CPUs in 2060 Smarter Than Every Human Brain

[ringfinger]

I agree with seapagen... I can understand where all of you are coming from who don't think it will be possible but how do we really know? Granted it is a badly written article, it does present an idea that most likely at some point in the future of humanity will become fact. Who's to say that computer won't be able to handle parallel processing or even better? In this day and age with technology advancing like it is, I believe anything's possible.

[mark]

The cells in human brains have connections going in many directions to other cells. CPU's aren't built the same way. Humans are born with instinct- to suckle, breath, walk on two legs (yes, it's an instinct), curiosity (watch movement, lights, follow sound with eyes, grab something and put it in the mouth), etc.... A baby's brain is not finished with it's growth. The brain will change, physically, with knowledge. The brain will grow to accomodate the language it learns. The hypocampus will grow larger in people who have to learn a lot of patterns and and manipulate those patterns. The brain changes to accomodate it's environment.

Will computers be able to do this? I don't know.

DL

[ringfinger]

It wouldnt suprise me at all...

[Anonymo]

Yes... you can do all this, but there is so much more to being smart. There's the concept of what is best... can you really trust a computer/machine to decide what is the "best" way to do things?

Anyone who wants a really really really good read about all this stuff (computers being "smart", making decisions, etc) read the Dune saga by Frank Herbert as well as the prequels by his son Brian Herbert and Kevin Anderson. There are now 12 books in all, and the last three that were released (that are actually the beginning of the whole story) dive into this topic a lot. I'm working on the last one and I love every last bit of it. There's always the everlasting question of "what does it mean to be human?"

<{POST_SNAPBACK}>

Yeah, a robot named Erasmus will have us living in pens, experimenting on us and watching us suffer, taking from us our uteri while a fleet of humans will try to overcome this by destroying all thinking machines. The series is very good and the only series I have read completely.

Dune

Dune Messiah

Children of Dune

God Emperor of Dune

Heretics of Dune

Chapterhouse: Dune

House Atreides

House Harkonnen

House Corrino

The Butlerian Jihad

The Machine Crusade

The Battle of Corrin

THE ROAD TO DUNE

[cyprod]

sorry, but I think anybody who says robots can learn, need to actually read up on AI. The number one thing in AI programming is developing tree traversal, if you know what that means, you know why AI really is a joke. And then there's also the problem that computers are purely derterministic. When designing, as my proffesors constantly mention, there are choices and consequences. A deterministic machine is not capable of making different choices based on different consequences for different design constraints. And if you don't mind me throwing in some philosophy, nothing can make anything greater then itself. If you believe that fundamental bit of philosophy, then it's completely impossible for a human being to create something bigger than itself. But even if we do pull designing something greater than ourselves, there's still the problem that design is often a matter of opinion. And the second you put opinions into a machine, well, I don't think I need to finish that statement.

[seapagan]

Everyone is basing this upon current technology, philosophy and knowledge. All of these concepts have changed over the past 50 years (or been proven just plain wrong in cases). Describe an X-box to someone of that era and they will give the same 'it aint possible' responses as for this topic. All knowledge evolves. Just because something is not possible by todays understanding of a problem does not mean that a new way of looking at a concept is not possible.

Glad they finally found out the earth wasn't flat, or i'd have fallen off it years ago.

Lets just wait 50 years more, look this thread up on an archive and see who was right (assuming of course that we are not all breeding slaves for a robot empire.)

SP.

[ringfinger]

(assuming of course that we are not all breeding slaves for a robot empire.)

LOL

[Zxian]

Yeah, a robot named Erasmus will have us living in pens, experimenting on us and watching us suffer, taking from us our uteri while a fleet of humans will try to overcome this by destroying all thinking machines.  The series is very good and the only series I have read completely.

Dune

Dune Messiah

Children of Dune

God Emperor of Dune

Heretics of Dune

Chapterhouse: Dune

House Atreides

House Harkonnen

House Corrino

The Butlerian Jihad

The Machine Crusade

The Battle of Corrin

THE ROAD TO DUNE

<{POST_SNAPBACK}>

Score... someone else who actually understands what I mean when I say some of this stuff... I just finished reading Battle of Corrin (just came out in paperback - hard cover is too expensive).... so GOOD!!!

And if you don't mind me throwing in some philosophy, nothing can make anything greater then itself.  If you believe that fundamental bit of philosophy, then it's completely impossible for a human being to create something bigger than itself.

<{POST_SNAPBACK}>

I don't think anyone minds that at all... it's a good point of discussion as well.

I will however refute it a little bit. We have already made computers that are able to compute far faster than all of us combined (Blue Gene anyone?), so wouldn't that be a bit "greater" than us? It may not have all of our other abilities (parallel thinking, reproduction, etc etc), but it is greater than us in one of them. it all depends on how you define "great".

@DL - Think of instict as biological ROM...

[cyprod]

"We have already made computers that are able to compute far faster than all of us combined (Blue Gene anyone?)"

Actually, I think that's a very arguable point. It's hard to say whether we've made a faster adder than we , ourselves, are capable of. It couldn't be known until we can get our brain to simply do addition and stop all that other "emotion" and "learning" that it constanty does, and compare them on identical tasks.

And to those who argue assumptions are being based on current technogoly, I'm simply stating that 50 years isn't long enough. There's a fundamental problem with learning and problem solving machines that are bounded by the fact that logic is fundamentally sequential and works poorly on parallel platforms, in it's current state. And problem solving is so complex, and our current microprocessors are starting to be limited by simple molecular and physical limitation. As of right now, processors are having a hard time running any faster for sequental instruction streams. If modern sequential processing is too slow to do any real problem solving and problem solving is inherently a sequential problem that doesn't scale well to parallel platforms, this causes a severe problem for developing any sort of decent artificial intellegence. Until someone can figure out how to parallelize logic, no matter what building block logic processor we use, it will be too slow. And if you think why not just parallelize a sequential process, especially one like logic, to a point where multithreading it would actually be useful, then please go ahead and do it, because nobody has the slightest idea as to how to do it for a general case logic problem.

Edit: fixed some really poor grammer that was especially bothering me.

Edited September 19, 2005 by cyprod

[ringfinger]

Wow... very well put man...

[Zxian]

"We have already made computers that are able to compute far faster than all of us combined (Blue Gene anyone?)"

Actually, I think that's a very arguable point.  It's hard to say whether we've made a faster adder than we , ourselves, are capable of.  It couldn't be known until we can get our brain to simply do addition and stop all that other "emotion" and "learning" that it constanty does, and compare them on identical tasks.

<--- That's all I have to say.

And to those who argue assumptions are being based on current technogoly, I'm simply stating that 50 years isn't long enough.  There's a fundamental problem with learning and problem solving machines that are bounded by the fact that logic is fundamentally sequential and works poorly on parallel platforms, in it's current state.  And problem solving is so complex, and our current microprocessors are starting to be limited by simple molecular and physical limitation.  As of right now, processors are having a hard time running any faster for sequental instruction streams.  If modern sequential processing is too slow to do any real problem solving and problem solving is inherently a sequential problem that doesn't scale well to parallel platforms, this causes a severe problem for developing any sort of decent artificial intellegence.  Until someone can figure out how to parallelize logic, no matter what building block logic processor we use, it will be too slow.  And if you think why not just parallelize a sequential process, especially one like logic, to a point where multithreading it would actually be useful, then please go ahead and do it, because nobody has the slightest idea as to how to do it for a general case logic problem.

<{POST_SNAPBACK}>

This here point to consider is the fact that Moore's Law will break down in the next 10-15 years by all guesses and estimates. By the time we hit 2060, chances are our computers will not be based upon silicon. There are many more reasons why silicon is turning out to be a poor choice for the material used in core components (resistance for one -> heat -> bad), but the fact that by the time we hit 30-40 nm cores, the people who make these things start having to deal with quantum effects between the different electrons, which is a very difficult thing to do (do some quantum mechanics if you want to understand this a little better... it's a 4th year physics course at most universities... ).

[mark]

@DL - Think of instict as biological ROM...

<{POST_SNAPBACK}>

Exactly! And this ROM is an intrinsic part of the whole computer. This ROM is manufactured into the computer at the time the computer is manufactured. The ROM is developed selectively and its blue print is stored in the genes. The genes inside a cell, if given the correct environment, turns itself into an organism. All that information is twisted up in microscopic thread. Take what others have said about the design of the chip and the coding that makes it run and compare it to the brain and there are some major hurdles.

If we do manage to make something greater than ourselves, let's hope it's friendly. But the breeding slave part..............

DL

[FthrJACK]

The naysayers are.. to a point right. But wait till Quantum computers start being built, and become more powerfull, and then i think you will be outsmarted

The way Quantum computers will work is much more "brain" like, and with the power there as well... i think we might just render ourselves obsolete.

Edited September 21, 2005 by FthrJACK

[Zxian]

How do you program imagination...?

[FthrJACK]

i expect back when we where using valves, people asked how you could program a weather simulator also.

and imagination is basically trial and error.

something like problem + past experiances + random = idea > functional? yes/no

let me counter that also with.... how did you learn to imagine? its not something you where born with. so you were... programmed, like you where programmed to eat with a knife and fork and why a pink dress with frilly ribbons on really doesent go with your beer gut and tattoos.

no?

Edited September 21, 2005 by FthrJACK