[Nihilated]

Late 2020s. I will make a long story short.

Hardware: With the advances in nanotech, the world's fastest supercomputer will achieve 10^16 calculations by the end of this decade. This matches the processing power of the human brain. By the end of the next decade, computers with 10^16 cps will cost less than $1000.

Software: Real-time scans for each interneuronal connection made possible. We have already mapped out more than 1/2 of the human brain, including the cerebrum, neocortex, etc. There have been ample tests that show that silicon can effectively replace neurons to make it function properly. If we cannot find a software to model this by the 2020s, we will be able to "brute force" our way to AI by mapping every single portion of the brain onto a computer. At the latest, this will be around 2030.

What say everyone?

This post has been edited by Nihilated: 05 May 2008 - 09:28 PM

[mikeblas]

What's your definition of "AI"? depending on the definition, it's something we've already achieved, might achieve in a couple of decades, or might never achieve.

[Martyr2]

You know, we have a bunch of obstacles to AI and not just from the technology standpoint. Regulatory blocks have to be lifted, human ignorance will have to be a thing of the past, governments need to stop being afraid of technology and understand it, and Microsoft needs to be gone.

Then perhaps we have a shot. I estimate that most of this can be accomplished by the 2040s.

[sam_benne]

Well looking at some of the new technology at the moment and the rate that new technology is being developed i say it will be in the next decade or so because some scientist are managing to combine computers with human brains by adding a memory chip to a human brain. So it is possible to get AI its just finding the people to get it going.

http://youtube.com/watch?v=BCNTNKq-LOE

[gabehabe]

Martyr2, on 5 May, 2008 - 10:27 PM, said:

Microsoft needs to be gone.

haha! so true...

[ccrasherdeb]

Technically speaking, we will never have AI. Computers are all based on a binary platform and therefore are only capable of doing what is programmed into them. Computers are restricted to 100% logical , arithmetic operations and can never use any type of emotion to make decisions. Even a system that can learn from itself and process 2^100000000.... power of instructions per second are still restricted to whatever the programmer tells it what to do. It may appear that a system behaves with some degree of AI, but this is not possible now or in the near future. The human brain can still store and retrieve more information and also use this data faster than any computer system. By this I mean retrieve, process both logically and emotionally without anyone telling it what to do.

Think of a computer system as using a form of instinct to do things like many/most living species except for humans and ape species.

[sam_benne]

Actually we can program a human brain to do something like a computer as we can program a computer to react to something like when it has an error we can program it to do something like when a human get something we have taught them but not really noticing that we have but we have taught them to be angry or another emotion. So when we can understand the brain we will be able to copy it. But we can get very close by we can program a robot to act anger when it errors or can't do something.

[Martyr2]

ccrasherdeb, on 7 May, 2008 - 01:15 PM, said:

Technically speaking, we will never have AI. Computers are all based on a binary platform and therefore are only capable of doing what is programmed into them. Computers are restricted to 100% logical , arithmetic operations and can never use any type of emotion to make decisions. Even a system that can learn from itself and process 2^100000000.... power of instructions per second are still restricted to whatever the programmer tells it what to do. It may appear that a system behaves with some degree of AI, but this is not possible now or in the near future. The human brain can still store and retrieve more information and also use this data faster than any computer system. By this I mean retrieve, process both logically and emotionally without anyone telling it what to do.

Think of a computer system as using a form of instinct to do things like many/most living species except for humans and ape species.

The thing is, AI is supposedly going to get the point where it updates its own source code upon mistakes. The first computer with true AI will have the mind of a child and through exposure to stimuli will gradually learn, reprogram ITSELF and use this updated code to then tackle new problems.

Its the idea that humans won't be doing the programming, computers will do it themselves. They will continuously optimize and thus execute instructions faster. I figure after several thousand or more cycles the language which it will program itself in will be something that humans can no longer understand and it will be writing its own computer language, with its own logic and with its instructions.

You are certainly right about the start, but with the right mechanism humans will only need to kick it off (give the computer a basic set of rules by which it will update itself with) and it will then continue on its own. I mean, even Lt Commander Data on STNG updates his own programming to better understand humanity and thus becomes more human.

This post has been edited by Martyr2: 07 May 2008 - 02:52 PM

[KYA]

Martyr2, on 7 May, 2008 - 03:50 PM, said:

The thing is, AI is supposedly going to get the point where it updates its own source code upon mistakes.

Hopefully that never happens in my lifetime

[Sonic88]

No Kidding! I can only imagine what my computer would want to do to me if it had a mind of its own.

[Nihilated]

The only way to go would be to merge with the exponentially improving computers.

[Ambercroft]

Quote

Nihilated
The only way to go would be to merge with the exponentially improving computers.

WELCOME TO THE BORG

[Wardy]

Just thought I would point out that AI is already here and has been for quite some time.

Read this then think about all the individual aspects of what defines intelligence.

Computers do all that.
Also it might be worth noting that the human brain can be compared to a highly complex "chemical" computer that relies on electrical impulses much like every computer out there today.

To answer to the "A computer is based on binary instructions ... ect ect" comment above, think about that then look up "quantum computing" which technically allows a computer to hold mutliple states at the same time hense solving problems that would typically take years in a fraction of a second.

having thought about all that what if we gave a very complex computer program access to the internet and told it to "update itself" based on the best information it found ... has anyone every wrote something that could do that ?

As i understand it the ability to reason / make a decision is based on our experience / wisdom from experiences we have had, technically a computer program connected to the internet would have the experience of every person on the planet that ever added anything to the internet from day 1 would it not ?

What about google bot ?
Is that not intellegent ?

It knows what its doing and dynamically updates its own workload through crawling to new locations, some combination of google bot with a compiler might prove interesting.

[gondal]

Ambercroft, on 17 May, 2008 - 09:11 PM, said:

Quote

Nihilated
The only way to go would be to merge with the exponentially improving computers.

WELCOME TO THE BORG

helo there

as most of us or all of us here have a sound understanding of how software and hardware works i must shre my point of view on AI

AI is all about simulating a human brain by some machine

we all know that a human infers from his/her surroundings and these "inferences are then STORTED in the brain" right!?

when we study AI its about machines with evolving experiances based on certain unique conditions

THE PROBLEM is that human mind has millio of processor working in "PARALLEL" to process that info

like the GRID we hear of these days , but brain is in a more "FERILE or DIVERSE" sense

for computer to be laike brain u need vision = machine vision
also the sense of touch and all human senses

suppose computers get all that and a chip like human brain

but u know what! there comes teh constraint , the limitaiotn of hardware
and with increasingly complex "software"(the AI code) and the ability of that hardware to run that much software(that big interrealted database) with the efficiency of human brain(there comes the constrainst we see these days in prcessing power by a single core)......

as software becomes more complex , the need to process that much interrelated data increases exponentially , so has to hardware to cope...

so i dont think so that AI is possible

this is simple math

more complex sftwre
more complex hardware needed

just suppose how prblms we are facing in moving to 62 bits

to think of 1024 bits or more

the odds are astronimical i think

i hope i covyed my point

[baavgai]

Here's the classic example of how far AI needs to go.

Got to a park and find a bored six year old. Have them gather up all the leaves they can find. From the pile of random leaves, have them organize the piles in to groups, based on which are the same. The child doesn't have to know the name of the trees, but if they had enough brains to preform the initial task for you, they should be able to group the leaves in to ordered piles.

Now, give that pile of leaves to the best AI computer available. Unless the developer is really into trees for some reason, the AI that can preform the simple task of a preschooler doesn't exist.

Speed doesn't matter. We have damn fast computer now and they're very good at well defined decision trees. At this point I can't even beat a moderately advanced MP3 player at chess. However, abstraction, pattern recognition, inference, and a thousand other things that humans take for granted every day have proved far more difficult to reverse engineer into the digital world than anyone expected.

The reason is we really don't understand ourselves well enough to create something that behaves like us. Perhaps it may even be a Zen riddle, like looking at your own face; from where we stand we just can't see it.

I have no doubt we'll one day be able to create things that can behave in a reasonably intelligent fashion. And perhaps even simulate all the elements our perceptions tell us constitutes intelligence. However, if we ever get to a truly thinking machine it will probably be by accident.