That's why you only see them used in servers.

What moron simulates something at an atom by atom level? Once you know the algorithm you just implement it properly in whatever hardware you have available. You can build specialized hardware if necessary.

Like literally we have evidence that intelligence is at least possible. Probably not in OP, but in most humans.

>Something is going to happen. I don't know what that something is, but it's totally going to happen. Believe me.

>he wrote a what if
>therefore the if will happen!

don't read what-if.xkcd.com

Yes and Einstein wrote "what ifs" about the atomic bomb. He was just joking guys, it can't possibly ever be built...

>that reminds me
podomatic.com/podcasts/independz/episodes/2018-06-07T19_18_04-07_00
this guys shows lately have been shit

wouldnt a consious be a original thought? and computers can only do WHAT THEIR TOLD. so it can NOT produce a consious thought. computers. unless your a total homo faggot. homooes are probbably retards. more people need to be classified as retards or a spectrum of retard

Attached: leftys.jpg (465x960, 72K)

>but
not if quantum physics is right
but its not.
they dont even understand what quantum is
youtube.com/watch?v=fPqbe75_kwM

>dude AI lmao
saged

your jokes arent funny

Yeah, except the machine has to be explicitly programmed to do that, and at the end of the day, it will be the only thing it can do. A baby can recognize and pattern match many things much more easily, and without being instructed to do so.

Then read some neuroscience papers.

>What moron simulates something at an atom by atom level?
People who think we should just brute force brain simulation. Neurons DO fire differently based on atomic-level interactions.

>claims to know the nature of reality

This. We can't even define consciousness yet.

Or the OP could instead provide evidence for their claim.

Transistors also fire differently based on atomic level interactions. Doesn't mean I need a physics simulator to simulate an atari.

Attached: 2018-06-09-034840_789x610_scrot.png (789x610, 97K)

Reinforcement learning and deep neural networks will lead to the development of AGI within 10 years.

Prove me wrong instead.

Transistors are used to create binary logic elements, and the secondary atomic influences are small enough to not matter in most cases.
Also, we know how ataris work.

Neurons are much more sensitive and have way more influencing factors, along with feedback and feed-forward loops and what have you. Read up about neurotransmitters and neurons in general.
I dabbled in neuroscience at my university, and I also have an EE BSc. Brain simulation only seems simple if you haven't tried it. Transistors are a child's toy in comparison.

Attached: 1466515627466.jpg (958x764, 66K)

You miss the point. We don't need to simulate neurons. We can just figure out whatever algorithm the brain is running and execute that. Brain simulations are incredibly primitive. Imagine if the wright brothers tried to build an airplane by making a giant feathered mechanical bird.

The "instruction" comes from their DNA you retard.

>We don't need to simulate neurons.
I never said we needed to. I said some people want to do that.

>We can just figure out whatever algorithm the brain is running and execute that.
"Just". A whole lot of people way smarter than you and me together have tried for decades, and neuroscience is still barely above black magic.
But anyway, I think we're kinda losing the point here. I don't really know what either of us is trying to prove.

My point was machine = need for explicit programming, baby = no need for explicit programming
DNA is not explicit programming.

Attached: 1358658941133.jpg (595x652, 260K)

DNA is explicit programming and if you argue otherwise you're just being intellectually dishonest.

Our brains are good at pattern matching because evolution has added that to our DNA for survival. DNA is to us what code and hardware would be to a true AI.

Then maybe I used the wrong words. What I meant was that programming a computer is an external, human controlled thing, DNA is not.

I agree completely with your second line.

I get where you're coming from now, but I'd argue that eventually we'll get to the point that humans can control DNA of a newly born baby. I don't think just because it's human controlled or controllable means it's not true AI. If someone were to write a true AI, there would eventually be a point where they could leave it alone and let it grow on it's own.

Any AI smart enough to pass the turing test is smart enough to know to fail it

the body is not the soul
the spirit is the soul

it can communicate directly with the creator
ai are just very complex machines
they are missing the direct connection to god

unless you have machines in the future with souls
I would assume this would be a very ungodly thing
and the souls inhabiting them would be very far from god

Attached: Ghost-in-the-Shell-Anime-2.jpg (970x545, 78K)

>I get where you're coming from now, but I'd argue that eventually we'll get to the point that humans can control DNA of a newly born baby
Both "eventually" and "control" are slippery as hell in that sentence. We can technically do rudimentary DNA control already, anyway, but I don't see what that has to do with AI.

How do you define "AI" and "true AI"?

wrong your a faggot go suck penis
prove me that your not wrong in 100 years
hAHAHA..

they only want us afraid of machines
thinking we cant tell what their going to do
>a.i. artificial intelligence or calling computers smart
this dumbass arguement can be debunked byu just showing computers are dumb.

your new car sucks its dumb

Life started as the molecules , who didn't gave up their integrity and instead went to become more complex.
We are result of that.
The reason , we just know many things ,it's because we have millions of instructions hardcoded in us and have a giant set of algorithms built in.
AI can't have this, right now

lmao what the fuck "bruteforce calculations"

also you dont need large datasets doofus look up Bayesian machine learning

>Reality (at all levels. yes even quantum.) is governed (to anyones' best knowledge) by smooth structures.
user, do you even know what the word quantum means?

There is no difference betwen the two.

If it is well defined, then by definition it can be calculated. If it is not well defined, then you're literally just moving the goal posts.

>>Consciousness is just a collective of analogue algorithms mostly set in stone by your genetics and environmental factors
what does the namefag even mean by analogue in this context?

>quantum is smooth structures
Get back to the 19th century, moron.

kek the crux of the whole matter in 1 meme arrow. most people are blind to the extend of our ignorance and blindness and make retarded claims in their small reductionist box.

Attached: 1526913976501.png (734x788, 491K)

Well, have you seen one?
inb4 burger propaganda

You're conflating consciousness with cognition there buddy.
You can have a perfectly healthy body + brain but have it be stuck in a coma

It's like you think specialized hardware has never been made. It could very well turn out that reliably simulating a neuron only takes a couple dozen transistors and capacitors in a specific configuration. Then if you want 86 billion neurons you just make an ASIC cluster of neuron hardware.

This is really going about it the wrong way though. You don't need to simulate neurons to get AI. Neurons are just nature's solution to general computation and signal processing, they aren't special.

As for digitized brains, simulation is also the wrong approach. You're going to get a ship of theseus regardless, so why not replace the ship with an airplane? I'm not saying such a thing would be easy, to accomplish or endure mentally, of course. It's just the most efficient way to do it.