Have you ever wished that your CPU was way more powerful and that it could just work with any software and do any task you/it needed it to do. Well maybe building a CPU that could adapt and rebuild its architecture for certain needs and be able to actually learn and learn automatically, and how we have instincts to do certain things maybe you computer could automatically shut off communication and be an antivirus in other ways or have specially designed "instincts" for your computer's hardware (and software), than maybe a CPU as powerful as the human brain is the thing to build!

Now let's talk logicly:
How...
Maybe build a new type of logic gate that can be multiple types at different times like an AND gate at one time and a OR or NAND at another, then maybe put thousands, millions of these in a matrix and can switch/rebuild the CPU architecture.

And for RAM it can have a new type of RAM that stores what's needed currently and as new data comes more memory used it can send old unused (anyway currently) data to permanent storage (and to be clear, the RAM will also be permanent (to some degree)) and when needed to retrieve it will read the old (now needed) data.

And how to actually build this... I have no idea, but I believe that if we try hard enough (or if not us, somebody) tries hard enough no matter the difficulty and no matter how many tries it takes, then I believe that it is possible, and that is my gole

I don't even know if this is the right place to talk about it

 

"In the human brain, some 86 billion neurons form 100 trillion connections to each other..."

We're getting close to the requisite complexity: https://cerebras.ai/product-chip/

 

The human brain certainly is complex, but it is also almost completely useless as a model for a Von Neumann machine.

There is no real direction provided by your post for any innovation. How would adaptive logic gates help improve things over current computers? The idea of shifting data resident in RAM to disk is currently used and is probably considerably older than you are.

It's nice to see you thinking in terms of innovation, but you need the prerequisite background in the field you are targeting before your ideas will make sense. This is very much like your "50G antenna" post—a response to a very shallow understanding of components of a system and no solid idea of their function or connections.

Read about computer architectures and then start thinking about improving them.

 

Google NPU.

 

what kind of human brain? because even defective ICs can easily match or exceed brain power of many humans. not to mention brains that are old enough, say 2-3 centuries...

 

As SW noted, you likely need a very large neural network type processor to even begin to emulate the processing power of the human brain.

 

what kind of human brain?

There's more than one kind?

because even defective ICs can easily match or exceed brain power of many humans.

Only those that are dead.

 

brain can be alive but incapacitated. and they are not same (just like fingerprints). am not ware of ANY human that can use ALL software. i would be pretty much useless with just about any of the software that i use daily if GUI interface was to change to Chinese or Korean or Arabic.... so when one tries to start a project, it would be a good idea to define parameters or constraints, preferably with some milestones. the objective as defined so far is to develop something that exceeds COMBINED brain power of ALL humans on this planet - dead or alive. that is rather tall order... and it will probably be done not by humans but one of the next AI generations.

 

what kind of human brain?

Abbey Normal.

 

LOL... such a classic.

 

A neural network is an approximation of how we think brains work ,
So our artificial neurons are an approximation , it seems to model what we know.
A few years ago, we knew earth was at centre of solar system / galaxy , and we knew continents did not move. It was a model that fitted , till it did not .
Our model of a brain is the same ..
Not least, one thing that always amazes me about babies ,is they want to learn, how do they get that ?
If we created a brain , it should want to learn without us having to program it ..

 

Have you ever wished that your CPU was way more powerful and that it could just work with any software and do any task you/it needed it to do. Well maybe building a CPU that could adapt and rebuild its architecture for certain needs and be able to actually learn and learn automatically, and how we have instincts to do certain things maybe you computer could automatically shut off communication and be an antivirus in other ways or have specially designed "instincts" for your computer's hardware (and software), than maybe a CPU as powerful as the human brain is the thing to build!

Now let's talk logicly:
How...
Maybe build a new type of logic gate that can be multiple types at different times like an AND gate at one time and a OR or NAND at another, then maybe put thousands, millions of these in a matrix and can switch/rebuild the CPU architecture.

And for RAM it can have a new type of RAM that stores what's needed currently and as new data comes more memory used it can send old unused (anyway currently) data to permanent storage (and to be clear, the RAM will also be permanent (to some degree)) and when needed to retrieve it will read the old (now needed) data.

And how to actually build this... I have no idea, but I believe that if we try hard enough (or if not us, somebody) tries hard enough no matter the difficulty and no matter how many tries it takes, then I believe that it is possible, and that is my gole

I don't even know if this is the right place to talk about it

Hi,

There is a lot that can be said about this. In fact, probably too much to take in within a single thread. It takes time, money, and knowhow along with an immense dedication to the project.

To start, you could probably build your own CPU to do some very complex tasks, but it's going to take a lot of time and a LOT of money to build because that CPU will have to be made from a LOT of smaller CPU's like a supercomputer. So maybe if you have years to dedicate to the project and millions of dollars to spend, you can probably do it. Funny thing is, that's the easy part!
Next, you have to train it. That's where the time consumption really comes in. You may have to train it for the rest of your life.
If you don't need it to be fast, you can use a currently made multiple core CPU. Even an 8 core CPU would work, but it would be much slower than having thousands of cores. But then there's that training period again, which could take years.

I considered something sort of like this but did not want to focus on 'ai' I just wanted to be able to do a hell of a lot of parallel processing much like a super computer. It would also be a novel device and show off some of the newer micro computers. The idea was to pack a LOT of micro computers into one box, maybe a full tower case or maybe larger, then connect them all on a single network so they can communicate.
The problem was, even though the micro computers are much less expensive than regular computers, the cost would still be very ,very high. I'd run into thousands of dollars.
Then after all is said and done, there's the end of life problem. The entire project life would probably only be maybe 5 years or so, and after that there would be cheaper chips that can do the same thing. That would be a lot of money down the drain more or less.

To do something like you are talking about (although some aspects are already in place) it would take a team of experts working together. These things usually are not done by just one person working alone, it's just too much work and takes too much time, and you need some sort of sponsor. You would have to convince a company that could put up the financing to purchase the needed hardware that you could actually do this and be successful. I considered going to a Chinese company I communicated with, but it's hard to ask them for 100 or 500 or more computers to be shipped to my address. They would have to be damn sure it was going to pull in more business for them.

 

Hi,

There is a lot that can be said about this. In fact, probably too much to take in within a single thread. It takes time, money, and knowhow along with an immense dedication to the project.

To start, you could probably build your own CPU to do some very complex tasks, but it's going to take a lot of time and a LOT of money to build because that CPU will have to be made from a LOT of smaller CPU's like a supercomputer. So maybe if you have years to dedicate to the project and millions of dollars to spend, you can probably do it. Funny thing is, that's the easy part!
Next, you have to train it. That's where the time consumption really comes in. You may have to train it for the rest of your life.
If you don't need it to be fast, you can use a currently made multiple core CPU. Even an 8 core CPU would work, but it would be much slower than having thousands of cores. But then there's that training period again, which could take years.

I considered something sort of like this but did not want to focus on 'ai' I just wanted to be able to do a hell of a lot of parallel processing much like a super computer. It would also be a novel device and show off some of the newer micro computers. The idea was to pack a LOT of micro computers into one box, maybe a full tower case or maybe larger, then connect them all on a single network so they can communicate.
The problem was, even though the micro computers are much less expensive than regular computers, the cost would still be very ,very high. I'd run into thousands of dollars.
Then after all is said and done, there's the end of life problem. The entire project life would probably only be maybe 5 years or so, and after that there would be cheaper chips that can do the same thing. That would be a lot of money down the drain more or less.

To do something like you are talking about (although some aspects are already in place) it would take a team of experts working together. These things usually are not done by just one person working alone, it's just too much work and takes too much time, and you need some sort of sponsor. You would have to convince a company that could put up the financing to purchase the needed hardware that you could actually do this and be successful. I considered going to a Chinese company I communicated with, but it's hard to ask them for 100 or 500 or more computers to be shipped to my address. They would have to be damn sure it was going to pull in more business for them.

What you said about training it, yes just one person training it would be a lot of time but if it is designed to have pre built parts that it has automatically learned that the factory added to do certain things like a humanoid robot using this can be designed to work with the hardware and to walk, talk, move, etc maybe that would take a day of learning off of the learning and it can want to learn want to learn and understand more and will note EVERYTHING it views as important or thinks it should save, witch would be a lot, like what 10TB a day, and to last 10 YEARS! I will definitely make my own type of file that is very small but can store a lot

And what you said about AI MrAI, I'm not building this just for the AI part but also for the rest the unlimited storage, rebuild to be better and the insane processing that it will handle not just AI.

And it will still have programming, as no computer (even this one) can run without programming or OS.


And no, I will NOT be using a 8 core CPU instead 25-70 cores depending on the model (assuming I make multiple)

I will name this Z Core 290 or NeuroTech Z Core 290 to make it sound more futureistic

 

What you said about training it, yes just one person training it would be a lot of time but if it is designed to have pre built parts that it has automatically learned that the factory added to do certain things like a humanoid robot using this can be designed to work with the hardware and to walk, talk, move, etc maybe that would take a day of learning off of the learning and it can want to learn want to learn and understand more and will note EVERYTHING it views as important or thinks it should save, witch would be a lot, like what 10TB a day, and to last 10 YEARS! I will definitely make my own type of file that is very small but can store a lot

And what you said about AI MrAI, I'm not building this just for the AI part but also for the rest the unlimited storage, rebuild to be better and the insane processing that it will handle not just AI.

And it will still have programming, as no computer (even this one) can run without programming or OS.


And no, I will NOT be using a 8 core CPU instead 25-70 cores depending on the model (assuming I make multiple)

I will name this Z Core 290 or NeuroTech Z Core 290 to make it sound more futureistic

I'm not trying to insult you, but nothing in this post makes any sense. It's so far from sensible that, as Wolfgang Pauli said, "It's not even wrong". You really need to consider getting a grounding in the fundamentals of computer architecture and systems design before you start inventing things that have no connection to... reality.

Seriously, this is not meant to belittle or insult, but there is nothing to refute in what you say, it doesn't conect to anything.

 

What you said about training it, yes just one person training it would be a lot of time but if it is designed to have pre built parts that it has automatically learned that the factory added to do certain things like a humanoid robot using this can be designed to work with the hardware and to walk, talk, move, etc maybe that would take a day of learning off of the learning and it can want to learn want to learn and understand more and will note EVERYTHING it views as important or thinks it should save, witch would be a lot, like what 10TB a day, and to last 10 YEARS! I will definitely make my own type of file that is very small but can store a lot

And what you said about AI MrAI, I'm not building this just for the AI part but also for the rest the unlimited storage, rebuild to be better and the insane processing that it will handle not just AI.

And it will still have programming, as no computer (even this one) can run without programming or OS.


And no, I will NOT be using a 8 core CPU instead 25-70 cores depending on the model (assuming I make multiple)

I will name this Z Core 290 or NeuroTech Z Core 290 to make it sound more futureistic

Hello again,

Well, I am not sure I understand everything you are talking about there or what you real goal is, but I'd love to see the end product when it is finished, if you actually do go ahead with this.

I do have a note about the number of cores. The Threadripper series CPU's go up high in the number of cores and threads. You can get a 64 core CPU for under $5k USD.

I also have a question.

I will definitely make my own type of file that is very small but can store a lot

Isn't that just a compressed file, which uses some form of compression to store more data than the number of bytes used normally would? If not, what kind of file are you talking about?

 

Sounds ambitious, but it could be a game changer. The idea of a CPU that adapts is cool, but who knows how hard it would be to implement? Those logic gates and RAM concepts are wild. It might take ages and a lot of brainpower. Not sure if it’s all feasible, but dreaming big, I guess?

 

Neural Nets don't work the same way as a computer, which means it is gonna be very hard to duplicate the function of a human brain. Personally I believe it's just a matter of time but we are not even close yet.

 

We don't have a good way to measure the power of the human brain. We can't agree on how to measure or how it works.

Here are the brains of two Flat-Earthers.

 

We don't have a good way to measure the power of the human brain. We can't agree on how to measure or how it works.

Here are the brains of two Flat-Earthers.
View attachment 340103

Hey, no porno pictures allowed in this forum (ha ha ha)

 

We don't have a good way to measure the power of the human brain. We can't agree on how to measure or how it works.

Here are the brains of two Flat-Earthers.
View attachment 340103

When it comes to biology, the biggest thing I learned is how little we know about the how the body and mind work and what they are as physical objects.

For example, this message is being transferred from my mind to yours, is this purely physical or electromagnetic or something else?

In my opinion, asking how the body or mind came to be is akin to asking how the universe came to be. The complexity of our being cannot be easily separated from the complexity of existence itself without serious consideration. This is a discussion of it's own with many facets.

From my read of the scientific literature in this area, all computational models like neural networks barely scratch the surface of what's actually going on inside us. As an example, there are countless compounds in the body that affect more than one organ. In other words, we are being modulated by forces which we haven't even discovered yet.

As per the topic title, it's more like a super computer or the internet is a very specific aspect of our shared cognitive ability (what I call the hive mind) that's targeted towards solving a deterministic set of problems (Turing machines solving Turing problems).

I don't think it's reasonable to take this methodology and try to reverse engineer it into a living organism like ourselves. In fact, I'm convinced this is exactly why we have never been able to create life. We can only emulate parts of it.

And even if we could, we run into brutal logical problems like this: If I were to clone you completely, would the entity be alive as you are? Not only is this not currently possible, but we don't seem to have any way to tell even if it was.

How then do we go about defining what YOU are? Sure you are homo sapien but that's not all you are. You are one of many in a sea of others and it's not at all obvious where you end and the other begins.

More often than not I see people discuss physics without entertaining the phenomenological aspects of existence like this. They take seem to take what we "know" as core truth and worse, they apply it to what we don't know.

That's what truly troubles me about the scientific enterprise. As a whole it has the ability to make things into what they are not by virtue of defining what's "logical" and "possible". I for one am very hesitate to form any conclusions of the sort! It's very humbling to recognize how little we actually know.

That's my rant for the day.