Thursday, July 26, 2012

AI, Sci-Fi and Silicon

Fairly recently my buddy and I got into a discussion regarding the future of computing; specifically, Artificial Intelligence (AI). Having grown up with television shows such as (classic) “Star Trek”, I have been amazed at how reality has actually surpassed our fictional imagination regarding computer capabilities. The cell phones we routinely carry with us today far surpass the computational power of the Space Shuttle. But are we on a pathway to someday creating a real “Lt. Cmdr. Data”? Well…

The silicon micro chip and modern science fiction has suggested, perhaps somewhat mythically, that future AI might manifest itself as some type of Silicon Life Form. Unfortunately, as chemists point out, silicon as a basis for a living organism has some major drawbacks compared to carbon. Though silicon has the ability to bond with other atoms, such molecules are relatively unstable. As a result the largest silicon compound ever observed was limited to only six silicon atoms. Carbon, on the other hand, is able to produce long chains of hundreds of thousands of atoms. Large complex chains of carbon atoms have the ability to form “left-handed” versions, which build amino acids, and proteins, and “right-handed” versions which form sugars. Carbon dioxide is a gas which dissolves easily in water, whereas Silicon dioxide is a solid.

Ok, so perhaps then we could conclude that a silicon life form might exist less as a carbon-based “living” organism and instead more like a super sophisticated robot. After all computing technology is increasing inversely to size and energy requirements. Well, not so much – silicon as a substrate is soon to reach its limits of it’s capability as a substrate for micro-processing. Moore’s Law has postulated that we could expect the number of transistors on a substrate to double every couple of years. That has happened to the point where now close to 200 million transistors can be placed on a single microchip resulting in smaller processors requiring less energy. But as size drops down to the 32 to 22 nanometer range, as it will within the next two decades, quantum effects and manufacturing technology limitations hit their bottom limit. For the trend of Moore’s law to continue, silicon must be abandoned.

It is true that silicon based computing has taken us very far. Earlier this year IBM’s Watson, a remarkable computer that demonstrated the ability to learn and reason, had its debut competing against two of the smartest human contestants ever to appear on the TV game show “Jeopardy”. But as sophisticated as was Watson, it was “tasked” with sole purpose of competing on a game show. Without being given a question to ponder, Watson is essentially inert – not capable of creativity, imagination or any of the other emotionally driven neural processes that make the human brain such a powerful generator of abstract ‘thought’. A computer with similar capabilities of the human brain would require a hydro dam amount of electrical energy consumption to even remotely emulate.

What works better than silicon, then? It turns out – Carbon; more specifically the carbon molecules that form the DNA molecule. DNA may be the future of computing technology that takes us beyond the limits of Moore’s Law. DNA is abundant and cheap, self replicating, consumes low amounts of energy; remarkable amounts of data can be processed in amazingly small amounts of physical space.

Though they are still in their infancy, DNA computers have been around since 1994 and have been used to solve some complex mathematical problems. These natural supercomputers already exist in our bodies; research into DNA computing promises to help us unravel both the complexity of the human brain as well as usher us into the newest realm of computer technology. With the epitaph of silicon looming on the horizon, science and industry is already working to build a more compact, efficient and accurate computer which replaces the need for electrical switching with chemical bonding.

But don’t sell your Intel stock just yet.
References and further reading:

Could Life be based on Silicon rather than Carbon?, NASA Astrobiology Institute

Where Is My Silicon-Based Life? Ask a Biogeek

Limitation Of Silicon Based Computation And Future Prospects, Shazia Hassan, Department of Computer science, The Islamia University of Bahawalpur, Punjab, Pakistan

Silicon shrinking will end about 2020, what will replace it?, Geek.Com

Is DNA Surpassing Silicon? George Klington Fernandez, St. Eugene University, Zambia


Infidel753 said...

My computer says it's offended by this posting:-)

The problems with silicon chemistry you discuss are more relevant to the possibility of silicon based quasi-organic life forms, which is really a separate issue from artificial intelligence. Life based on a silicon imitation of carbon chemistry may well be impossible, but silicon chips in computers don't work that way anyway.

The big remaining hurdle to artificial intelligence is reverse-engineering the brain to figure out how it functions and how it generates phenomena like self-awareness and will. We're further along in doing that than most people realize. Once we know how a true artificial consciousness needs to be physically structured, it won't matter what the computing mechanism is made of, just how it's organized (though most likely nanocircuitry is the way to go).

If we decide to give an artificial consciousness a physical body to control, it could be a carbon-based organism, a metal robot, whatever.

The real potential of strong AI isn't to create new intelligent conscious minds (we already have seven billion of them and creating new ones is fairly easy), but integrating them into the organic brain to enhance human intelligence. Stephen Hawking has discussed this point. Once strong AI gets going it will very quickly far exceed human intelligence. We need to make sure it's integrated into our own species rather than becoming, in effect, a rival species.

DJan said...

Fascinating, and I wish I had a comment as erudite as the previous one, but I don't. I appreciate your extensive links.

Robert the Skeptic said...

Infidel I have heard several hypothesis from other scientists involved in AI development voicing their concerns regarding other created intelligent life competing with our own. It brings to mind the demise of the Neanderthals as possibly resulting from pathogens from Cro-Magnon man or just simply competitions from Homo sapiens. What you elude to could well involve the eventual extinction of Homo sapiens by the evolving Homo electronicus.

DJan Indeed, if one can conceptualize it, it's probably already somewhere on the internet.

Beach Bum said...

Believe it or not I have tried to post a comment several times but something always blowup here at house stopping me.

With that out of the way while I freely admit I do not know all the specifics I have heard speculation that as MRI scans continue to become detailed it might one day become possible to create AI from a holographic brain simulation. It is further speculated that these holographic brains while being based on homo sapien neurology would not have the same limitations.

That is the limit to my knowledge on the whole subject except for the fact that some computer scientists do not discount a Skynet situation where an AI or group of them decided to off us carbon based naked primates.

Now I would not be opposed to being a pet to some superior AI as long as I am not neutered.

Grundy said...

It bothered me that Watson won on Jeopardy. It bothered me that Deep Blue won at chess. It really bothers me that computers prowl on-line poker games and continuously profit playing Texas Hold-em mathematically perfect.

But they still can't decipher the CAPTCHA I entered to write this comment. Advantage, humans!

Robert the Skeptic said...

BeachBum Dotto on the neutering thing.. though.. uh, does a vasectomy count?

Grundy One of the advantages I heard that Watson had was the ability to learn that certain 'shapes' represented letters of the alphabet. So the CAPTCHA may not be long as the barrier between AI and human intelligence.

Anonymous said...

Maybe these DNA computers will save us from ourselves?

Robert the Skeptic said...

Dissenter More likely they will hasten our extinction as a species, some believe.

Heidrun Khokhar, KleinsteMotte said...

Once the reverse brain function is better understood we may be able to create a much more efficient system of education so that our unused brain portions work at a higher intellect and then we may not want or need AI.
We have never reached the human potential because we have yet to know how to stimulate many parts yet untouched.

Heidrun Khokhar, KleinsteMotte said...

Once we have the understanding of how the brain works we will likely be able to learn and use much more of it. Perhaps AI will not be wanted once we function at our true full capacity?

joared said...

Maybe the "unused" portions of our brain are actually being used in some manner, but our technology and/or science has not yet developed to the degree of being able to decipher or detect that happening.

I'll not welcome being content to be the "pet" of some electronic creation.

Entre Nous said...

I have embraced the change - after a century of avoiding.... Taking a class in computer programming in October. I'm hoping it doesn't change the synapse in my brain or something weird........

Robert the Skeptic said...

Apologies to my followers, but clearly I have been losing momentum (motivation) to continue to plug away at blogging... I feel like a feeble voice lost in a hurricane. Ideas come to me, I sit at the keyboard then wonder "why bother?" I appreciate your comments and it warms my heart that people still follow me. I have actually stopped following some of your blogs (mostly the political and non-theist) simply because I am so disheartened. For that I am sorry.

Heldrun I wasn't sure if your comment was a duplicate/correction so I didn't delete either one, will let you make that call.

No, I think humans will crave AI if for no other reason, because we can.

Joared The fact that our unused portions of our brains are actually in constant use has been a theme throughout my blog. What we term cognition is mostly a concerted practice of suppressing the huge amount of information our senses deliver to us each second. It's why we don't remember every minute of every day, for example; why we can concentrate or manage impulse control.

On thing to consider: If you every watch the TV game show "Jeopardy" you can appreciate how the answer just pops into your head; normally you don't ponder and get the answer nor do you sift through a pile of data to get to it. This is NOT how computers normally work and why computers rarely make mistakes. The human brain constantly makes mistakes it is surprising that the capacity for making errors is one of the things that will be an impediment to true AI. Mistakes are crucial for abstract thinking and creativity... functions still out of the grasp of AI.

Entre Nois Actually, learning computer programming WILL INDEED change the synapses in your brain which is a GOOD thing. It was mistakenly thought that the brain is hard-wired at birth but now we know that the continual creation of new synaptic pathways is paramount for brain health. Learning new skills, especially things that require hand-eye coordination are excellent. Good for you - it will keep your brain healthy and active!

joared said...

Yes, addressing attention issues, tapping into cortical information storage areas can be part of what is required with interventions I provide some individuals.

As for your waning blogging interest -- that seems to be a fluctuating pattern for each of us. Take a break, then maybe you'll want to resume, post less frequently or more -- or whatever.

Antares Cryptos said...

Good post, but I am biased.
Centuries away from creating a self-aware "android", but the research is fascinating.

Saw your comment above, I commiserate, but do hope you do not give up the blog. Skepticism can be applied to any topic after all.;)

merlen hogg said...

Hi Robert,
I loved reading this piece! Well written!

Merlen Hogg