Future Tech ... on nanobots and the human soul in the 26th Century

POSTED BY: Karnejj
UPDATED: Thursday, January 5, 2006 21:49
VIEWED: 13373
PAGE 2 of 8

PREV PAGE 1 2 3 4 5 6 7 8 NEXT PAGE

CITIZEN

Sunday, January 1, 2006 11:08 PM

The Magic comment was basically trying to say it was further in the future than you seem to imagine. Would, with the possible exception of the walkie talkie, your examples actually appear to be all that fantastical. They had Prosthetic limbs and the basic building blocks of the technology. Now if a cloud of Nanites suddenly formed in front of them...

Quote:
Originally posted by Karnejj:
Links

I see nothing there which couldn't be considered science fiction. These robots require miniature gearing and motors which are far from current capabilities. The prototype is 5cm, which doesn't seem to be exactly Nano scale

. It's an interesting concept, yet a concept is all it is, it's not much more developed than the Alcubierre warp.

Quote:
Eeek ... a squirrel more sentient than Star Trek's "Data" ... that's

Erm, I don't know how to break this too you, erm, well best to come right out and type it, Data isn't real, he's a guy who wears too much make-up

.

Quote:
It seems that you agree that computers can simulate this in software ... why discount that it can be done physically, as well (not that I believe it's strictly necessary, ... but for the sake of argument).

Actually I discount it as anything more than a simulation, for similar reasons as I discount the wooden Automata of the 19th century as being anything more than a simulation of life. Some of the Animatronics we have around today are incredibly life-like, am I too believe they experience life the same as the animals they are modelled after?

Quote:
There are what ... 4 billion devices connected to the internet. What if we add a program to each of these computers that allows them to communicate in a way that more closely models the way neurons work? This is still a far cry from 100 billion neurons and the quantum effects therein (not to mention lag throughout the internet), but it would possibly allow us enough processing to simulate the personality of a human.

What you'd have is a large and complex Neural Net, which just like software driven ones would share some of the properties of the Human brain, such as learning and pattern recognition, but nothing besides.

Quote:
Ahhh... but why can't billions of processors be lumped together similar to the brain. One of today's most complex computers is the SGI Altix which is powered by a total of 10240 Intel® Itanium® 2 processors. And these are full blown CPUs.

The Atrix is a cluster server which is actually more analogous with a group of computers linked via a network. Also it is actually the CPU which is the computer; the rest of the machine is the box it sits in. The CPUs are not linked anything like the way Neurons are, they can't share working memory, or interoperate. Given modern computing architecture the CPU's will operate independently on separate threads, not together.

Quote:
If we were simply modelling neurons, they'd only have to be able to do two calculations (one being: "fire or not fire") and it would have to store up to about 1000 or so addresses of neighbors which it sends a signal to if it does fire.

This shows one of the major differences between a Neuron and a CPU. Firstly the signals passed between neurons are Analogue and weighted depending on which dendrites received the initial signal. The strength of the incoming signal is also taken into account, both when deciding whether or not to fire but what strength the resulting signal should have. Add to this that Neurons have direct links as well as indirect, which allow the generation and propagation of action potentials.

A Neuron can also be linked to tens of thousands of other Neurons, with the average being around 7000, In comparison the SGI Altrix has no interconnectivity or interoperability between either the CPU's of a single node, or the node's themselves.

Quote:
These processors could be made quite small and would certainly be subject to quantum effects as well. And that's only today's technology.

A CPU to

KARNEJJ

Monday, January 2, 2006 1:04 AM

I suppose we just differ in what we believe is possible.

On the subject of nanobots, well, the concepts are there, and in 500 years, I would like to think that the field would be heavily researched. There are quite a large number of obstacles to overcome, so my guesses here are pretty shaky. However, onto something I do feel more convicted about ....

I personally think compelling unprogrammed behavior can emerge from a complex (but quite acheivable) software program. And that "Data" will one day exist (and not just be a guy with make-up ...). However, you seem to discount totally the possibility of sentience from software alone, so let's stick to what can be done in hardware.

Most research I've seen indicates about 1000 connections per neuron, but even if the number is 7000 interconnections per node, it seems to be well within reach extrapolating current technology. The only limitational difference between the internet and the human brain is that the internet is too large to account for the quantum effects that you maintain to hold the key to sentience/consciousness. I believe this scale issue could be solved without any new radical advances, and just with the expected improvements in the current methods of processor construction.

Neurons operate in a fairly simplistic fashion. They accept inputs of various types and may or may not send out a signal in response. That signal, if sent, is sent to certain various neighbors in varying strengths. (Ignoring for now the issue of updating connectivity - rewiring - which would ONLY add one or two more calculations,) that means only one calculation and storing the address of neighbors. If done efficiently, one could assume that it would take roughly 16kilobytes of memory space and one simple dedicated calculation unit per neuron. That would be a fairly small processor, so bundling 512 "neurons" per chip should be feasible. Put, say, 64 chips per board (computer) and you'd already have hi-speed interconnectivity between 32,000 neurons. And the connection speed would be over 500 times faster than the chemical signal propagation in the brain. It's conjectured that most neurons in a healthy human brain are unused, so let's assume we need 30% of the 100 billion human neurons to achieve similar results in creativity and intelligence. That means you need about a million "lobes" (processor boards) to recreate the interconnectivy of the human brain, with silicon neurons small enough to even be subject the the occasional quantum weirdness. I'd say it's well within reach if IBM, NASA, or the National Security Agency wanted to plunk down the cash to have it built. (And I wouldn't be surprised if the NSA has done it already.)

That's just what can be done with today's silicon. We haven't gotten to the silicon of the year 2020; I've already mentioned bucky-balls, and you don't wanna get me started on man-made quantum computers (ion traps / quantum dots / etc...); some researchers have even found a way to store 2 bits per atom!

Really, I still think all the talk of hardware is irrelevent. I firmly stand by the supposition that sentience can surface from sufficiently sophisticated software [yeah for alliteration ]. I've even got some stuff designed and coded up myself. My old college psych professor seemed impressed with what I've termed my "Model of Natural Thought Process Pathways." So, maybe in 10 years or so, I'll actually be the one who gets to provide you with the computer program that becomes golf-happy of it's own accord .... (of *COURSE* I wouldn't cheat and plug that in anywhere .. no .. NEVER! )

GUNRUNNER

Monday, January 2, 2006 3:07 AM

Quote:
Originally posted by Karnejj:
Say Guy #2 doesn't run .. instead he gets his bucket of nanobots and takes his remote control and pushes a few buttons. The nanobots rise out of the bucket and merge together to form bulletproof armor that covers Guy #2. Say, he pushes a few more buttons and some more bots get together to form Jayne's good-ole Vera with other bots forming .50 caliber slugs ... whereupon he asks Guy #1, "Do you feel lucky punk? ... do ya??!" ... That first shot is gonna do more than sting a little .....

Is that supposed to impress me? Modern Weapons Tech can do stuff like that! Take an SM Series 'Standard' Missile, standard armament on US Navy Arlegh Burke, Ticonderoga, Oliver Hazard Perry, Kidd, California and others I'm forgetting class warships (Also on some NATO and US Allied Warships.) It can be used defensively to shoot down incoming anti-ship missiles ranging from old subsonic Silkworms to high tech sea skimming supersonic "streakiers" like the SS-N-22 Sunburn. It can shoot down aircraft, it can be used against surface ships, it can attack land targets, it can shoot down ICBMs in space, if you are feeling real evil you can light shit on fire with the guidance RADARs, it even makes Julian fries and if you order now with your credit card will throw in a Tomahawk cruise missile free! With a Standard missile you and a reliable ready weapon that can be fired in seconds (you don’t even need to push a button when the system is set for auto) and can be stored in the launch tube for months with nothing but occasional diagnostics, while if you had nano-bots you would need to wait for them to make a comparable missile, run a test to make sure they did it right, synchronize the missile to your radar and … oh wait that SS-N-22 just sunk your ship…

Yes nanobots are powerful but when you need something done quick and on a large scale they aren’t the tools to use.

EV Nova Firefly mod Message Board:
http://s4.invisionfree.com/GunRunner/index.php?act=idx

KARNEJJ

Monday, January 2, 2006 4:07 AM

That's not quite fair ... the weapons that "sink my ship" had to have been made at some time too ...

You just might be impressed to see the exploded bits from your missiles be smelted back down by the nanobots and put back into place to repair the holes..

GUNRUNNER

Monday, January 2, 2006 5:10 AM

Cruise missiles are cheap and can be built by the hundreds ahead of time. Nanobots are expensive and difficult to produce, making sufficient nanobots to build one missile is a time consuming process as is building the missile its self. Even if nanobots could built them faster than factories weapons tend to have a long shelf life, in the Gulf War the US Navy used shells and bullets from WWII, in the Falklands the Royal Navy used torpedoes from WWII (and arguably won the war with them). You can build weapons on demand but it takes twice as long or you can have weapons built at the normal rate and just store them for when conflict comes (which could be anytime) I would take the pre-built weapons. Having a system that can make weapons on demand only to have them sit around for years hardly makes sense. A unit cut off from resupply would find it useful but a proper force has large mobile stockpiles of ordnance at its disposal that has been built to spec and checked constantly for quality.

BTW would you really want to incorporate potentially dangerous missile debris in to you're ship's hull? Imagine if the missile explosion damages the nanobot circuitry and they take bits unexploded explosives from the missiles warhead and try to build a wall out of it? At a minimum you risk a secondary explosion. If they smelt down an aluminum missile wing and put it in to your damaged steel bulkhead you may have weaker bulkhead.

EV Nova Firefly mod Message Board:
http://s4.invisionfree.com/GunRunner/index.php?act=idx

KARNEJJ

Monday, January 2, 2006 4:01 PM

Quote:
Originally posted by GunRunner:
Cruise missiles are cheap and can be built by the hundreds ahead of time. Nanobots are expensive and difficult to produce, making sufficient nanobots to build one missile is a time consuming process as is building the missile its self.

Well ... of course it'd be silly to use the nanobots to make missiles out of, I was just pointing out the unfairness in production time in your example. If you really, really, really wanted to use nanobots in war.... You could really do some ugly things. I'm pretty sure nano-war would be considered a worse atrocity than using nukes (which means the US would jump at the chance to be first

.. funny how we have more nukes than any other country in the world, are the ONLY country that has proven that we WILL USE them

, and YET STILL demand other countries de-arm

... it's good to be on the side with the upper-hand

, but I guess that's another thread ...)

Assuming your enemy doesn't have nano-tech, it'd be conceivable to just program them with the geography of the enemy country and have them assimilate any biological material they find. The bots would float over the country and "eat" all the humans and other critters they come across and use those materials to multiply. It'd be something like a slightly more safer (and totally nastier) form of biological warfare. Just like disease though, it could conceivably backfire. You'd likely want military units (bases/satellites) positioned around their country as a back-up line of containment. They could pump out an "ARE YOU THERE?" code and then a "DIE!" response to kill off nanobots that were leaving their geographic limits.

If the enemy does have comparable nano-tech, then you'd have to blanket your own country with "Defense Nanobots" first, to make sure they don't capture and reprogram your own bots to use against you or to nano-battle their 'bot swarms.

CITIZEN

Monday, January 2, 2006 4:59 PM

Quote:
Originally posted by Karnejj:
On the subject of Nanobots, well, the concepts are there, and in 500 years, I would like to think that the field would be heavily researched. There are quite a large number of obstacles to overcome, so my guesses here are pretty shaky. However, onto something I do feel more convicted about ....

Very possibly, but remember concepts can be around long before the technology (if the technology ever materialises). Leonardo Da Vinci created concepts of gliders in 1490, yet no one built one till the late 1800s. Gliders are far simpler tech, even relatively speaking, than Nanobots.

Quote:
I personally think compelling unprogrammed behavior can emerge from a complex (but quite acheivable) software program.

There's reasons you can't do it in software. One in a very real way a software program is limited to concurrent serial processing, the Brain is not, and in a very real way I believe this concurrent processing is also responsible for the emergent property of consciousness. This IS not possible without the CPU's to back it up. Remember also that computer software has a certain level of precision inherent when dealing with real numbers, which would prevent it from dealing with the range of values dealt with by the Neuron. This would mean that no matter how convincing the behaviour of the system it would still be only an approximation of consciousness, not that property itself. Surely real consciousness must arise as a property of a real-world physical system?

Computers are calculators, they use mathematics to produce a convincing approximation or simulation of real phenomena, but it is just a simulation. A computer can simulate gravity within the bounds of the simulation, but it can not create gravity. What I mean is that a simulation of consciousness may give a convincing outward appearance of consciousness, but by the virtue of it being a mathematical simulation it cannot be true consciousness.

Quote:
Most research I've seen indicates about 1000 connections per neuron, but even if the number is 7000 interconnections per node, it seems to be well within reach extrapolating current technology. The only limitational difference between the internet and the human brain is that the internet is too large to account for the quantum effects that you maintain to hold the key to sentience/consciousness. I believe this scale issue could be solved without any new radical advances, and just with the expected improvements in the current methods of processor construction.

A Neuron has three different way's of communicating with other cells/Neurons. These are the Dendrites, Axons, and Soma. The Dendrites are usually where incoming signals are received, and there can be a 1000 of these, which enables connections to tens of thousands of other cells. Now Axons are usually transmitters, but can also accept incoming singles. There are often also connections to the Cell body (Soma) as well. In total there is, on average, 7,000 synapses for an individual Neuron. This is an Average, not a maximum, and in small children it is believed that there may be a total of 1,000Trillion Synapses, which is an average of 10,000 Synapses per Neuron. Again an average, not a maximum.

The physical structure of the Dendrites also effects how the signals from other Neurons are integrated (this is the weighting I alluded to earlier). This integration depends on both the summation of stimuli that arrive in rapid succession as well as the excitatory and inhibitory inputs from separate branches.

Now add to this the fact that Neurons actively include feedback from their own output via the Axons, with action potentials and voltage gated ion channels, this sort of feedback (or indeed any) is not present in any IC currently in production, or even planned.

You seem to be working from a false precept that we merely need processing power to 'ape' a brain. This is simply not the case; we need a vast amount of true interconnectivity, and dynamic interconnectivity.

GUNRUNNER

Monday, January 2, 2006 5:16 PM

The thing about nano-war is that if nations can’t use them on a tactical scale they won’t put much funding in to the weapons. Even NBC weapons are geared towards tactical combat, strategic uses are just political (even ICBMs are considered tactical in many respects). You have to assume that enemy units will have their own defensive nanobots as you said and any use would just piss off the other side and bring the international community down on you if you kill lots of civilians. On today’s battlefield chemical and biological weapons are really useless against modern militaries like those in NATO, militaries like Iran and Iraq during their war are susceptible due to the total lack of proper functioning gear. Their only use would be to force the enemy to take shelter as you launch a blitzkrieg and if you try that on land occupied by non-combatants well you are not going to be liked by the rest of the world. Have you read Tom Clancy’s ‘Red Storm Rising’? It deals a bit with NBC warfare and how to use it for tactical level operations in the beginning of the book.

EV Nova Firefly mod Message Board:
http://s4.invisionfree.com/GunRunner/index.php?act=idx

KARNEJJ

Monday, January 2, 2006 5:50 PM

Arghh, CITIZEN! Seems you've redefined sentience now

Before, it seemed you'd be satisfied with considering sentient any being capable of displaying compelling, novel unprogrammed behaviors. This, I still firmly believe, can be done with software only, mostly independant of the architecture of the hardware. Of course, the software would be complex and there would be a lot of concurrency necessary. So, either it'd need a fair amount of concurrency in hardware or sufficiently fast single processors capable of simulating concurrency (as most computers today). In other words, you'd need a bitchin' system to run "SoftHuman 2020," but that system need not have a billion linked processors.

Now, however, you seem to believe that sentience is basically defined as an emergent property of extreme inter-connectivity in adaptive analog circuits.

If the hardware is a must, then .. as for floating point vs. binary calculations, ... well, even in nature, there is no such thing as "real numbers" or "floating point numbers of infinite precision." Even nature is discrete or "binary" in that the smallest units are still limited by Planck's constant... (ie. the "quanta" in quantum physics).

I'm not sure how adamant you are on true sentience requiring quadrillionth place (or better) precision in signal calculations, but I'll assume that you wouldn't make it a strict requirement.

Other than that, the system I described above should be a fair approximation of the hardware of the human brain. Self-feedback is easily achieved as a low-weight address of itself stored in my electronic "neuron." I still don't see any insurmountable obstacles in the signal propagation, and extrapolating out CURRENT tech to the year 2020 would mean that it would only require about 2000 multi-proc boards, which is probably about the size of a large van.

If we can keep Moore's 'law' going until 2030 (which is questionable) then we can get my design proposal minituraized down to about the size of a 27-inch TV or so. Keep it going much further and we can build "Data" in this century alone. Seems kinda pessimistic to believe we can't achieve this by the 26th century.

Again, there are still more radical techs that I haven't really touched on, which could potentially more easily fit (and even exceed) the hardware requirements you have.

KARNEJJ

Monday, January 2, 2006 6:43 PM

Quote:
Originally posted by GunRunner:
The thing about nano-war is that if nations can’t use them on a tactical scale they won’t put much funding in to the weapons. Even NBC weapons are geared towards tactical combat, strategic uses are just political (even ICBMs are considered tactical in many respects). You have to assume that enemy units will have their own defensive nanobots as you said and any use would just piss off the other side and bring the international community down on you if you kill lots of civilians.

I think the precedent of the indiscriminate "Fat Boy" and "Little Man" kinda contradict the nicey-nice world you're describing. Whispered opinions aside, most countries don't have too many bad things to say very loudly about a country that can obliterate them in ugly ways.

Even barring indiscrimate killing, if a nanobot can be programmed with the geographical borders of a country, it'd be just as simple to limit them to the location of a particular enemy military installation.

Most countries would rather kill the human enemy and preserve their buildings and other material, but if you wanted a clean war, you could theoretically program the bots to eat glass and metal instead. That would limit them to destroying enemy equipment and be "world-friendly."

PREV PAGE 1 2 3 4 5 6 7 8 NEXT PAGE