The pulseless man

[Ignuus66]

Turns out, we really dont need a heart or a pulse.

http://mobile.designtaxi.com/news/351464/First-Heartless-Man-You-Don-t-Really-Need-A-Heart-Or-A-Pulse/

[llearch n'n'daCorna]

I've seen this before, and I'm muchly impressed, but...

... you still need a heart. The gentleman in question still has a heart. It's just not pumping by compression, but by turbine. But that's a minor quibble...

[Ignuus66]

I've seen this before, and I'm muchly impressed, but...

... you still need a heart. The gentleman in question still has a heart. It's just not pumping by compression, but by turbine. But that's a minor quibble...

That depends on how you define hearts or not, if you define it by the fact that it pumps/makes the blood flow through the bloodstreams, then yes, this is a heart.


Also if we are on a topic: what do you think are the boundries between human and machine?

Say if a person had a uncureable brain disease, and they get a mechanical brain instead that functions EXACTLY (behavior-wise) as the human brain, is the person a human or a machine?

[VAE]

Neat!
Also, duh.
All you need is supplying the tissues with oxygen. The only reason a heart *beats* is that it's far, far easier to implement in a body - rotary motion isn't exactly common in organisms.

[joshofspam]

Though I am impressed, I wonder if continuous blood motion from turbines is truly as effective as a natural heart.

Things like rise of blood flow to carry more oxygen to the rest of the body with a rise of physical activity, the issues of fresh oxygen gathered and waste gases expelled. One wonders if a constant flow system is as good as one with a pulse. Hmmmmm...I gather it won't really be tested until its used on a fairly active person.

[littlekreen]

Whether biological or machine you're both just a quantum computer operating at a number of abstraction levels above the base units and diverging halfway toward the top. So I'd say whether organic brain or silicon machine isn't particularly more different than different architectures of computer hardware. We are getting close to designing bacteria from scratch to do specific things and that makes them largely complex chemical apparatus. The definitions aren't going to hold up very strictly in the long term. In any case a mix to any degree would be a cyborg and a mix from pure machine in the other direction would be biorobotics (biot).

[VAE]

Though I am impressed, I wonder if continuous blood motion from turbines is truly as effective as a natural heart.

Things like rise of blood flow to carry more oxygen to the rest of the body with a rise of physical activity, the issues of fresh oxygen gathered and waste gases expelled. One wonders if a constant flow system is as good as one with a pulse. Hmmmmm...I gather it won't really be tested until its used on a fairly active person.

What do you see that would make it less effective?

[Ignuus66]

Whether biological or machine you're both just a quantum computer operating at a number of abstraction levels above the base units and diverging halfway toward the top. So I'd say whether organic brain or silicon machine isn't particularly more different than different architectures of computer hardware. We are getting close to designing bacteria from scratch to do specific things and that makes them largely complex chemical apparatus. The definitions aren't going to hold up very strictly in the long term. In any case a mix to any degree would be a cyborg and a mix from pure machine in the other direction would be biorobotics (biot).

Correction: we are NOT a quantum computer, grey matter is actually only slightly more compact than modern circuitry, and the main advantage of grey matter is that it has a very, very low power usage, and is very compact. some Supercomputers surpass the brain's (raw) computing power.
But me and my brother think that being human is based on behavior.

[littlekreen]

I say that as you're composed of interacting turing-complete components being atoms and such. The matter of a fully simulated brain would arrive at the same function just at a different abstraction of components than yours.

[Tapewolf]

I came across a remarkable paper not long ago which suggested that memories may in fact be digital, using a 6-bit system of logic gates implemented in phosphorylated tubulin.

http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1002421

...my initial reaction was "No, that's impossible...!" but I don't really have the background to make a proper judgement.

[Ignuus66]

I came across a remarkable paper not long ago which suggested that memories may in fact be digital, using a 6-bit system of logic gates implemented in phosphorylated tubulin.

http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1002421

...my initial reaction was "No, that's impossible...!" but I don't really have the background to make a proper judgement.

The science behind it seems pretty solid...
But I'm not knowledgeable (yes, that's a word) enough to decide either.
But as far as I know, human brains have a capicty of 1 too 1000 terabytes (some people say even more)

[joshofspam]

Though I am impressed, I wonder if continuous blood motion from turbines is truly as effective as a natural heart.

Things like rise of blood flow to carry more oxygen to the rest of the body with a rise of physical activity, the issues of fresh oxygen gathered and waste gases expelled. One wonders if a constant flow system is as good as one with a pulse. Hmmmmm...I gather it won't really be tested until its used on a fairly active person.

What do you see that would make it less effective?

Well...I have to wonder in-between pulses allows for greater exchange between waste gasses and oxygen. So if you illuminate the pulse, does that lesson the effectiveness of the exchange?

Obviously it works for this old man, but I still wonder if it can keep up with all the requirements needed by a more active person.

[Ignuus66]

Actually, blood is flowing continously either way, thus there is no impact on the effectiveness of oxygen dropoff.
The question is that the heart normally quickens when the body needs more oxygen, but the turbine doesn't... of course this can be solved by using a chemical detector in the turbine which detects the levels of hormones in the bloodstream, and slows/quickens the flow rate accordingly.

[VAE]

Actually, blood is flowing continously either way, thus there is no impact on the effectiveness of oxygen dropoff.
The question is that the heart normally quickens when the body needs more oxygen, but the turbine doesn't... of course this can be solved by using a chemical detector in the turbine which detects the levels of hormones in the bloodstream, and slows/quickens the flow rate accordingly.

I wonder whether simply checking for pass-through blood oxygenation would be enough? I mean,higher usage rate would mean a decrease in the leftover oxygen, unless i'm failing.

[littlekreen]

Actually, blood is flowing continously either way, thus there is no impact on the effectiveness of oxygen dropoff.
The question is that the heart normally quickens when the body needs more oxygen, but the turbine doesn't... of course this can be solved by using a chemical detector in the turbine which detects the levels of hormones in the bloodstream, and slows/quickens the flow rate accordingly.

I wonder whether simply checking for pass-through blood oxygenation would be enough? I mean,higher usage rate would mean a decrease in the leftover oxygen, unless i'm failing.

That would be the simpler way to test for it I'd think and the amount of blood oxygen needed for the average person is far more stable between individuals afaik. Detecting complex hormones would be non-trival. You'd have to put in a limiter in I'd think to account for edge cases and such.  If I recall right the only real long-term risk for these continuous flow pumps is atrophy on the part of the valves in the legs and such things since they're always open for years at a time. lungs would likely work just fine as then it'd be a continuous average supply to move over rather than having it mostly sit idle after diffusing.

[Ignuus66]

I wonder whether simply checking for pass-through blood oxygenation would be enough? I mean,higher usage rate would mean a decrease in the leftover oxygen, unless i'm failing.

Unless I am misunderstanding what you mean, I think you are failing  , the red blood cells always drop all the oxygen they are carrying, that is why when your organs need more oxygen, your heartbeat, and with it your oxygen intake and blood flow quickens, so your red blood cells can pick up/ drop oxygen off at a much higher rate.

[Alondro]

Actually, memory storage and processing in the brain appears far more complex than anyone imagined.  Single ionic channels may store data for several memories, and those memories are stored across synaptic networks in some fashion while at the same time a single neuron seems to hold onto a memory in some specific capacity. 

There also seems to be no limit on how much information the human brain can store!  At least none that's been quantified.  The details of images, faces, complex scenes with sounds and smells, entire conversations over decades, the sheer amount of memory storage required when compared to computer memory is vast.  And then there's the ability to think and create in abstractions, to logically process nonsense, to invent and investigate... no one has any clue how the brain manages to do those things. 

There's really no comparison between how a computer and the brain function anymore.  Organic neural processing is very 'fuzzy' with information constantly overlapping and being reinforced, altered, or erased.  A computer cannot operate that way because the inorganic circuitry can't function in the same way as biological molecules.  It's the layers of flexibility from the gene expression upward to entire regions of the brain that lead to life-based intelligence. 

Let's face it, even an ant can perform more functions and act with more independence than the most advanced robot.  And it only has a tiny ganglion in its head.

Also, when comparing 'life' to machines, are we talking all life or only intelligent life?  If it's compared to all life then it's possible we may one dqy crete a 'species' or simple robots which can gather materials and replicate themselves, even allowing random alterations in their programming to simulate evolution.  Remember, not all life has intelligence.  The question of whether a machine can simulate life is rather tricky, because we're still not sure how simple life can be.  If we find a living crystalline organism in space (aka Andromeda Strain", that would REALLY muddy the waters!@

[Ignuus66]

Actually, memory storage and processing in the brain appears far more complex than anyone imagined.  Single ionic channels may store data for several memories, and those memories are stored across synaptic networks in some fashion while at the same time a single neuron seems to hold onto a memory in some specific capacity. 

There also seems to be no limit on how much information the human brain can store!  At least none that's been quantified.  The details of images, faces, complex scenes with sounds and smells, entire conversations over decades, the sheer amount of memory storage required when compared to computer memory is vast.  And then there's the ability to think and create in abstractions, to logically process nonsense, to invent and investigate... no one has any clue how the brain manages to do those things. 

There's really no comparison between how a computer and the brain function anymore.  Organic neural processing is very 'fuzzy' with information constantly overlapping and being reinforced, altered, or erased.  A computer cannot operate that way because the inorganic circuitry can't function in the same way as biological molecules.  It's the layers of flexibility from the gene expression upward to entire regions of the brain that lead to life-based intelligence. 

Let's face it, even an ant can perform more functions and act with more independence than the most advanced robot.  And it only has a tiny ganglion in its head.

Also, when comparing 'life' to machines, are we talking all life or only intelligent life?  If it's compared to all life then it's possible we may one dqy crete a 'species' or simple robots which can gather materials and replicate themselves, even allowing random alterations in their programming to simulate evolution.  Remember, not all life has intelligence.  The question of whether a machine can simulate life is rather tricky, because we're still not sure how simple life can be.  If we find a living crystalline organism in space (aka Andromeda Strain", that would REALLY muddy the waters!@

Actually there are Ai's that are much more intelligent than ants, and are the size of a human brain.
the simple thing is, grey matter is just a very advanced computer. as  I said the raw processing power of machines is higher, and the human brain does NOT have unlimited storage, it just compresses memories very intelligently, making use of space much more efficiently than computers.
I would bet 1000 dollars that in 20 years we have created a fully sentient AI (they have already created learning robots, robots that work together to solve programms they weren't programmed to solve, and we are already seeing the begings of intelligence in some machines.
humans tech is accelerating very fast, I wouldn't be surprised if we reach the technological singularity within 60 years.


Now there is still my life's goal of legalizing The nuclear pulse propulsion.
that's basically my life's goal: allowing a technolodgy that has been around for 50 years that would make it (theoretically) possible for a unmanned probe the size of a corvette to accelerate to 10% the speed of light in around 4 years. With that speed we could easily reach Alpha centauri within 50 years.
Now that would be my dream....

[Viaxeiro]

As a trained Medic. it would be really creepy for me to find this guy in a deep sleep. Check his pulse and through my thorough searching and finding non, have him wake up and inquire what was going on.

I'd be inclined to keep Searching regardless, through disbelief and astonishment.

Other than that; Grab your bibles and pitchforks. Zombies are real.

[LionHeart]

I would bet 1000 dollars that in 20 years we have created a fully sentient AI (they have already created learning robots, robots that work together to solve programms they weren't programmed to solve, and we are already seeing the begings of intelligence in some machines.

The promise of fully sentient AIs has been "twenty years from now" for about the last half-century...

[littlekreen]

if moore's law holds it would be about 30 years to hit the ~100 million MIPS required to match the human brain. Somewhere after that point (entering actual industrial production) you'd be hitting a software bottleneck to make a sentient AI rather than both which is where we are now. Given we're approaching optical routing transistors and graphene CPU hardware I expect Moore's law may have a bit of a speedbump currently but should pick up again soonish. At the worst another generation or two past that you could simply brute force it by somehow taking a working copy of a smaller intelligent brain and figuring out how it processes via simulation. Emulation of hardware usually being inefficient after all.

Making a working copy of something that likely is highly sensitive to that kind of systemic copying delay being a rather high bar I'd think. We already have the storage problem somewhat comparable at least for long-term memory. Holo dvds can be found in development that purportedly will hold some 1TB though 200GB initially. At 120MB/s they'd be good for that kind of archival with their innate error-correction.

[Ignuus66]

if moore's law holds it would be about 30 years to hit the ~100 million MIPS required to match the human brain. Somewhere after that point (entering actual industrial production) you'd be hitting a software bottleneck to make a sentient AI rather than both which is where we are now. Given we're approaching optical routing transistors and graphene CPU hardware I expect Moore's law may have a bit of a speedbump currently but should pick up again soonish. At the worst another generation or two past that you could simply brute force it by somehow taking a working copy of a smaller intelligent brain and figuring out how it processes via simulation. Emulation of hardware usually being inefficient after all.

Making a working copy of something that likely is highly sensitive to that kind of systemic copying delay being a rather high bar I'd think. We already have the storage problem somewhat comparable at least for long-term memory. Holo dvds can be found in development that purportedly will hold some 1TB though 200GB initially. At 120MB/s they'd be good for that kind of archival with their innate error-correction.

Considering that if moore's law holds we wouldn't be able to make transistors any smaller by 2022, and I wasn't talking about the general public harware having sentient ai's I was talking about 1 or 2 supercomputers may have sentience.

Keep in mind that AI's can simulate intelligence pretty well already (not truly intelligent though).
Optical transistors and graphene transistors aren't the only alternate computation that can be developed, and another possibility of a sentient cloud AI.... A bit like skynet. (also what's with everybody fearing sentient AI's? )

The promise of fully sentient AIs has been "twenty years from now" for about the last half-century...

Considering we are actually REACHING the raw computational power of a human brain, it's actually slightly more feasable than 20 years from now with computers that have the computational power of your microwave.

[justacritic]

I came across a remarkable paper not long ago which suggested that memories may in fact be digital, using a 6-bit system of logic gates implemented in phosphorylated tubulin.

http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1002421

...my initial reaction was "No, that's impossible...!" but I don't really have the background to make a proper judgement.

The science behind it seems pretty solid...
But I'm not knowledgeable (yes, that's a word) enough to decide either.
But as far as I know, human brains have a capicty of 1 too 1000 terabytes (some people say even more)

We have even less due to not having anti-virus programs. I say 80% of that capacity is taken up by spam and mal-ware

[littlekreen]

I came across a remarkable paper not long ago which suggested that memories may in fact be digital, using a 6-bit system of logic gates implemented in phosphorylated tubulin.

http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1002421

...my initial reaction was "No, that's impossible...!" but I don't really have the background to make a proper judgement.

The science behind it seems pretty solid...
But I'm not knowledgeable (yes, that's a word) enough to decide either.
But as far as I know, human brains have a capicty of 1 too 1000 terabytes (some people say even more)

We have even less due to not having anti-virus programs. I say 80% of that capacity is taken up by spam and mal-ware

Heh! Well 8% of noncoding DNA is composed of inactivated viruses so I'd say you might be right with that much of the source code infected. As much data sorting as the brain does running storage would be hard to pin down I'd think. However our storage logic works the relational nature of the storage wouldn't likely expand linearly with time.

edit:alter derp phrasing

[Tezkat]

if moore's law holds it would be about 30 years to hit the ~100 million MIPS required to match the human brain. Somewhere after that point (entering actual industrial production) you'd be hitting a software bottleneck to make a sentient AI rather than both which is where we are now. Given we're approaching optical routing transistors and graphene CPU hardware I expect Moore's law may have a bit of a speedbump currently but should pick up again soonish. At the worst another generation or two past that you could simply brute force it by somehow taking a working copy of a smaller intelligent brain and figuring out how it processes via simulation. Emulation of hardware usually being inefficient after all.

Making a working copy of something that likely is highly sensitive to that kind of systemic copying delay being a rather high bar I'd think. We already have the storage problem somewhat comparable at least for long-term memory. Holo dvds can be found in development that purportedly will hold some 1TB though 200GB initially. At 120MB/s they'd be good for that kind of archival with their innate error-correction.

Considering that if moore's law holds we wouldn't be able to make transistors any smaller by 2022, and I wasn't talking about the general public harware having sentient ai's I was talking about 1 or 2 supercomputers may have sentience.

Keep in mind that AI's can simulate intelligence pretty well already (not truly intelligent though).
Optical transistors and graphene transistors aren't the only alternate computation that can be developed, and another possibility of a sentient cloud AI.... A bit like skynet. (also what's with everybody fearing sentient AI's? )

The promise of fully sentient AIs has been "twenty years from now" for about the last half-century...

Considering we are actually REACHING the raw computational power of a human brain, it's actually slightly more feasable than 20 years from now with computers that have the computational power of your microwave.

Moore's Law is pretty much on its last legs (unless you chose to read it as a more generalized rule regarding exponential progress a la Kurzweil). However, simulation of an entire human brain function at the molecular level should be possible shortly after the advent of exascale supercomputing, which probably won't much more than a decade away--still well within the reach of Moore's Law for semiconductors alone. There are a number of projects working towards that goal. The Human Brain Project is the most developed of these (and one of the best funded, especially if it wins its bid for €1 billion in FET funding from the EU). It's unlikely to create a sentient machine at first (at least, not on purpose), but it will definitely point us in that direction. Plus there are all the efforts to straight up reproduce neuron functionality in hardware (such as Stanford's Neurogrid or the DARPA-funded SyNAPSE project) rather than going the pure digital route.

20 years until a supercomputer has a thinking capacity comparable to humans is not at all unreasonable.

[RobbieThe1st]

if moore's law holds it would be about 30 years to hit the ~100 million MIPS required to match the human brain. Somewhere after that point (entering actual industrial production) you'd be hitting a software bottleneck to make a sentient AI rather than both which is where we are now. Given we're approaching optical routing transistors and graphene CPU hardware I expect Moore's law may have a bit of a speedbump currently but should pick up again soonish. At the worst another generation or two past that you could simply brute force it by somehow taking a working copy of a smaller intelligent brain and figuring out how it processes via simulation. Emulation of hardware usually being inefficient after all.

Making a working copy of something that likely is highly sensitive to that kind of systemic copying delay being a rather high bar I'd think. We already have the storage problem somewhat comparable at least for long-term memory. Holo dvds can be found in development that purportedly will hold some 1TB though 200GB initially. At 120MB/s they'd be good for that kind of archival with their innate error-correction.

Considering that if moore's law holds we wouldn't be able to make transistors any smaller by 2022, and I wasn't talking about the general public harware having sentient ai's I was talking about 1 or 2 supercomputers may have sentience.

Keep in mind that AI's can simulate intelligence pretty well already (not truly intelligent though).
Optical transistors and graphene transistors aren't the only alternate computation that can be developed, and another possibility of a sentient cloud AI.... A bit like skynet. (also what's with everybody fearing sentient AI's? )

The promise of fully sentient AIs has been "twenty years from now" for about the last half-century...

Considering we are actually REACHING the raw computational power of a human brain, it's actually slightly more feasable than 20 years from now with computers that have the computational power of your microwave.

Moore's Law is pretty much on its last legs (unless you chose to read it as a more generalized rule regarding exponential progress a la Kurzweil). However, simulation of an entire human brain function at the molecular level should be possible shortly after the advent of exascale supercomputing, which probably won't much more than a decade away--still well within the reach of Moore's Law for semiconductors alone.

Of course, things may be more complex than just molecular interactions - what if there's some quantum stuff in there? Or frequency-based-interactions; stuff that might not be part of your simulated molecule interaction and could increase the computational cost even more?

The other side of the coin, however, is that we might be able to do it sooner than expected... the replacement device might be the size of a large datacenter, however.
With modern wireless tech, however, we might be able to have a useful AI in a compact 'body' even /if/ it required a huge building for all the processing capacity.

[littlekreen]

Of course, things may be more complex than just molecular interactions - what if there's some quantum stuff in there? Or frequency-based-interactions; stuff that might not be part of your simulated molecule interaction and could increase the computational cost even more?

The other side of the coin, however, is that we might be able to do it sooner than expected... the replacement device might be the size of a large datacenter, however.
With modern wireless tech, however, we might be able to have a useful AI in a compact 'body' even /if/ it required a huge building for all the processing capacity.

Heh, yeah I can see that being a gigantic problem. I suppose we'll have several steps through inefficient models before we get anything close to a strong AI even if we have the raw processing power? Servers that can upgrade in place without stopping threads entirely and starting from a certain point from scratch aren't even written more than single threaded nowadays even assuming they exist. (I don't know of any such interacting asynchronous multiversioned program designed for anything. Do we even have any such conventional tihngs?) Much less being able to have different models of interacting with the data for the same task then ending up with a result that's consistent and self-correcting for abstract data integrity issues.

Just coming up with the algorithms we use to store things would be a hell of a leap let alone being able to track the firing of individual neurons to see how it's processed. Most things that read mental processing like MRI or whatnot are still averages of oxygen consumption or such aggregate values IIRC? I'm sure watching memory consumption and allocation would help puzzle out a conventional program but you'll still eventually have to figure out how the machine language works to get anywhere important.

[Spellblade Kelzon]

Well hopefully he doesn't get declared deceased on accident.

Going back to what makes the difference between man and machine. my thought on what makes something a worthwhile form of life is sentience. If something can question its own existence it should be considered life worth preserving. Hopefully AI doesn't go insane and try to kill us first thing it becomes self aware. My views are highly influenced by Chobits and more recently Mass Effect as well as a few other things here and there. Just my two cents, hopefully relevant enough.