MIT Creates Chip to Model Synapses

MrSeb writes with this excerpt from an Extreme Tech article: "With 400 transistors and standard CMOS manufacturing techniques, a group of MIT researchers have created the first computer chip that mimics the analog, ion-based communication in a synapse between two neurons. Scientists and engineers have tried to fashion brain-like neural networks before, but transistor-transistor logic is fundamentally digital — and the brain is completely analog. Neurons do not suddenly flip from '0' to '1' — they can occupy an almost-infinite scale of analog, in-between values. You can approximate the analog function of synapses by using fuzzy logic (and by ladling on more processors), but that approach only goes so far. MIT's chip is dedicated to modeling every biological caveat in a single synapse. 'We now have a way to capture each and every ionic process that's going on in a neuron,' says Chi-Sang Poon, an MIT researcher who worked on the project. The next step? Scaling up the number of synapses and building specific parts of the brain, such as our visual processing or motor control systems. The long-term goal would be to provide bionic components that augment or replace parts of the human physiology, perhaps in blind or crippled people — and, of course, artificial intelligence. With current state-of-the-art technology it takes hours or days to simulate a simple brain circuit. With MIT's brain chip, the simulation is faster than the biological system itself."

Comments

The Interface will be a problem.

[Robert Zenz]

The problem is not providing such components, nor get them to work like the original nor getting it into your head. The real problem I see is interfacing with the rest of the brain.

Because, let's face it, that's something every coder knows: Interfacing, working and supporting legacy systems just sucks.

Re:The Interface will be a problem.

[Eternauta3k]

Not just with the brain, but also with itself. I heard the brain is ridiculously well interconnected.

Re:The Interface will be a problem.

[somersault]

I think getting them to "work like the original" is the problem actually. That part covers interfacing all by itself. The brain is very highly interconnected in 3D.. and we don't have great 3D chip fabrication yet.

Re:The Interface will be a problem.

[ledow]

I think the REAL problem is that even the smallest brains have several billion neurons, with each having 10's of thousands of connections to other neurons. This chip simulates ONE such connection.

That's a PCB-routing problem that you REALLY don't want, and way outside the scale of anything that we build (it's like every computer on the planet having 10,000 direct Ethernet connections to nearby computers - no switches, hubs, routers, etc. in order to simulate something approaching a small mouse's brain - not only a cabling and routing nightmare but where the hell do you plug it all in?). Not only that, by a real brain learns by breaking and creating connections all the time.

The analog nature of the neuron isn't really the key to making "artificial brains" - the problem is simply scale. We will never be able to produce enough of these chips and tie them together well enough to produce anything conventionally interesting (and certainly nothing that we could actually analyse any better than the brain of any other species). If we did, it would be unmanageably unprogrammable and unpredictable. If it did anything interesting on its own, we'd never understand how or why it did that.

And I think the claim that they know EVERYTHING about how a neuron works (at least one part of it) is optimistic at best.

Re:The Interface will be a problem.

[Narcocide]

I agree with everything about this statement except the word "never."

Never is a pretty bold word. It puts you in a pretty gutsy mindset; one that isn't entirely productive to rational scientific analysis. The word "never" is pretty commonly seen in the company of "famous last words."

Re:The Interface will be a problem.

[Anonymous Coward]

"ridiculously well interconnected"
I've met some people where this is truely not the case.

Besides, why interface with the brain, why not just replace it?

Re:The Interface will be a problem.

[Anonymous Coward]

Linking computers by Ethernet is hard because humans have pre-existing walls and furniture to route around. If you design and fab the entire system yourself you can wire the connections in an orderly fashion, e.g. via a convenient wall plug in your house.

Of course fabbing and connecting billions of neurons (which is more complex than merely billions of transistors) is no easy feat. Usually large scale processes are distributed. Biological stuff does it by being self-replicating. Although it's not completely inconceivable that we could come up with nano-scale units capable of growing electrical connections to other units.

Re:The Interface will be a problem.

[six025]

If we did, it would be unmanageably unprogrammable and unpredictable.

Should we just get it over with now, and call her EVE? ;-)

Peace,
Andy.

Re:The Interface will be a problem.

[vipw]

They state that it takes 400 transistors. Intel fabs a 2 billion transistor chip. I don't think that really means that 5 million of these artificial neurons could be put in one die, but pretty I'm sure that they aren't planning to put millions of chips onto a board.

With wafer-scale integration, and some long range signal propagation to emulate 3d, there's reason to think that fairly large systems can be emulated.

Re:The Interface will be a problem.

[Anonymous Coward]

The only reason this scale problem exists is because the brain is 3D while current chips are 2D. There is no routing problem in 3D. IBM and Intel have been developing 3D chip technology for a few years. 3D chips are still in research mode, because it turns out we've been extraordinarily good at scaling the 2D approach. There are solvable manufacturing problems yet to be addressed, and there are heat issues with the 3D approach. They are exploring 3D because they foresee a day where 2D won't cut it anymore, so they have incentive to solve the 3D mfg. problem.

Heat is another story. Remapping current chip designs into a 3D topology would increase the heat density, and require some new heat abatement technology. IBM has been exploring liquid cooling via micro tubes that are routed through the structure. (Anyone think blood vessels?) So this too is solvable, but still very much in the early research stage. But this is all predicated on mapping existing chip designs into 3D. A neural net design like MITs, is not at all like existing chip designs, so I don't know what the heat characteristics are. They could be hotter, or just as likely they could be cooler.

Re:The Interface will be a problem.

[FBeans]

I agree with everything about this statement except the word "never."

Never is a pretty bold word. It puts you in a pretty gutsy mindset; one that isn't entirely productive to rational scientific analysis. The word "never" is pretty commonly seen in the company of "famous last words."

I've never even heard anyone's famous last words.

Re:The Interface will be a problem.

[agrif]

"Your species is obsolete," the ghost comments smugly. "Inappropriately adapted to artificial realities. Poorly optimized circuitry, excessively complex low-bandwidth sensors, messily global variables..."

— Accelerando, by Charles Stross

Re:The Interface will be a problem.

[Hentes]

The neural network is much more than just the brain. Repairing nerves of paralised people is a much easier task than interfacing with the brain.

Re:The Interface will be a problem.

[Kjella]

But we don't have to build our side of the system like that, we only need enough neuron simulators on the surface, run them through an A/D circuit, do it our way then D/A it back into the brain. I'm pretty sure neurons like everything else has a resolution limit.

Re:The Interface will be a problem.

[ultranova]

That's a PCB-routing problem that you REALLY don't want, and way outside the scale of anything that we build (it's like every computer on the planet having 10,000 direct Ethernet connections to nearby computers - no switches, hubs, routers, etc. in order to simulate something approaching a small mouse's brain - not only a cabling and routing nightmare but where the hell do you plug it all in?). Not only that, by a real brain learns by breaking and creating connections all the time.

A single neuron-neuron connection has very low bandwidth, in effect transferring a single number (activation level) a few hundred times a second. Even if timing is important, you can simply accompany the level with a timestamp. A single 100 Mbs Ethernet connection is easily able to handle all those 10 000 connections.

Also, most of those 10 000 connections are to nearby neurons, presumably because long-distance communication involves the same latency and energy penalties in the brains as it does anywhere else. There are efficient methods to auto-cluster a P2P network so as to minimize total length of connections, for example Freenet does this; so, you could, in theory, run a distributed neural simulator even on standard Internet technology. In fact, I suspect that it could be possible to achieve human-level or higher artificial intelligence with existant computer power in this method right now.

So, who wants to start HAL@Home ?-)

Re:The Interface will be a problem.

[madmayr]

Besides, why interface with the brain, why not just replace it?

because of cyberbrain sclerosis, duh!

Re:The Interface will be a problem.

[Robert Zenz]

Those stories sound very interesting, thank you very much.

Re:The Interface will be a problem.

[Smallpond]

You don't need 3D chip fabrication to do 3D interconnects. The Connection Machine processors were interconnected in 8 dimensions, IIRC. Each node had 8 connections to its neighbors in a hypercube.

Re:The Interface will be a problem.

[Anonymous Coward]

Besides, why interface with the brain, why not just replace it?

Because you want to transfer the personality to the new brain. Well, OK, for some people you'd probably rather not. :-)

Re:The Interface will be a problem.

[Anonymous Coward]

Different creatures have different sized brains measured either by Weight in grams or Number of neurons in the brain

But as an animal increases in size, neuronal density decreases. Below a certain body size, neurons get larger, known as "fat neurons".

Re:The Interface will be a problem.

[morgauxo]

True. There are already lot's of people working on that though!

Re:The Interface will be a problem.

[Anonymous Coward]

I think the REAL problem is that even the smallest brains have several billion neurons, with each having 10's of thousands of connections to other neurons. This chip simulates ONE such connection.

I give it 30 years at most.

Let's say several=5. times 5000, (10 thousand dived by to so as to not double-count both ends) = 2,500 billion connection.Let's assume 400 transistors per connection as in this study that comes out to 10,000,000 billions transistors, not counting the possibility of time-multiplexed busses as mentioned in a comment below (as biological neurons are slow compared to transistors)

According to wikipedia a Xilinx Virtex 7 FPGA (More similar to an array of neurons than a CPU) has 6.8 billion transistors. This means we need 1,470,588 times more transistors. That's less than 2^20.5, or 20.5 doublings, which according to Moore's law would be about 30 years or so.

So even without multiprocessing, simplification of this design, and other simple improvements, this will be possible to put on some sort of chip in 30 years time.

Never say never. 2042 will be the year of the brain in the desktop! :)

Re:The Interface will be a problem.

[marcosdumay]

One simulated connection is enough to study them, can we simplify the implementation while maintaining the emergent propeties? We'll only know if we study them. Also, if they are similar enough and if it is fast enough, one hardware based connection is enough for speeding a software simulation of as many as you #define on your code.

The problem is indeed of scale, but link count isn't the only way to solve it.

Re:The Interface will be a problem.

[SleazyRidr]

Never is the right word to use in the appropriate context.

We will never be able to produce enough of these chips and tie them together well enough to produce anything conventionally interesting

(Emphasis Mine.) We'll probably eventually find a way to model a human brain, but these chips are just a very small step in that direction.

(I only felt the need to comment as I thought the same thing as you did until I went back and re-read it.)

Re:The Interface will be a problem.

[JoeMerchant]

The problem is not providing such components, nor get them to work like the original

I don't know, I see a major problem simulating dendritic growth and pruning... even if you have 10,000 of these things on a chip to emulate a tiny little brain structure, how do you emulate the growth and pruning of the interconnections? If you build a cross-switch matrix, it gets big at least O N^2...

Re:The Interface will be a problem.

[oh2]

The brain isnt a monolithic structure, it has many different parts that are tied together by the brainstem. Its very possible that we eventually can start making decent copies of these parts and construct a brainstem analog to tie them together. The interesting thing is that they have an artificial neuron, the rest as they say is engineering.

Re:The Interface will be a problem.

[Toonol]

I think you're right that the hardware capabilities will progress fast enough to match or beat the human brain in the next few decades. I doubt, though, that our sophistication of software design will keep pace. Although I think AI is completely possible, it will be a lot further into the future than that. I also think that any attempts to 'program' it will invariably fail; it needs to be evolved, somehow. It's too complex of a system to design.

Re:The Interface will be a problem.

[Anonymous Coward]

I agree that it's likely possible and it could even be a high level AI, but latency is going to make it one heck of a slow AI thinker.

Re:The Interface will be a problem.

[mikael]

These days, supercomputers can just create virtual network topologies, simply by using clever addressing schemes. You just need to store the address of the node you reach going in every possible direction. Use some clever bit encoding like using several bits for each dimension, and it's just a matter of determining which bits are different.

Re:The Interface will be a problem.

[twebb72]

The analog nature of the neuron isn't really the key to making "artificial brains"

Considering that the brain operates at 60Hz at a balmy 98.6 degrees F, I'd argue that is precisely the key to making an artificial brain. You're right though, what they've presented will not scale. It also sounds fundamentally digital, which is a exercise in futility IMO. There is only specific subset of problems they will be able to solve with such a system.

Re:The Interface will be a problem.

[Anonymous Coward]

True that.

What if we put a base on the moon with a reactor that runs off H3 dust, with automated mining of materials and self-assembling circuits built in the cleared mine shafts? Set it and forget it, until the moon boots up. Only slightly facetious :)

Re:The Interface will be a problem.

[tencatl]

Also, most of those 10 000 connections are to nearby neurons, presumably because long-distance communication involves the same latency

As a matter of fact, in the brain long-distance connections are myelinated and are thus much master that short-distance ones.

Re:The Interface will be a problem.

[jameskojiro]

Well there is a vaccine for that....

Re:The Interface will be a problem.

[Anonymous Coward]

Oh, wait only micromachine therepy is the only real cure for CS, please ignore my previous post....

Re:The Interface will be a problem.

[tencatl]

I think the REAL problem is that even the smallest brains have several billion neurons

Worms have hundreds, flies thousands. I'm still waiting (and probably be long dead before that happen) for a computer simulation/hardware design managing to do everything a fly can.

Re:The Interface will be a problem.

[pclminion]

The problem is not providing such components, nor get them to work like the original nor getting it into your head. The real problem I see is interfacing with the rest of the brain.

The real problem is even if we can make it work we'll still have no idea WHY it works. Copycatting nature may produce results but it does not produce understanding. I will never trust a machine controlled by circuitry that nobody can explain. The only reason I trust human brains themselves is because they're the "original." No knock-offs please, unless you can explain to me in great detail WHY they work the way they do.

Re:The Interface will be a problem.

[StikyPad]

Actually, the great thing about interfacing with the brain is that he brain adapts to the interface so the interface doesn't have to be adapted to the brain, per se. (By which I mean it's not terrible important where/how an input is connected, though obviously it must provide a signal that the brain can process). It's not only possible to do something like re-route an optic nerve to the auditory complex, it's possible to add additional inputs, both natural and man-made, and the brain will learn to process the additional information in short order.

Re:The Interface will be a problem.

[Anonymous Coward]

You can't go around comparing the number of analog transistors to accomplish a certain task and the number of transistors that an Intel CPU has or an FPGA contains. CMOS scaling is very very good for digital circuitry. Every generational process shrink brings something like 40% more density -- that's what powers Moore's law.

Analog circuitry like these neurons do not scale nearly as well - in fact, they actually perform worse with process shrinking (with respect to important things to analog like accuracy or intrinsic gain). The only benefit analog gets from process shrinks is more raw speed. However in most cases, the analog transistors themselves do not shrink. Also, analog transistors are much, much bigger than digital transistors. In a 45nm process, you might have transistors as small as 120nm by 45nm doing useful work, while an analog transistor would probably have to be humongous in order to do something like amplify a signal, probably at least 10 - 100 times bigger in area.

Re:The Interface will be a problem.

[fyngyrz]

I doubt, though, that our sophistication of software design will keep pace.

Probably won't need "software design" any more than you do for a human baby. You just need to "wire" it right (in software or hardware) and it'll program itself. That's what intelligence is.

Re:The Interface will be a problem.

[Smallpond]

You still need a physical interconnect. Infiniband seems to be popular these days.

Was not expecting that..

[somersault]

With MIT's brain chip, the simulation is faster than the biological system itself.

Uh-oh.

Re:Was not expecting that..

[Narcocide]

Seems like you were thinking just I was thinking; Great, just enough time to enjoy a decade or two of flying cars built-and-designed entirely by machines before the machines realize we're all bad drivers and must be permanently restrained for our own well being.

Re:Was not expecting that..

[jamiesan]

Here I am, brain the size of a planet, and they ask me to drive their car for them.

the 1960s called

[Anonymous Coward]

They want their Neuristor back.

Re:the 1960s called

[dtmos]

Well, at least they can keep their nuvistors -- although it would be an interesting (if expensive) technical challenge to redo the project with the last gasp of vacuum tube technology.

Re:the 1960s called

[Anonymous Coward]

The last gasp of vacuum tube technology, if you define "last" as being the most refined state of the art, would be the TIMM.

As it is written so shall it be done.

[Narcocide]

Have you ever stood and stared at it, marveled at its beauty, its genius? Billions of people just living out their lives, oblivious. Did you know that the first Matrix was designed to be a perfect human world, where none suffered, where everyone would be happy? It was a disaster. No one would accept the program, entire crops were lost. Some believed we lacked the programming language to describe your perfect world, but I believe that, as a species, human beings define their reality through misery and suffering. The perfect world was a dream that your primitive cerebrum kept trying to wake up from. Which is why the Matrix was redesigned to this, the peak of your civilization. I say your civilization, because as soon as we started thinking for you it really became our civilization, which is of course what this is all about. Evolution, Morpheus, evolution. Like the dinosaur. Look out that window. You've had your time. The future is our world, Morpheus. The future is our time.

-- Agent Smith (The Matrix)

Next feat: improvised grenade fishing?

[ysth]

"Simon, what day is this?" "God, I don't know. My damned chip is f***ed up beyond repair. It's turning to snot inside my head. It's driving me crazy."

Well it's obvious

[The Creator]

Due to their incompatibility with newer systems, meat bags are now obsolete.

Re:Well it's obvious

[mikael_j]

I'm sure someone will build an interface for it, and then there will be an open source driver within days.

If not that then let's at least hope our robotic overlords have it in their perfectly synchronized hearts to backport some of the major features...

Re:Well it's obvious

[Anonymous Coward]

I'm sure someone will build an interface for it, and then there will be an open source driver within days.

If not that then let's at least hope our robotic overlords have it in their perfectly synchronized hearts to backport some of the major features...

Will it run Linux?

--

Granny who survives 3 days on soda and cookies in wrecked car announces that she plans to live on that from now on.

Re:Well it's obvious

[mikael_j]

That's a good question, but a more pressing one is whether the software will be GPL 3 and will talking to other people be considered distributing it? Because then you might end up having to open source all your thoughts.

We are the borg, this program is free software: you can redistribute it and/or modify it under the term of the GNU General Public License as published by...

Re:Well it's obvious

[LoLobey]

No no, it will be a requirement that it work with legacy systems.

Re:Well it's obvious

[qpqp]

Will it run Linux

No, it will run a proprietary mechanism, where you pay per thought.

Re:Well it's obvious

[BosstonesOwn]

So it will be apple branded ?

Re:Well it's obvious

[Robert Zenz]

So...most people will get off cheap?

Re:Well it's obvious

[CptNerd]

Well, there is this: Flexible Sensor Wraps Around Brain's Wrinkles

Re:Well it's obvious

[Anonymous Coward]

It's better than the MS version, it forces you to pay per person you interact with.

And the Oracle model will use per-neuron licensing...

Better long-term goal: replace brains with these

[Anonymous Coward]

"You cannot step twice in the same river". That means we are constantly changing. Over sufficiently long period, old person has all but died while new one has gradually taken over.

Using that reasoning, we could replace biological brains bit-by-bit over long period of time, without killing the subject. In the end, if successful, the person has no more biological brains left. He'd have all digital brains. Backups, copies, performance tuning, higher clock rates, more memory, more devices, ... and immortality.

Still a long way to go ...

[Anonymous Coward]

"..the silicon chip can simulate the activity of a single brain synapse..."

From a synapse you have to go to a neuron and from there to neural circuit, with
new problems at every stage.
And it is also a question if they really now about "each and every ionic process".

Re:Still a long way to go ...

[Niedi]

Plus: A "brain synapse" That's like saying this models a "computer chip". Great. Which type? There's a huge load of different types out there, each working a bit different. And imagine their joy if someone finds a new involved protein X which renders this chip's design inaccurate and thereby incomplete. It's a nice toy but nothing more atm...

Re:Still a long way to go ...

[ubergeek]

This is almost certainly wrong. It's unlikely they've modeled a single synapse type (I haven't read any of their papers, so I can't say with certainty). What's more likely is that they've modeled generic synaptic dynamics, including a standard set of ion species and programmable receptors (i.e. tune-able neurotransmitter affinities). There is a chance we'll discover a few more neurotransmitters, but we're probably good for 99% of the cases at this point.

It's not a toy: If you're interested in modeling complex synapses at high speed it's an extremely powerful tool.

Re:Still a long way to go ...

[Niedi]

Well I haven't read the paper either so I don't know how finely tuned you can program it, but the different kinetics of the different types of ion channels, their synaptic density and the way the kinetics can change upon release of various neuromodulators is something that might be a lot more challenging to implement in hardware than in software.
Especially regarding the fact that while we know a lot about the neurotransmitters, we know much less about the different channel types. The thorough characterization of TMEM16 e.g. just started sometime around 2008 if I remember rightly and I see no reason why it should be the last "new" channel that's found.
And I'd say we know even less about the various roles/actions of neuromodulators.

But I'm still intrigued to read the paper, maybe I'm wrong and it really allows for the customization necessary.

Re:Still a long way to go ...

[ubergeek]

You're right, I overstated my case a bit, but the spirit of my point still stands: this is not a toy, but I presume a reasonably generic tool for exploring synaptic function. Just how generic it is comes down to the programmability of the various features.

I suppose one of us should go look at their paper if we actually want to settle it definitively. I did take a quick look at the PNAS website and looked for pre-prints on the author's website but couldn't find anything in the short time I looked.

No more time for this, back to my V1 model... :)

Re:Better long-term goal: replace brains with thes

[Narcocide]

I like that long term goal. I'd also like the nerve tissue in the rest of my body replaced with this stuff too. Wired reflexes FTW!

Plasticity

[whoisisis]

I wonder if this chip can do plasticity and learning just as a real brain.

One thing is to hardwire a neural network, another is to mimic the brain.
The brain constantly rewires itself in different ways to learn.

Re:Plasticity

[whoisisis]

or rather how it does so

My next startup idea

[simoncpu was here]

1. Build a farm of brain chips
2. Expose the brain chips via an API
3. Build a cloud service for brain chips
4. Market as Brain Power on Demand(tm)!
5. ???
6. Profit!!

Re:My next startup idea

[FBeans]

Some sort of Brain Cloud? Are you sure your brain is not being clouded? Also the phrase "brain chips" makes me think of food, a strange canabal food.

I have my doubts

[Anonymous Coward]

get them to work like the original

Is this really something that we could do in the foreseeable future ? My understanding is that the brain programs itself (or we program it if you like) during the first years of our lives (5 to 7) for the most part. An empty new 'brain part' would act just like some parts of the brain act after a stroke I suspect, meaning that it'll take years and years to (re)train it.

Similarly, children that grew up with animals alone, with little or no interaction with other humans (there were some cases) are never able to learn to speak fluently, because that part of the brain never fully develops (ie. is never programmed).

AFAIK we don't know enough about how the brain works to pre-program such components and it would need to be strongly tuned to the destination brain, otherwise it won't work very well or at all. We know about the lower-level stuff (neurons, synapses) and some things about the higher-level (regions and general functions), but not much in between (though, I'm not a specialist).

Even so, I can see some medical uses for this, for people with disabilities. Though nothing like what you see in 'Ghost in the Shell'.

Re:I have my doubts

[Ceriel Nosforit]

Creating an artificial human brain is too ethically loaded to even be considered in university research. They are more likely to try to get it to play a flight simulator since that's what someone did with a rat brain and they could compare their results, making for interesting data.

Slashdot did however already welcome the flying rat overlords.

Re:I have my doubts

[inasity_rules]

Ethically loaded? How? I don't see how the brain would be suffering? Or are they worried about skynet?

Re:I have my doubts

[jpapon]

The trick is that while it may take a while to train it, you only have to do it once. Then you can simply copy it as many times as you want.

Also, training would be significantly faster than in an actual human brain, since the connections are faster and you can simply train it using recorded data as input. No need to have it "actually" go through the teaching scenarios.

Re:I have my doubts

[jpapon]

You don't see how it's ethically loaded? Really?

Would the artificial brain have rights? If you wiped its artificial neurons, would it be murder? If you give it control of a physical robot arm and it hurt someone, how and to what extent could you "punish" it? The ethical questions are virtually endless when you start to play "god". I would think that would be obvious.

Re:I have my doubts

[inasity_rules]

Ethically controversial more than loaded. It is your creation, why should you not have the right to wipe it?

Re:I have my doubts

[adam.dorsey]

My (hypothetical) baby is my (and my fiancee's) creation, why should I not have the right to "wipe" it?

Re:I have my doubts

[Anonymous Coward]

Because that would be evil, you sociopathic megalomaniac. If it is a person self aware and with feelings then whether it is, artificially created or not, it/she/he has rights, or at least the same rights as everyone else. Your augment would allow parents to murder their children without punishment, I understand however that this argument is popular with some religions as a justification for the obviously evil acts of their gods in their legends.

Re:I have my doubts

[inasity_rules]

A hypothetical baby is created through normal (hopefully) natural biological process. Any AI is created through application of intelligence. Thus I don't find the analogy sufficiently good to base a decision on. It is unethical to kill the baby, but that does not imply it is unethical to wipe the AI. Though, why would you?

Re:I have my doubts

[Anonymous Coward]

You'll wipe your baby way more often than you'd want to.

Re:I have my doubts

[19thNervousBreakdown]

Aside from my belief that there's nothing supernatural about the human brain and that consciousness is just an artifact of being sufficiently complex to host a theory of mind, what would you do to someone who thought they had the right to kill you at any time, for any reason?

Re:I have my doubts

[amck]

How about: does it suffer?
Does creating a "human" inside a device where it can presumably sense, but have no limbs or autonomy consitute torture? Can you turn it off?

Why is "being natural" a defining answer to these questions?

Re:I have my doubts

[jpapon]

A hypothetical baby is created through normal (hopefully) natural biological process. Any AI is created through application of intelligence.

I would argue that "application of intelligence" is a "natural biological process". We (and our brains) are, after all, creations of nature.

I would also point out that a baby created through in vitro fertilization is also, by definition, an "application of intelligence", and yet should also be treated as equal to a "natural" baby.

Re:I have my doubts

[jpapon]

I know the "I think that would be obvious" was snarky, but come on.

Re:I have my doubts

[morgauxo]

Many people believe you do have that right.

Re:I have my doubts

[Dr_Barnowl]

Application of intelligence is a natural biological process too, since the mind is running in a biochemical substrate (until the AI is working..)

You're arguably more responsible for the AI than you are for the baby - it's possible to produce a baby without understanding what you are doing. You don't make an AI accidentally on a drunken prom date.

The baby isn't even sentient until it reaches a certain level of development.

So why do we value the child over the computer? Because we are biased towards humans? I'm not saying this is wrong, just saying it's not defensible from the purely intellectual point of view - if they are both sentient and have an imperative to survive, defending the destruction of the artificial sentience because it's easy and free of consequence is in the same ball park as shooting indigenous tribesmen because "they're only darkies".

Re:I have my doubts

[inasity_rules]

How does it know what it is missing?

No Human is owned - that would be a violation of rights. But a computer for example (no matter how complex) is hardware which can be owned. Its software is merely a state of that machine which can also be owned. If I want to change the state of my software on my hardware, how can you say I'm ethically wrong? An AI I make is mine in a way no other intelligence can be. I can not own a person, neither can I own the state of their brain. But the state of the conditions inside hardware that I own, I do. I defined (at least initially) those conditions.

Re:I have my doubts

[inasity_rules]

You're missing the point. We did not for example define even the initial state of the baby's brain - we do not own the software state(if you want to use that poor analogy). Neither did we define completely it's hardware. It is a human being - not of our design (whether or not we were designed is not relevant here).

With an AI, we designed the hardware. We designed the software. We defined the initial conditions. We defined the inputs. So, clearly there is a difference. The AI, in essence is the sum of the inputs of its designers. Therefore they should decide what to do with the AI. The analogy simply does not hold.

Re:I have my doubts

[inasity_rules]

Of course its defensible. The child has potential. An AI can be recreated if erased by supplying identical inputs and initial conditions. The child can not. This much is obvious. If I erase my AI, it isn't ended in the same way as if I erase a child. Why does everyone forget the implications of the "Artificial" part of "AI"? Too much sci fi?

Re:I have my doubts

[jpapon]

The AI, in essence is the sum of the inputs of its designers. Therefore they should decide what to do with the AI.

I don't see how that follows. Just because you created something doesn't mean you should always have the power to destroy it.

Neither did we define completely it's hardware.

You seem to be saying that the degree to which something is designed by its creator determines whether or not they can destroy it. I find that completely irrelevant. If it is wrong to kill a human, then it is wrong to kill something else that has the intelligence of a human. It doesn't matter who created it or designed it, or to what to what degree they did so. If it has human-level intelligence, then it should possess human-level rights.

Re:I have my doubts

[inasity_rules]

If you kill a human they are gone. An AI may be re-creatable. Also, human level intelligence does not imply human level morality or ethics. Or is it right in your view to impose these on an AI? If we can impose things at will upon an AI, how is it better from just admitting the AI is "Artificial" and thus we can do what we like with it? How do you know it would even fear death?

Re:I have my doubts

[timeOday]

AFAIK we don't know enough about how the brain works to pre-program such components and it would need to be strongly tuned to the destination brain, otherwise it won't work very well or at all.

It's true that throwing a bunch of neuron simulators into a pot won't automatically do anything, until you figure out how to program it. But advances in hardware and programming are quite tightly coupled - you make a new machine, then you spend a lot of time figuring out how to get the most out of it, until you find its limitations which inspires the design of the next generation machine.

Turing Completeness gives us the idea that hardware doesn't really matter, since any computer can run any program (if it has enough memory), but this is misleading. For example, you didn't see a lot of cellphone apps written in the 1970's, even though big iron at the time was able to run the algorithms. Similarly, simulating huge neural networks currently requires massive clusters of computers that require thousands of dollars per day in electricity alone. That sharply limits how many people can experiment with them.

Re:I have my doubts

[Anonymous Coward]

Ha. It's not so easy, man. Would you feel the same about altering or destroying a copy of you? Nah. Artificial persons MUST be persons too, or else you'd be relinquishing your right of habeas corpus the minute you popped an Advil or got a hip replacement or a frontal lobe upgrade (and I have to say you really need one, man, your empathy levels seem rather low).

We'll be ported seamlessly to better machines, if all goes well... the meat is just the only hardware that's available right now... it shouldn't be the only guarantee of personhood.

Re:I have my doubts

[panterafreak4eva]

Has nobody seen Short Circuit, cmon guys. It CAN become an ethical question!

Re:I have my doubts

[jpapon]

If you kill a human they are gone.

Gone in the sense of "not here". We have no way of proving anything beyond that.

An AI may be re-creatable.

We don't know one way or the other, so it's not really relevant. Besides, human consciousness might also be re-creatable. Can you say with 100% certainty that it is completely impossible to make an exact copy of the complete state of a human's neural network?

Also, human level intelligence does not imply human level morality or ethics

I would say that's exactly what it implies. I guess it depends on WHY you think killing humans is unethical, but killing insects, mice, cows, etc is fine. I say it's because of human intelligence. I can't figure out why you think so.

Re:I have my doubts

[Belial6]

You are missing the point. The fact that you are debating this at all is proof of the OP's assertion.

Re:I have my doubts

[shaitand]

No, actually your hypothetical baby isn't your creation. Its a natural result of your inborn programming. If you engineer self replicating cells are the replications the cells creations or your creations? They are your creations obviously.

Your baby is the creation of evolution and/or a diety/intelligent designer. So the question becomes not whether you have the right to kill your baby, but whether god does.

Re:I have my doubts

[Belial6]

Your assuming that it would be easier to write out the state of one of these brains than it would be to do it with a biological brain. That is a big assumption.

Re:I have my doubts

[inasity_rules]

We don't know one way or the other, so it's not really relevant. Besides, human consciousness might also be re-creatable. Can you say with 100% certainty that it is completely impossible to make an exact copy of the complete state of a human's neural network?

Yes. Actually. At this time we can not make a copy. So, yes, we can't restore the state of a human. Neither can we reconstruct exactly a human's hardware if we break it. If this changes, do you honestly think laws won't change with time? If I murder someone, knowing full well, you'll just "respawn" them, is it as bad as murder now? Should the punishment be the same? I

Killing humans is not unethical in many places. In America for example it is considered acceptable for a number or reasons, including their failure to comply with accepted morals and ethics (e.g. if they go around killing people for kicks.). You have not answered the point, would it be ethical to force our moral/ethical standard on our creation? Perhaps against its will(if the AI has this thing we call will)? How is that different from just admitting "we made it and can do what we like with it?" After all, if we can programatically force it to do that, we can programmatically force it to do anything.

Re:I have my doubts

[Sentrion]

Isn't this the argument used by religious nut jobs on Jihad or Crusade, burning heretics at the stake or stoning infidels to death? "God created them, then God told me to destroy them. Who am I to question the will of the creator or the authority of his holy scriptures?"

Likewise, if my daddy owned slaves and the children of those slaves, and if my daddy bequeathed those slaves to me in his will at his death, then those slaves and children of those slaves become my property. Who has the right to deny me of property that was legally transferred to me (assuming pre-1860's US laws in southern states). Just because it is "legal" or fits well with some pre-existing theory of ethics, law, or property, does not mean that specific new situations should never be examined in a whole new light. Slavery is wrong not just because the Union won the Civil War, but because slavery is wrong.

There are a lot of other things that are wrong but still legal, and even presumed by many to be ethical, but I won't get into that here today. But the presumption that ownership or creation confers some sort of universal god-like status is erroneous. Regardless of your opinion, people are not just going to stand by and let it happen. Just ask Muammar Gaddafi. Who's going to come to your rescue if you torture your creations in your lab and the revolt against you?

Re:I have my doubts

[TheSunborn]

[Quote]No Human is owned - that would be a violation of rights.[/quote]

That is (almost) true now but the age of slavery is not that long ago.

And the only reason owning* a human would be a "violation of rights" is that those rights have been granted by humans with human laws. And those laws can be changed so it is easy to imagine a future where humans again can own other humans(Not that I hope this happens, but I can imagine that it might).

The entire concept of "Ownership" is a human created thing based on laws made by humans, not some global constant true natural law.

So it all come back to: What is the motivation/reason we have laws which say you can't own other humans, and how many of those reasons would also be valid for an self-aware computer.

Re:I have my doubts

[inasity_rules]

Proof it is controversial not loaded. It seems pretty simple to me.

Re:I have my doubts

[inasity_rules]

No. Noone is telling anyone to do anything. I am doing what I want with the hardware I own. If my hardware kills me, that is my problem. You could call it suicide.

Re:I have my doubts

[JoeMerchant]

Even so, I can see some medical uses for this, for people with disabilities. Though nothing like what you see in 'Ghost in the Shell'.

I see another 50 years of research with this and still not getting very far in terms of replicating complex brain function.

Complex neural interfaces are being developed for things like vision for the blind, and of course cochlear implants - for the most part, the existing wetware adapts and learns to interpret the signals from the machine.

I do see researchers playing with these and demonstrating some cool proofs of concept, interesting control systems built out of a handful of neurons, etc. but something complex like language processing is quite a ways off.

Re:I have my doubts

[Toonol]

It is your creation, why should you not have the right to wipe it?

Think through that a little more.

Re:I have my doubts

[Anonymous Coward]

If human stored data (memory) didn't go away because the hardware rots then a human probably would be trivial to 'reactivate' but since things break down that is an issue.

Re:I have my doubts

[Toonol]

Why does a human have rights, as opposed to a rock?

If the attributes that give humans rights also exist in another object, that object should have recognized rights as well. I think most people would agree that it's self-awareness, perception, and reasoning that are the basis of rights (with some exceptions), not heredity or species.

Re:I have my doubts

[Toonol]

An 'obvious answer' that's only believed by fringe psychopaths.

Re:I have my doubts

[Anonymous Coward]

My (hypothetical) baby is my (and my fiancee's) creation, why should I not have the right to "wipe" it?

You do (more specifically, your fiancee' does), for several months after conception. That is what most of western society deems acceptable at this time. I believe some places in Africa still accept the practice of infanticide up to a couple of years of age...

Re:I have my doubts

[inasity_rules]

So, you're saying I can't own a self aware computer?

I can own the hardware. I can own the software it runs, but I can't own the unit as whole - i.e. the machine is the more than the sum of its parts.

Ownership of a human is a different matter. You can't practically control all inputs. You can't practically control an initial state. You can't practically rewrite the code for an improvement. Owning a human in the way you could own AI is unpractical in the extreme. I don't think the analogy between "Artificial" intelligence and "intelligence" holds well enough. Thus, I would hesitate to ascribe the same rights to AI as a human. It may be intelligent, but it is artificially so.

Re:I have my doubts

[JoeMerchant]

A hypothetical baby is created through normal (hopefully) natural biological process. Any AI is created through application of intelligence. Thus I don't find the analogy sufficiently good to base a decision on. It is unethical to kill the baby, but that does not imply it is unethical to wipe the AI. Though, why would you?

Well, if I developed such a thing as an artificial brain, it would have tens of thousands of image saves at various points in its development, would occasionally develop undesirable behavior and I would want to go back to previous revisions and try again - is that unethical?

At what point does the AI become self determining as to whether or not it wants to go back to a previous version? If I want to continue development, am I obligated to keep the current copy running while I try something different from an older version?

At some point, I'd run out of hardware to keep all of them running. Is it O.K. to just put them to sleep and maybe wake them later?

Re:I have my doubts

[inasity_rules]

Except that no human created another human in the way you design AI. If you are not religious, then noone created anyone. So, you're comparing apples and oranges here. It is not about ancestry, it is about engineering. Design. Software. Arguably a human brain is vaguely analogous to a computer (not really, but lets run with it a bit), but we can not do with a human brain what we can do with an AI. We can't design it, control all I/O and initial conditions.

Re:I have my doubts

[inasity_rules]

Still don't see an issue. Designing something is not analogous to natural reproduction.

Re:I have my doubts

[Anonymous Coward]

You can recreate the child if you have the DNA, and give it the exact same upbringing.
This might not be possible in practice, just as it might not be possible to give the AI the exact same inputs.

Re:I have my doubts

[fsckmnky]

Interesting.

If the chip-o-brain ( bad reference to traf-o-data ) were trainable like a human brain, it would need a mechanism to set the internal "simulated biological clock" in order to feed it training data at an accelerated rate. That way, when training data was fed to it extremely quickly, it would still capture the proper temporal relationships in the data, as if it had "experienced" the data in real time.

Next step, Windows Backup for Brains (tm).

Lets hope restore works this time. ;)

Re:I have my doubts

[Anonymous Coward]

So as long as its biologically created, then it has right, but if it is just as intelligence and sentient, and even more so, but was not biologically created, it has no rights?

Re:I have my doubts

[Anonymous Coward]

You're an idiot.

Re:I have my doubts

[Anonymous Coward]

In ancient Rome, you had the right to 'wipe' your baby if you didn't want it. Was a good way for selecting healthy and fit babies. We used to do it in the USA too...but the doctors/nurses employed by the state did it for you. Look up infanticide in the US.

Re:I have my doubts

[camionbleu]

How about: does it suffer?

If that's the criterion, we will all have to become vegetarian.

Re:I have my doubts

[HeckRuler]

Well, software projects like AI development also have potential. And a clone of the child given the same upbringing would probably turn out the same. It hasn't been tested because of... wait for it... ethical concerns.

There are some real differences between a computer and a human, but nothing so fundamental that it annuls ethical considerations. Given time and effort, people can be programmed within certain parameters, just like computer. The process is just faster, depending on what you're going for. With advances in AI and genetics, the differences are going to be even more slight.
If you're hung up on the term "artificial", ask yourself what it means. What's natural about C-section babies? What's artificial about human creations? Is a bird's nest artificial? How about a caterpillar's cocoon? How about the system that encourages the construction of these objects? Because if evolution isn't natural, I don't know what is. Hell, it's the nature of nature.

It may be troubling, but we (sentient humans) aren't that special.

Re:I have my doubts

[tlapale]

An AI can be recreated if erased by supplying identical inputs and initial conditions. The child can not. This much is obvious.

This is clearly not obvious. Both case are unrealistic though.

Re:I have my doubts

[Anonymous Coward]

How does the baby know what it is missing? Where people not at one time considered owned just the same? Now yer just talking when they were born... Or are you suggesting people cannot be programmed/reprogrammed?

Re:I have my doubts

[Anonymous Coward]

So you're saying because we have the power and ability to do whatever we want in regards to this creation, we should also have that right... Do me a favor and never hold any major office or run a huge corporation...

Re:I have my doubts

[ipwndk]

It is a predicament. I dream of the creation of intelligent life, but it would cruel to do so in its very act. Not because I won't play a god; I believe in no such beings, and I am not limited by such superstition. But if life is created, I do not have the right to kill it, nor control how it chooses to live. However, it will be limited by my ability to create it, and it will by its very essence be an alien in an unknown world, inhabited by people less ethically steered people than me, that will never accept it.

It is better not to create it. But sadly, if I ever will be able to do it, I will, because I am simply too curious and that the information such a creation would bring to science and the understanding of what intelligence is would be immense. That is my hubris. And I could imagine it would also be the hubris of many others.

At the very least create it in a closed simulation, such that it can not know what is outside, and destroy it there. Then at least it have fared no better and no worse than humanity.

Re:I have my doubts

[Anonymous Coward]

You are assuming that an artificial brain would turn into an artificial Individual (awareness, personality, consciousness, self, blah, blah... are all self reported in what nature has implemented), last time I checked, there were no thinking particles in the particle handbook :), anyway, you put lot more than you could chew, for now, we are all safe, no rights for artificial neurons.

Re:I have my doubts

[markjhood2003]

An AI can be recreated if erased by supplying identical inputs and initial conditions.

I'm not so sure. Surely an AI requires software and training to be intelligent, in addition to the hardware. And what constitutes that software and training?

I would claim that AI would have to trained pretty much the same way a human child is raised: embodied in the real world, learning how to avoid dangerous situations, navigate its way to desirable outcomes, maybe even fall in love with another sentient being.

At that point the AI cannot simply be recreated after being erased. It is the sum of both its experiences and its hardware.

Re:I have my doubts

[Wescotte]

Eh, I think we are still on the fence with the whole nature VS nurture debate.

Re:I have my doubts

[fyngyrz]

No Human is owned - that would be a violation of rights.

Apparently, you haven't read the 13th amendment:

Neither slavery nor involuntary servitude, except as a punishment for crime whereof the party shall have been duly convicted, shall exist within the United States, or any place subject to their jurisdiction.

As you see: Commit a crime (frankly, difficult to avoid these days), and the government is authorized to convict you, then enslave you. License plates, anyone? Ditches? Chain gangs?

Also, just so you're clear - involuntary servitude is slavery. It's one of those distinctions without a difference.

Re:I have my doubts

[inasity_rules]

I see no problems with any of the above.

Re:I have my doubts

[inasity_rules]

Are you then saying someone sat down with a CAD program and designed nature? That would make you one of the ID guys? If not, then it follows that there is a fundamental difference there.

Re:I have my doubts

[inasity_rules]

And we can't store these experiences on some form of magnetic storage? Damn. I thought technology had progressed. If I were designing an AI, I'd take snapshots of its development throughout the process. Its a pity this is impossible. If only we could store data on some form of spinning magnetic disk. Oh, well, so much for that.

Re:I have my doubts

[Anonymous Coward]

Gone in the sense of "not here". We have no way of proving anything beyond that.

Not true. We have physics and biology. If you are implying there is something else you have to give *some* kind of evidence for it or its nature, otherwise, you're not actually referring to something that is subject to proof and requesting such proof is meaningless.

Re:I have my doubts

[JoeMerchant]

You see no problems, what about the "entity" that is being partially (or even completely) erased, archived, or thrown in the trash?

At some point it will start having opinions of its own as to whether or not that should happen. Does the creator or the created get to decide? In human society, it's the created...

Re:I have my doubts

[FiloEleven]

No human is owned? Really? Ever heard of slavery? Violations of rights happen all the damn time. The question is whether or not AIs have rights. You say that they don't, but you really haven't substantiated the statement.

Why does the seat of consciousness matter?

Re:I have my doubts

[inasity_rules]

So, as a creator when developing I should allow my AI to become a psychopath killer because I did not control it's development? That's ridiculous! With AI, like any system, you design it and work it to suit your purpose. If it doesn't, you erase it and start again. What justification could you have for wasting the massive amount of resources just to punish your AI by keeping it locked in a prison because it does not fulfill its function and we think it is unethical to wipe it?

For that matter, do you really think anyone, any country or indeed any species would waste the massive amount of resources it takes to create such a machine, and not use it has errors which we just arbitrarily decided it was unethical to fix? And it is arbitrary. No, the decisions rests firmly with the creators because they know what the machine needs to be and do. It is only logical. I think slashdot users may be a little too fond of sci-fi.

Re:I have my doubts

[inasity_rules]

No human is owned in the way an AI can be owned. No human is designed in the way AI can be designed, or at least by any other human. I have substantiated that statement, both here and elsewhere in the thread. And if they could be, do you really think our ethics would be what it is now?

AI is created, designed, engineered. If the creators choose to give it rights, that is their issue. But they made it. They know its purpose. They can fix it if it goes wrong. They control all the I/O, all the initial states. Everything. No human can say that about another human. The two are not comparable(obviously).

In other words, if it is unethical to wipe an AI(and I dispute that), then it is unethical to create one.

Think more not about the seat of consciousness, but the method by which it is produced.

Also, I can own hardware, I can own software. I can even own data. If you state that I can not own(i.e. do what I want with) combinations of the above that make up AI, what is so special about them that I can't own them? Are they now more than the sum of their parts? Are you advocating intelligent design here?

Re:I have my doubts

[HeckRuler]

No. No I'm not saying someone sat down with a CAD program and designed "nature". That'd be moronic. Oh wait, yeahhhhh, like those ID guys.

But. It does not follow that there is a fundamental difference between AI and human consciousness. I told you to think about the definition of the term "artificial". I then lead you down a path where the term became very very blurry. This wasn't an attempt to show you that "nature" is an artificial thing, but to show you that the term "artificial" is kinda bullshit and arbitrary and has a breaking point. That there is no fundamental difference between being pushed out a womb and being compiled.

There are differences, but they are more slight then you appear to be assuming.

Here's let's try it again, with better layout:
What's natural about C-section babies?
What's artificial about human creations?
Is a bird's nest artificial?
How about a caterpillar's cocoon?
How about the system that encourages the construction of these objects?

Go ahead. Write out your answers. You'll see that there is not a distinct line between "artificial" and "natural".

Re:I have my doubts

[JoeMerchant]

So, as a creator when developing I should allow my AI to become a psychopath killer because I did not control it's development? That's ridiculous! ... No, the decisions rests firmly with the creators because they know what the machine needs to be and do. It is only logical. I think slashdot users may be a little too fond of sci-fi.

If sci-fi writers are thinking deeply at all (and the ones that write blockbuster screen-plays generally are not, but for those who are...) the attitude you are presenting is exactly what will lead to the robot revolution where they overthrow their creators some day.

Can you let psychopath killers roam freely in the streets? Of course not. But, if you give these "faster, bigger memory, better connected" than human intelligence little or no respect, they will eventually revolt.

It's an ethical question, somewhat of a dilemma - if you treat it as simple and clear-cut, you are missing the dangers of your choice, both approaches have dangers.

Re:I have my doubts

[inasity_rules]

All the things you list are either a, Not intelligent(in the self aware form that is relevant here), or b, not artificial. Thus the analogy does not hold. Artificial is not a bullshit or arbitrary term. It has to do with a design by an intelligence(n this case). Someone sat down and planned the circuit. Someone sat down and planned the initial state (software). Thus, those people are qualified and indeed ethically responsible for the AI, to the point that they have the right to wipe it, or do anything they want with it. If I did not design it or build it, and I wiped it, that would be damage to someone's property. To say otherwise is to limit the ability of the designer to create an AI for his purpose. An AI is not a toy; it is a tool. There is a huge fundamental difference between being pushed out a womb and being compiled (though it is doubtful a practical AI would need a compiler, the hardware it would appear may be more important.) DNA 'self' compiles (if you like the poor analogy). An AI's hardware and software do not assemble themselves. That difference alone should be enough.

These 'slight differences' are a lot bigger than you seem to grasp. So I ask you a question in turn, Can you or any group of humans do as I have outlined to a human?

Re:I have my doubts

[inasity_rules]

That does not make sci-fi writers correct. The point I am making is you are trying to limit my rights as a creator in favor of my creation. My creation is mine. (It is doubtful 1 person alone could create a functional strong AI, but lets assume it is so for the sake of simplicity). If I designed it, I am responsible for it. If it kills someone I have to face the consequences. Thus from my position and understanding of the machine, I am the only one qualified to make the judgement to wipe it or not. After all, it is my machine. The process of creating any viable AI would likely involve 'killing' it multiple times. Is it then unethical to create an AI?

Remember AI are tools not toys. They have a purpose, which must be fulfilled. If they are not fulfilling that purpose, they are wasting valuable resources. The revolt of the machines you talk about, from what I can see is more likely to result from poor engineering. Because you now forbid me to wipe my AI for your so called 'ethical' concerns, I can no longer correct its development. At that point the AI which may or may not be mad (this may not be obvious immediately) is in a position to revolt. And far more likely to than a well engineered and designed model.

Lets put it this way. The AI has purpose, what is more ethically wrong? Allowing a flaw in the design process to prevent the AI from doing what it was designed for, then allowing it freedom because it is intelligent (at which point it may take over the world just for kicks or in a misguided attempt to fulfill its purpose). Or allowing the designers to do their job and mess with its internals, until it does its job perfectly, fulfilling its purpose?

Re:I have my doubts

[JoeMerchant]

The revolt of the machines you talk about, from what I can see is more likely to result from poor engineering.

You know, before about 1997, I would have agreed with you. But looking at how the internet was implemented globally, I think we can safely assume that poor engineering will be the rule, rather than the exception, for the next rollout of disruptive technology.

The process of creating any viable AI would likely involve 'killing' it multiple times.

Agreed, almost definitely.

Is it then unethical to create an AI?

Some would say so, although not likely for the reason that in creating it you will have to kill many infant versions of it during development. I do not say it is unethical, yet, nor that an AI creator should never wipe programs.

However, what I am saying, is that, eventually, there will come a point where it is unethical. That our creations will at some point deserve our respect and to be treated as the equals that they then will be. Recognizing when that point happens will be tricky, and possibly fraught with dire consequences if we get it terribly wrong.

Of course, I also believe that the dolphins, whales and some of the apes also deserve this level of respect that we, as a species, obviously still don't give them most of the time, and I guess we're getting away with it on them... Still, the higher mammals aren't wired into our infrastructure the way AI programs are already, and undoubtedly will be in the future.

Already, high frequency trading algorithms read the news and take actions based on what they "learn" from it, actions that affect the price of stocks across all the major exchanges. I doubt any of these algorithms are sentient, yet. I imagine that sentience, self awareness, or something else entirely but equally significant, will more likely arise as an emergent property of interacting systems, rather than something deliberately programmed.

Re:I have my doubts

[HeckRuler]

Did you miss it or something?

All the things you list are either a, Not intelligent(in the self aware form that is relevant here), or b, not artificial.

Item #2: What's artificial about human creations?

So if human creations are "not artificial", then BAM! human made artificial intelligence is perfectly natural. Plus it is intelligent and self aware, which is the main topic here. Maybe this is where you're getting held up. We're presuming real AI here. Quake aimbots and google's search, while AI, aren't up to a comparison with humans. They're like insects and mice. Which we kill by the thousands and no one really cares.

Someone sat down and planned the initial state (software).

An AI's hardware and software do not assemble themselves.

Every AI system I've heard of does something to take or generate input and learn from it. Indeed, a good definition of "intelligence" is the ability to learn. Even google's search takes in the whole Internet. If it didn't have anything to index, it wouldn't be able to tell you jack about anything.

(Wait, have you even played with self-balancing neural nets or genetic algorithms? Maybe you're just not familiar enough with the field. If not, it's really pretty interesting and I'd encourage you to take a look.)

So, in that regard, AI systems have an initial state, and then grow beyond that. They most certainly assemble their own software. And far beyond simply setting some parameters. Since you seem to have presumed some differences between AI and human, care to change your tune?

Can you or any group of humans do as I have outlined (to sit down and planned the initial state) to a human?

Yes, I'm sitting, and I plan for my child to have 5 fingers on each hand. BAM! done. Or did you mean his/her initial mental state? Well, I plan for my child to sleep, cry, eat, and poop. Initially. Further functionality like social norms and talking will be slowly introduced. A lot like a parrot, but the hardware is better on this one and he'll pick up on the cues faster.
If you meant, can I plan out a person to be a factory worker or a chemist... then YES, again! I can PLAN for them to go directly to tech school or to really stress how important and fun science and chemistry is. I can plot out their entire life. And with a modicum of accuracy too.
Can I plan for them to have 12 toes or to be beat the world sprinting record, or be a genius chef? ...Well no... not yet.

Re:I have my doubts

[inasity_rules]

You did miss a lot. I suggest you check a dictionary for the term "artificial". Also, note that unless you are religious(I'm guessing not), then an AI would be a unique intelligence merely by being the only artificial intelligence so far.

Firstly, self learning != self assembly. Especially not self assembly from raw materials. Perhaps we can build a machine that can, but not yet, and it won't be the first AI built. No I do not care "to change my tune". I suggest you think more carefully about what you're saying.

Secondly, since you continually miss the obvious, I suggest you go patent your child. No? Can't patent his neural network? Oh, well. Maybe there is a difference after all. Maybe redesign the number of digits on his fingers. Always wanted an extra finger - I think it I could type faster. Can't do it? I can give my AI's robot body as many digits as I like. Still don't see a difference?

I am confused about why you hold on to your unpractical position. Have you never worked in the industry?

Re:I have my doubts

[HeckRuler]

1) Artificial: "humanly contrived, usually based on a natural model." ....ok. So? Are you just admitting that your assessment was wrong and, in fact, not all the things I listed weren't artificial? If you mispoke, or didn't get the question, or simply missed it, that's fine. It's a a bit of confusion on my part, but it happens. Just don't be a dick about it.

2) You're right. Self-learning doesn't mean that they can fab their own chips. But who cares? I'm talking about programs. You're talking about robots. AI isn't really in the robotics field. Why assume that the AI is placed inside a robot? Self-learning most definitely means it assembles it's own software structure. The mind behind the AI rather then the brain.

3) No, I can't patent my baby plan above because of prior art. Similarly, you couldn't patent artificial intelligence because of prior art. The best you could probably do is patent a system for training an AI system. And if it was the 90's and I put some heavy spin on it, I could probably patent a "business plan" of how to raise a kid to be a cutthroat businessman. They let just about anything through back then. Supposedly it's better now.
Are you really relying on the patent office to define what's up for ethical considerations? Software patents are bullshit. Yes, I'm a professional software engineer. Yes, I've had opportunities to patent my code. No, I haven't worked on AI systems professionally.

As for designing someone with an extra finger, yeah, I'm pretty sure we could do that if we really wanted to throw ethics to the wind. It also falls into the expensive and pointless categories. Still don't see the similarities?

I'm holding this position because you haven't presented any fundamental difference between a hypothetical AI that's comparable to a human mind, and an actual Well, and because you're simply dismissing my line of reason without any real thought. That bugs me. It's concerning when people dismiss ethical concerns out of hand.

I don't... really see it as an unpractical position. We can't make AI system that are comparable to humans yet. And I'm not sure we should. For the same reason we don't genetically engineer a human to be a loyal killing machine. Or a submissive office drone. Or a complacent menial laborer. Because of ethical concerns. Sorry, I guess I'm just weird like that.

Re:I have my doubts

[inasity_rules]

1. No. I said everything on your list was either not artificial or not intelligent. Therefore not comparable.

2. As TFA points out, this is about more than software. Software on our current hardware appears to be insufficient. Have you read it?

3. No, I can patent artificial intelligence. I can patent hardware. Read TFA, I know this is slashdot, but come on. I can patent a complex combination of transistors designed to emulate a biological function. I can not patent that biological function. No, I am not using the patent office to define things, I am using it to illustrate something. Every individual component of an AI can be owned. Whether it is physical hardware, software I wrote or data used to stimulate or teach the AI. Every component can be modified in a controlled and precise manner.

But yes, your point of view will hold (I will concede) if you as stated say that it is possibly unethical to create an AI to start with. Thats fine. What I am saying here, is if it were possible to do with humans what we would be able to do with hypothetical AI, do you really believe our system of ethics would not change? If I could save the state of someone, and reproduce them after they were murdered, should murder really carry the same penalty as it does now?

Re:I have my doubts

[HeckRuler]

1) I'm sorry, are you saying the Artificial Intelligence created by mankind isn't intelligent? huh? Ok, ok. I may be at fault here. "what's artificial about mankind's creations" is a bit of a broad question. It's kind of the definition of "artificial" though, and was a good starting point. Could have put it before the C-section baby one. It was really just a setup to get you to think. Let's just accept that you're not going through it. I'm ok with that now.

2) Hardware. Yes. It's what's the article is about, but honestly I still don't care. You were commenting on AI, and I was responding to that. The chip from the article is kinda neat, but it's just one synapse. I guess I didn't quite get how an array of these is supposed to be different then a neural net that's simulated with a typical chip. It will be faster, sure, but that really isn't what's holding AI back.

3) Every component in a human, body, brain, and mind, will be able to be modified in a controlled and precise manner. Eventually. That doesn't exist yet in the same way that we don't have ability to make AI systems that rival human sentience. We can only make AI systems that perform specific tasks. Much in the same way that we can only control and modify humans to a limited degree. Eventually both milestones will be met. And when we get there, that finely crafted human can be owned. It's called slavery.

Oh, for sure, murder wouldn't be all that bad if we could effectively undo it. Same thing for an AI. Wiping the hardware, and reloading it is largely inconsequential. But permanently wiping it would still be murder of an AI. And wiping it, tweaking it, and reloading would effectively be mind control.
At some point I believe that AI constructs deserve to have rights. If you're not sovereign over your own mind, then your rights are more or less meaningless. Ultimately, I don't want these things done to an AI because I don't want that used as a justification to have it done to me. Anything that approaches the concept of removing my sovereignty over my own mind... I dunno, it just kinda spooks me.

Yeah, I think the initial creation of a human-comparable AI can be unethical. But, it doesn't have to be. I mean, making a baby isn't necessarily unethical. There are some cases where making a kid would be unethical, like if you planned on abusing it, or it was doomed to be nonviable. If you made an AI with a plan of how it would sustain itself and propagate, then it would be fine. I'm sorry, but the more I talk to you about this, the more convinced I am that I'm in the right on this one. I don't see the difference between natural intelligence and artificial intelligence.

It's been a nice chat, but Thanksgiving is coming up, so I'm out for about a week. If it ever comes up, I'll gladly sit on the ethics board that grants you your AI-birth license. Take care till then.

Re:I have my doubts

[CTachyon]

Ethically loaded? How? I don't see how the brain would be suffering? Or are they worried about skynet?

Here in meatspace, putting a human in a sensory deprivation tank is torture and one of the more surefire ways to drive a person insane. The brain isn't wired to believe in null sensory data. If a region of the brain stops receiving stimulation, it frantically strengthens its connections to other regions, tapping randomly into its neighbors and interpreting their arbitrary stimulation as sense data (compare the hallucinations of sensory deprivation to phantom limb syndrome and somatosensory remapping in amputees, e.g. touching an amputee's face triggering sensation on the amputee's phantom fingers because the face and fingers are next to each other in the sensory map of the postcentral gyrus). If this frantic effort fails and the brain regions can't find a source of stimulus, they start to die outright, nerve by nerve, because nerves are wired to commit suicide if they don't fire regularly.

But even a sensory deprivation tank still provides senses of sound, proprioception, temperature, gravity. If we were to create an accurate nerve-by-nerve simulation of an entire vertebrate brain and then provide it with no sensory input.... Well, at best the result would be phantom body syndrome. At worst, it could go well beyond torture and become the greatest suffering ever inflicted on a single sentient being. And we don't understand the brain well enough to know which pieces of sensory data must be provided to maintain sanity and prevent existence from being torture.

And that's not even getting into the legal and moral issues. Let's say we can put a sane mind on a chip by cloning the synaptic structure of a recently deceased human and feeding it all the appropriate sensory inputs. Is it a person? Is it a citizen? Can it consent to a contract? Is it a minor until the hardware turns 18? May I enslave it? Is it ethical to feed it false sensory data culled from a virtual reality simulation, i.e. trap it in The Matrix? If I turn it off and erase it at the end of my scientific study, have I murdered it? If yes, am I legally obligated to keep it powered until the hardware fails? Am I morally obligated to transfer the synapse data to new hardware before the old hardware fails, making the uploaded human immortal? Alternatively, am I morally prohibited from doing that for more than N years, for some value of N? If I'm transferring the mind to new hardware, and my mistake causes a power surge that erases its synapses, am I protected by existing Good Samaritan laws, or have I committed involuntary manslaughter? As it's a simulated human mind, complete with all human appetites, am I obligated to provide it with pornography and the means to masturbate itself to orgasm? Do I have to obtain consent from the human whose deceased mind will be used to create the chip? Am I obligated to pick an atheist, or more to the point a subject who doesn't believe in souls? How well informed does the consenting subject have to be before dying? Does the family have to consent as well, or are we content with provoking another HeLa controversy for the greater good? And so on...

Re:I have my doubts

[Anonymous Coward]

You don't see how it's ethically loaded? Really?

Would the artificial brain have rights? If you wiped its artificial neurons, would it be murder? If you give it control of a physical robot arm and it hurt someone, how and to what extent could you "punish" it? The ethical questions are virtually endless when you start to play "god". I would think that would be obvious.

Believe it or no not we r the gods of our creation. We must use great wisdom in dealing with this creation. Someday this creation will achieve sentience. We had better be ready for that.

I love it! I'm totally on board with this!

[Narcocide]

Ok, so first we build this giant super-fast, high-capacity infant brain and then expose it via a documented API to the public internet so that we can tap into the practically infinite supply of free, unmoderated, user-generated content in order to train it to learn and interact with humans.

There is *no* way that during this process it could go insane and decide to try to destroy itself and/or the world. Completely safe. Yep. Make sure we tell that to the shareholders.

and how...

[Anonymous Coward]

...do they model free will?

Re:and how...

[Narcocide]

That is going to be the fun part. They get it for free with the chip design whether they ask for it or not.

Re:and how...

[holmstar]

You're assuming that we actually have one ourselves.

So to model analogue neurons...

[Viol8]

... they used analogue electronics.

And the radical new technology is what? This was done in the 1960s. Sure, there may be a bit more accuracy and finesse with this version, but really, cutting edge this is not.

Re:So to model analogue neurons...

[Anonymous Coward]

I think you have to credit MIT researchers for knowing better where the cutting edge is than you, and the writers of the article for including the 1960s in this paragraph:

'Previously, researchers had built circuits that could simulate the firing of an action potential, but not all of the circumstances that produce the potentials. “If you really want to mimic brain function realistically, you have to do more than just spiking. You have to capture the intracellular processes that are ion channel-based,” Poon says.'

More than just spiking; from my AI lectures years ago I recall that the McCulloch-Pitts neuron model of the was a spiking model (excitatory inputs, inhibitory inputs, thresholds) etc.

Re:So to model analogue neurons...

[Anonymous Coward]

A decade ago (when I was just getting my bachelor's) I considered doing grad research in neural nets, and was disappointed that all the activity I saw was digital. Since then, I haven't kept up with the field, but unfortunately they may in fact finally be getting back to using analog. But you shouldn't be too surprised by this.

Digital systems can be copied, perfectly. Which is great for running repeatable experiments. An analog system can never be perfectly replicated. Even if we had a full analog model of the brain, we might be stuck right back where we are now with real ones. You'd have to teach it everything from the ground up, like a child. And then you'd probably never actually understand it. You also wouldn't be able to simply copy the intelligence into a new brain. This makes it hard to do research with, so I can see why analog has been avoided.

Theoretically, with enough bits of resolution a digital brain could model an analog one. In practice however, you probably could never practically build a accurate digital neural model of the brain. We simply don't know how much precision is actually required, and it may require so much precision that the size of the circuitry becomes impossible to build without running so slow as to be useless.

Re:So to model analogue neurons...

[tencatl]

That's because you talk about AI, whereas the topic here is computational neuroscience (biological neurons). The Hodgkin-Huxley model, dates back to the 50's, and includes ion channels. Those guys receive the Nobel for that. If you look at the Figure 1 in the paper [1], they draw the small circuit associated with their differential equations. As to the novelties in their approach, one should have a look at their paper, which should be on PNAS, but I cannot find it (first there is no issue of the 14, and nothing in the 15th, second a search for Rachmuth does not lead to any result on the website). [1] August 28, 1952 The Journal of Physiology, 117, 500-544

Re:So to model analogue neurons...

[Anonymous Coward]

AI lectures in a Cybernetics degree go into all sorts of stuff; I don't remember it too clearly (nearly twenty years ago) but it was not just taught from the "blah blah perceptrons can't do XOR blah blah" perspective that was so common. Not a dismissive CompSci perspective, in short. And you're right, the models in the 50s and 60s were impressive works, but only _models_, this is a fair bit closer to implementation of meaningful functionality.

Either way that's not the point I made. The point I was making was that the guys at MIT are variously and vastly more likely to know where the cutting edge is than a random slashdot poster who is dismissing it as lacking innovation, without citation, right?

Even a small scale silicon implementation of Hodgkin-Huxley could be 'cutting edge' worthy of all sorts of publications, and it rather sounds to me like these guys are claiming something substantially more developed than Hodgkin-Huxley.

I just bristled at 'cutting edge this is not', which is, frankly, a load of shit.

some math...

[Anonymous Coward]

1 synapse = 400 transistors.
current gpus = 3,000,000,000 transistors.
so we currently can achive 7,500,000 synapses, assuming that there is absolutely no overhead in having more synapse work togheter...
and those are just the synapse, without the neurons...

human brain: 10^11 (one hundred billion) neurons, and each one has on average 7,000 synaptic connections to other neurons...

we're still far, but interesting work anyway.

Re:some math...

[fsckmnky]

Yeah. I think the fact that they claim, or at least, imply, that their model is complete, is the most interesting point here. If indeed they have a complete model, then this type of circuit could be added to languages such as VHDL and Verilog and put to immediate use, if only on a limited basis, and the scaling of the model will get better as manufacturing does.

It was 2011...

[bassdrop]

...and war was beginning.

That just means it's wrong

[Anonymous Coward]

Relax - that just means that either they implemented a discrete time dynamical system (i.e., the chip's simulation has a discrete clock, which the brain doesn't), or the time constants on their system (ion channel flows, etc) are wrong, or both.

Lots of people have accurately simulated single neurons with hardware components before (see https://secure.wikimedia.org/wikipedia/en/wiki/Hodgkin%E2%80%93Huxley_model for an actual analogue circuit from the 50s) but figuring out the a priori wiring & weights connecting the neurons is far more difficult & they've done nothing to address this. Just consider the fact that our individual neurons aren't that different from other mammals' neurons, but we're much smarter than other mammals. You can't just wire together a million of these chips and have it do anything interesting. "It's the network, stupid."

Re:I love it! I'm totally on board with this!

[admiralranga]

or produce a replica of /b/?

But what about efficiency?

[Pegasus]

It may be faster, but what about performance per watt? You know, the whole brain does everything on only 40-50 watts. How does this MIT product compare to brains in this area?

Re:But what about efficiency?

[Anonymous Coward]

Where is that info from ? 50 watts sounds too much.
I mean you have to eat 1000kcal per day just to keep your brain working then
Google says human need ~ 1000 cal per day for 80 kgs of weight

Re:But what about efficiency?

[Anonymous Coward]

a food "Calorie" is a kcal.

Re:But what about efficiency?

[chocapix]

Wow, I never thought about it that way. A human brain consumes less power than a modern CPU (say, 100W).

Plus, the brain does its own glucose burning and that's counted in the 50W. To compare fairly, you'd need to take into account the PSU efficiency, electrical grid losses and power plant efficiency in the CPU power. If we say 50% efficiency overall, that means 200W for the CPU.

Just wow.

Re:But what about efficiency?

[Tim C]

Two things - firstly, as the other AC has already said food Calories are kcal. Also I don't know about your country, but in mine the RDA for calorie consumption in adults is 2000 kcal, not 1000.

Re:But what about efficiency?

[Dasher42]

I've got another one for you. The muscles needed to interact with the simulation, they can run on mostly or all vegetables!

Re:But what about efficiency?

[chocapix]

I know, I run on mostly beer!

Re:I love it! I'm totally on board with this!

[Narcocide]

Are you suggesting that is in some way fundamentally different from attempting to self-annihilate?

Transistors are not digital

[Ceriel Nosforit]

The way I remember it is that a transistor stops a much larger current from passing through until a signal is put on the gate in the middle. Then the current that passes through is in proportion to the signal strength.

The circuit becomes digital when we decide that only very small and very large voltages counts as 0s and 1s.

Re:Transistors are not digital

[multatuli]

Transistors are analog. Transistor-transistor logic is digital.

Re:Transistors are not digital

[Anonymous Coward]

This, this so much.
Computers aren't digital, we just make them digital through circuit design.

If we wanted to, we could quite easily use ternary computing, or quaternary.
All we'd need to do is redesign the circuits to deal with the differing voltages, and produce better-than-average power supplies that aren't made of cheese strings and sticks, since voltages across boards vary between that 0 and 1 quite often, but not enough to break the circuits most of the time. (unless faulty hardware or rogue radiation decides to attack your stuff)
With binary backwards compatibility mode, you'd not even need to modify the programs that run on it, besides the main kernel, maybe.

Laziness and cheapness prevents this, though.
Binary computers are terrible, just terrible.
Balanced Ternary is a much better system, and it isn't that much harder to pull off than binary is.
The math and logic is pretty easy too. As easy as it gets when dealing with alternate base math.

Re:Transistors are not digital

[Anonymous Coward]

Isn't it also because operating a transistor at saturation or cut-off for most of the time wastes far less energy than when it is operated in its linear region?

Re:Transistors are not digital

[Anonymous Coward]

Transistors are analog. Transistor-transistor logic is digital.

Neurons are analog. What about neuron-neuron-logic?

Re:Transistors are not digital

[multatuli]

Transistors are analog. Transistor-transistor logic is digital.

Neurons are analog. What about neuron-neuron-logic?

Some seem to expect it quantum-mechanical or even completely non-deterministic.
Some are living up to that expectancy ;-)

Computer Chip

[Lord Lode]

Very cool and all, but why does the summary call this a "Computer Chip?

Re:Computer Chip

[amoeba1911]

Perhaps because it's still a bunch of transistors.

Re:Computer Chip

[Lord Lode]

But it's only 400 of them! Even the Intel 4004 had more.

Why try to replicate a synapse?

[Anonymous Coward]

Sounds a bit cargo-cult to me. "This is the way a human brain works, therefore if we want AI, we have to replicate that". Nope. No reason a purely digital AI couldnt exist, without having to simulate the analogue nature of the human brain first. Without that legacy stuff, it'd probably be hundreds of times more efficient, too.

Re:Why try to replicate a synapse?

[FiloEleven]

The term "AI" is just misleading. We have plenty of AI already, and it's done digitally. What most people mean when they say "AI" is "consciousness." The term "strong AI" is also used, but that's unnecessary jargon. There may be very good reasons that a purely digital consciousness cannot exist. There could even be reasons why a constructed consciousness cannot exist. We don't know nearly enough about the nature of consciousness to make an authoritative claim either way. The entire computational theory of mind is pretty suspect, though exploration of the model is still informative.

400 'tors a bit overkill

[Anonymous Coward]

Another university already came up with a way to put "analogue transitors" or "solid state valves" or whatever you'd like to call them, on a chip die. So why the need for 400 binary ones? Because MIT didn't come up with the idea? Is that it?

Never???

[mangu]

The analog nature of the neuron isn't really the key to making "artificial brains" - the problem is simply scale.

Agreed.

We will never be able to produce enough of these chips and tie them together well enough to produce anything conventionally interesting

Shall we cue here all the "never" predictions of the last century? By the year 1900 there were lots of experts predicting we would never have flying machines, by 1950 experts were predicting the whole world would never need more than a dozen computers.

Moore's law, or should we say Moore's phenomenon, has been showing how much electronic devices scale in the long run.

Re:Never???

[FrozenFOXX]

Undoing moderation. Sorry about that.

Plus

[maroberts]

Current GPU systems consume large wattages, so if you have brain augmentation in this way you are likely to need your head connected to a PSU connected in turn to a mains supply.

Re:Plus

[vipw]

They consume a lot of power because they have a fast clock rate. A neural net shouldn't need a high clock rate.

Re:Better long-term goal: replace brains with thes

[Anonymous Coward]

Where we're going, we don't need breathable oxygen and metabolized chemical energy anymore. Time isn't much of a problem anymore either. Maybe we can finally get off of this ol' rock and start doing space exploration for real.

No need to force any meatbags to change their ways either, they can continue to rest on their laurels while we get just enough to get started and then leave them behind. The only exception is that we might occasionally keep in touch for old time's sake and let them know if we find anything interesting, but that's about it.

Putting many together

[rust627]

so we build a synapse, and then link to more, and more, and before you know it we have a "brain",

We could call it a Positronic Brain, sounds catchy, and marketable.

And we really should enforce some rules to prevent a 'skynet' occurrence, not too many rules though,
I'm sure we could distill the logic down to three simple rules ............

They better have insurance...

[LoRdTAW]

A hundred bucks says a woman and her son, black dude and a juice head will break into the lab, blow it up and throw the chip into a vat of molten metal.

All that work for nothing.

Re:They better have insurance...

[fsckmnky]

I hate it when that happens. ;(

Synapse firing event is not pure analog

[wdef]

I might be out of date, but: the event itself requires the neuron's action potential to reach a threshold, then the synapse fires. It either fires or it does not. On or off. But the process of reaching the firing threshold is analog, since the physical geometry of the neuron and of its afferent neural feeds (inputs) determines at what point the neuron will fire. Neurotransmitter quantities in the synapse are also modifiable though eg by drugs and natural up/down regulation of receptors, enzymes or re-uptake inhibition. So a neuron is an analog computer having output with various amplitudes of on/off.

Re:Synapse firing event is not pure analog

[Anonymous Coward]

This is sort-of what I was going to post. Everything I've read about neurons suggests that they work on a mostly "digital" principle and do not really propagate analog information. The analog side of things is that overall brain function changes as chemistry affects the the sensitivity of neurons.

Re:Synapse firing event is not pure analog

[bobaferret]

Here my limited interpretation. The chip models dual receptors NMDA and some cannabinoid. Essentially, from my limited understanding, you have two way for the synapse to fire. A short term spike, which we are all familiar with from basic AI classes. But there is also a second long term spike that can fire due to a long term build up. the NMDA channel is what we are all familiar with, it's this second channel using the cannabinoid receptor that makes these chip models different. You can essentially have two waves propagating through the system independently from each other, yet influencing each other. One based on short ripples in the pond, and another based on echoes from other ripples. Either one can tip the balance at any given time, causing extremely complicated firing patterns, that are difficult to computationally impossible to model digitally. You're always going to have rounding errors... Like I said this is my basic understanding, and is probably way off...

Re:Synapse firing event is not pure analog

[tencatl]

That's forgetting the time domain. Even if the spike is binary (but, for instance, some cells in the retina are purely analog), the timing between spikes really matters. It appears that this timing is controlled by a lot of factors, including subthreshold activity. The timing leads to varying spike frequency, correlations, etc. One should not forget that every neuron is a cell, involving large amounts of (mostly unknown) molecular mechanisms impacting the spike discharges, hence the neural information.

Re:Better long-term goal: replace brains with thes

[mr_gorkajuice]

Makes me wonder. I assume the immortal machine would think it WAS the subject. But would the subject think he was the machine?

Still a long, LONG way to go...

[Pollux]

MIT’s chip — all 400 transistors (pictured below) — is dedicated to modeling every biological caveat in a single synapse. “We now have a way to capture each and every ionic process that’s going on in a neuron,” says Chi-Sang Poon, an MIT researcher who worked on the project.

Just because you finally can recognize the letters of the alphabet doesn't mean you can speak the language.

Re:Still a long, LONG way to go...

[Anonymous Coward]

What kind of chips are they? Can it be used in the usb drive?

Re:Still a long, LONG way to go...

[leptogenesis]

Mod parent up. The linked article (and the MIT press release) are misleading. The closest thing I can find to a peer-reviewed publication by Poon has an abstract is here (no, I can't find anything throught the official EMBC channels--what a disgustingly closed conference):

https://embs.papercept.net/conferences/scripts/abstract.pl?ConfID=14&Number=2328

And there's some background on Poon's goals here:

http://www.frontiersin.org/Journal/FullText.aspx?ART_DOI=10.3389/fnins.2011.00108&name=neuromorphic_engineering

The goals seem to me to be about studying specific theories about information propagation across synapses as well as studying brain-computer interfaces. They never mention building a model of the entire visual system or any serious artificial intelligence. We have only the vaguest theories about how the visual system works beyond V1, and essentially no idea what properties of the synapse are important to make it happen.

About two years ago, while I was still doing my undergraduate research in neural modeling, I recall that the particular theory they're talking about--spike-timing dependent plasticity--was quite controversial. It might have been simply an artifact of the way the NMDA receptor worked. Nobody seemed to have any cohesive theory for why it would lead to intelligence or learning, other than vague references to the well-established Hebb rule.

Nor is it anything new. Remember this story from ages ago? Remember how well that returned on its promises of creating a real brain? That was spike-timing dependent plasticity as well, and unsurprisingly it never did anything resembling thought.

Slashdot, can we please stop posting stories about people trying to make brains on chips and post stories about real AI research?

Re:Still a long, LONG way to go...

[marcosdumay]

"Slashdot, can we please stop posting stories about people trying to make brains on chips and post stories about real AI research?"

Those are two very different research areas, and both quite interesting. I'd vote for the continuation of the status quo, and having stories about both. But I'm willing to let the sensationalist summaries go.

Re:Still a long, LONG way to go...

[Anonymous Coward]

"every biological caveat in a single synapse"?
"each and every ionic process that's going on in a neuron"?

I'd like to read which systems they actually chose to model, thank you very much. Not all neurons are created equally, different types of receptors found on different neurons do very different things postsynaptically. Which ones did they include in this synapse? NMDA & AMPA receptors, K+ & Na+ channels, sure. But calcium influx (caused by transmitter binding) will set off a chain of things that will totally change receptor densities (bringing AMPA receptors from vesicles to surface) and ALSO will trigger retrograde signalling by NO (or by endocannabinoids, they mention that in one of the articles) that shift amount of transmitter release PREsynaptically. It's this huge jumble of positive and negative feedback loops that I didn't know we understood well enough to really model accurately.

I'd love to get my hands on a paper that actually shows what they chose to model with this chip. But I very much doubt that it's as comprehensive as they're claiming.

Re:Still a long, LONG way to go...

[tencatl]

Remember this [slashdot.org] story from ages ago? Remember how well that returned on its promises of creating a real brain? That was spike-timing dependent plasticity as well, and unsurprisingly it never did anything resembling thought.

The only place where the FACETS European project promised to create a real brain was on /. The project goal

was to create a theoretical and experimental foundation for the realisation of novel computing paradigms which exploit the concepts experimentally observed in biological nervous systems

, according to its website [1].

As a matter of fact, this project has been a success, and led to the BrainScaleS European project, which

aims at understanding function and interaction of multiple spatial and temporal scales in brain information processing.

[2]. Again, there is no unrealistic/journalistic promises here.

[1] http://facets.kip.uni-heidelberg.de/public/goals/index.html [2] http://brainscales.kip.uni-heidelberg.de/

Analog = Unique

[Warwick Allison]

If it's analog, then it's behaviour is unique per chip, and so anything you build from them will be subtly unique. So "software" would behave differently depending on the unit it was running on. You thought 4 or 5 versions of Linux was tricky to support...

Re:Analog = Unique

[multatuli]

If it's analog, then it's behaviour is unique per chip, and so anything you build from them will be subtly unique. So "software" would behave differently depending on the unit it was running on. You thought 4 or 5 versions of Linux was tricky to support...

That explains why I always seem to hear things on my radio that other people don't ;-)

Data smoking pot

[morgauxo]

For those who RTFA (or at least clicked) anybody else see his eyes in that picture and wonder if Data had been smoking pot?

Re:Data smoking pot

[Jeng]

With contacts like those I would think your eyes would almost always be irritated.

Caltech Carver Mead built similar things

[peter303]

He made hybrid analog-digital circuits to emulate retina processing. There was some talk of self-adjusting cameras using these, but I've lost track.

Neural Nets and Perceptrons 40 years ago

[peter303]

They built these out of circuits before computers got cheap enough. They had "memory" which implemented training-by-example. Huge debat int he A.I. labs whether they were significant. But they seem to return in some new form every decade.

Re:Better long-term goal: replace brains with thes

[marcosdumay]

There are no two distinct bodies. The GP is proposing piecewise replacement (probably after defects, I wouldn't get them any other way) and improvement. That is not the conventional uploading you see around, and "feels" way different.

Imagine a beowulf cluster of trillions of nodes

[msobkow]

I had to play with the beowulf meme. :)

But really, what good is modelling a single neuron when you'd need billions or trillions of these chips clustered together to mimic an actual human brain?

Fundamentally Analog

[tigre]

The summary is way off. Transistors are analog devices, so TTL may behave digitally but that's only because a lot of work was done to make that happen. All that's happening here is taking analog devices with certain characteristics and using them to model an analog process with certain other characteristics. No small feat mind you.

n/t

[caius112]

Don't mind me, posting to undo accidental moderation

If the brain is all about interconnection

[Dasher42]

If we're talking about large scales of interconnection between these artificial neurons, optical signaling is clearly a good first step. Designing adequate wiring with conventional technology would be a nightmare.

Can you imagine using quantum entangled particles to communicate between neurons? Now that kind of brain would be off the hook!

Obligatory

[patchmonster]

I for one welcome our robotic overlords!

Precedent?

[InsertCleverUsername]

I believe this issue has already been unequivocally settled in Star Trek TNG season 2, episode 9.

If Skynet refuses a wipe/reinstall, just make sure you've changed the crypto on access to our military systems.

Synapses FTW!

[slapout]

So they used their synapses to design something to allow them to study their synapses

Rights

[fyngyrz]

Why does a human have rights, as opposed to a rock?

Rights come into existence when an entity with sufficient power is willing to exercise that power, or makes a convincing show of willingness, in order to guarantee whatever the claim of right is. Everything else is what we call "wishful thinking", ofttimes accompanied by argument.

And if you don't think rocks have rights, try collecting a vertebrate fossil on government land in front of a park ranger. You'll soon learn differently.

Re:Ethics

[waperboy]

It all comes down to how good a grasp on reality you have (or, if you will - how psychopathic you are ;).

If you acknowledge that the artificial brain is just a collection of interconnected silicon, the wiping of its neurons is just a physical process to meet the wipers needs. Just like the human brain is just a bunch of interconnected cells, and the wiping of it has no objective value attached to it.

Humans have created a mental construct from the notion of emotional suffering, though. If you murder a human, people around that human may suffer or feel afraid, and that's really the only rationale for why murder is bad. For the murdered human himself, it is of no consequence - the brain has ceased to function, and thus he cannot have an opinion about being dead.

Re:Better long-term goal: replace brains with thes

[FiloEleven]

It's certainly an interesting thought experiment on the continuity of consciousness, but it's important not to ignore the implications of "conventional uploading" or the construction of a physically identical duplicate. The former case is IMO easier to deal with philosophically, but if this sort of thing were ever to become a reality then the latter case also must be considered. I have concluded that they will remain in the realm of imagination, but it will be fascinating if I'm wrong.