Slashdot Mirror
Ray Kurzweil Does Not Understand the Brain
jamie writes "There he goes again, making up nonsense and making ridiculous claims that have no relationship to reality. Ray Kurzweil must be able to spin out a good line of bafflegab, because he seems to have the tech media convinced..."
The singularity is to nerds what the rapture is to fundamentalist protestant wackjobs....
Would be nice if the summary even hinted at what the ridiculous claim actually WAS...
Namely, that we'll be able to reverse engineer the human brain in the next 10 years.
If the complain here is about how media laps up these 'pundits', well, welcome to new age media. David Pogue is very popular too. But some of the nasty stuff he writes which are factually wrong goes completely un-noticed.
...he must be right! He used math, and everything! I'm a little shocked that Kurzweil equates blueprints with the functioning organ. I am not shocked, however, that the tech media latched onto this--at first blush it sounds so *reasonable.*
Kurtzweil is just the latest in a long line. How do you think publications like Fast Company and Wired get written?
And does anybody remember JonKatz?
Try reading the rest of the article, not just the pretty quoted stuff...
His latest claim is that we'll be able to reverse engineer the human brain within a decade
Amateur. I could put something together to simulate the human brain in about 8 months.
(Plus another 3 minutes at the start)
Summation 2
FTFA: The end result is a brain that is much, much more than simply the sum of the nucleotides that encode a few thousand proteins.
Likewise, the end result of a computer is much, much more than simply the sum of the commands that encode a CPUs instruction set.
Sejnowski says he agrees with Kurzweil's assessment that about a million lines of code may be enough to simulate the human brain.
Here's how that math works, Kurzweil explains: The design of the brain is in the genome. The human genome has three billion base pairs or six billion bits, which is about 800 million bytes before compression, he says. Eliminating redundancies and applying loss-less compression, that information can be compressed into about 50 million bytes, according to Kurzweil.
About half of that is the brain, which comes down to 25 million bytes, or a million lines of code.
Idiot. The design of the brain is encoded in the genome in the same way that the design of a 4KiB program is encoded in its load module: useful for running the program on its original hardware.
But then you have architectural issues. That 4KiB of information does not run unless it's supported by a complex operating system, which itself is supported by complex logic in an ISA and memory managment architecture backing it up. And all that is implemented on a specific design in a specific physics model.
Translating that program to SPARC takes work, and it comes out roughly the same size. Translating that program to a progression of chemical reactions produces something vastly different, especially since you need a new middle ware (chemical environment) running on top of different physics (chemistry).
Translating a physical architectural design from chemistry to computer logic on top a given ISA is the same problem. You now have odd issues that are messy, and then the program running on the brain needs to be built again. That program is even more complex and less known.
Support my political activism on Patreon.
I'm not even sure what to say about this statement
I wish I could get a job as a futurist....think about it:
"What do you think is going to happen in the future???"
"Um...dogs will bring soda to you when you whistle a Cradle of Filth song?"
"OMFG THATZ BRILLIANT. HERE R MONIES, PLZ HAS MAH BABBIES!"
Living With a Nerd
Ray Kurzweil is yet another computer programmer blathering on about things that he has no understanding on. The vast majority of software people I know do that, I don't understand why this guy gets to publish books on it.
Tesla was a genius. Edison however was a overrated hack who liked to torture puppies.
Ray Kurzweil has been making claims for AI for years. For example that we will have an AI singularity event and that
society will be completely replaced my machines. Well, decades later it still hasn't happened and the only things in the
field of computer science that seems to have a life of its own are spam and computer viruses. I'd like call them a
life form.
Will we reverse engineer the brain any time soon? I doubt it. Part of the reason is practical. This would be an
extremely expensive and time consuming undertaking. I'm not sure its even worth it, especially when this is
compared to other branches of science which have made rapid advances. For one example, take a look at
the field neuro-science and its use of fMRI scanning.
Reversing engineering the brain, probably is possible, but it's probably not worth it right now. Well have
to wait another decade.. again.
---- It won't be as bad as you fear or as good as you hope, but it will take twice as long as you plan.
What Kurzweil doesn't understand could fill a book.
Several in fact.
In Capitalist America, bank robs you!
Ray Kurzweil is yet another computer programmer blathering on about things that he has no understanding on.
Get Kurzweil a slashdot account, stat!
Would be nice if the summary even hinted at what the ridiculous claim actually WAS... Namely, that we'll be able to reverse engineer the human brain in the next 10 years.
It's a little more complicated than that. You see, the article actually breaks down the logic behind that statement and points out how poor it is. Here's the initial part of Kurzweil's argument:
Sejnowski says he agrees with Kurzweil's assessment that about a million lines of code may be enough to simulate the human brain.
Here's how that math works, Kurzweil explains: The design of the brain is in the genome. The human genome has three billion base pairs or six billion bits, which is about 800 million bytes before compression, he says. Eliminating redundancies and applying loss-less compression, that information can be compressed into about 50 million bytes, according to Kurzweil.
About half of that is the brain, which comes down to 25 million bytes, or a million lines of code.
I have only taken high school biology but I know that the genome doesn't magically become the brain. It goes through a very complex process to amino acids which fold into proteins which in turn make cells that in turn make tissues that in turn comprise the human brain. To say we fully understand this transformation entirely is a complete and utter falsity as demonstrated by our novice understanding of the twisted beta amyloid protein that we think leads to Alzheimer's. How amino acids turn into which proteins I believe is largely an unsolved search problem that we don't understand (hence efforts like Folding@Home). And he claims that in ten years not only will we understand this process but we will ... reverse engineer it?
The man is insane. I've posted about this same biologist criticizing him before and it looks like P.Z. Myers just decided to take some extra time to point out how imprudent Kurzweil's statements are becoming. Kurzweil will show you tiny pieces of the puzzle that support his wild conclusions and leave you in the dark about the full picture and pieces that directly contradict his statements. This is a dangerous and deceptive practice that -- despite my respect for Kurzweil's work in other fields -- is rapidly turning me off to him and his 'singularity.' He's becoming more Colonel Kurtz than Computer Kurzweil.
My work here is dung.
"Read it. Other than the solid date he predicts, it's pretty plausable."
No it's not. If it was possible to do in a million lines of code, it would have been done by now. Windows XP had something like 40 million lines of code. While we can agree it was probably coded relatively inefficiently, there is no way that any OS even comes close to the complexity of the brain.
Let's see. On another recent article it was stated that the average car has several million lines of code running in it. I haven't come across a sentient Prius yet.
And there's that pesky parallel processing the brain does. I don't think that a rack full of Nvidia Tesla cards can approach the average two year old's parallel processing capability.
I agree, Kurzweil is smoking something and not sharing.
Aside from the article, it would appear arbitrary to apply loss-less compression to the LOC.
Code must be very very compressible losslessly (I am betting like 90%, as plain text is often 80% when zipping). This would imply to me that one would need 10 times as many LOC as the (faulty) premise assumes.
The article itself points out that it is not just a matter of writing the code, but also simulating the machine. So yes, if we could accurately make a machine (real or virtual) that could compute the way that DNA computes perhaps we could then make a brain that functions in it with not too much code, but it does not follow that we can just a tersely describe it on a computer as we have them (Turing Machine?).
Wow, sent an e-mail as suggested when clicking on "use classic" banner, and got a fast response that addressed my msg
Ray Kurzweil never says we'll be able to write software to emulate the brain completely in ten years. He just says we'll have reverse engineered the signals we have to send to the cortex to interact with it. Reverse engineering involves analysing and understanding something. Reengineering anew would be the next step. But then again, when you're only interested in reverse engineering the various inputs you can give to the brain to get different results, you're still very far away from understanding how to build a brain.
PZ Myers clearly doesn't understand reverse engineering and writes up a useless article based on his erroneous comprehension of Ray Kurzweil's prediction.
What's so crazy about that?
A single line of code has more redundancy than our genes. There are very good reasons why duplicated genes survive.
I haven't read the article yet, but Ray Kurtzweil is a technology speculator - like a sci-fi writer except that he doesn't make up a story to go with his ideas and tries harder to convince people they're actually going to happen. He wrote "The age of intelligent machines" and "The age of spiritual machines" where he takes a hard AI stance that computer thought can become indistinguishable from human thought. He is also a proponent of technological singularity.
Generally his ideas aren't taken very seriously by academia in Computer Science, or at least that has been my experience. The philosophy department at my university sometimes enjoyed going over his ideas; but the philosophy department at my university was very fond of pseudoscience.
Sejnowski says he agrees with Kurzweil's assessment that about a million lines of code may be enough to simulate the human brain.
Kurzweil explains: The design of the brain is in the genome. The human genome has three billion base pairs or six billion bits, which is about 800 million bytes before compression, he says.Eliminating redundancies and applying loss-less compression, that information can be compressed into about 50 million bytes, according to Kurzweil.
Dude, the equations of quantum mechanics can be written on one page. General Relativity can be written on a second page. What more do you need ? Clearly, a few hundred lines of code (and a few do loops) should be enough to simulate the entire universe, brains and all.
Glad we cleared that up. All you physicists and astronomers can go home now and work on your resumes.
To simplify it so a computer science guy can get it, Kurzweil has everything completely wrong.
Best summary I've ever read. (Though a bit more respect for our ability to get it would have been nice.)
Well, I suspect that in the right (machine) language it could very well be possible to put together the necessary instructions for generating a perfectly simulated brain, the problem is that we have little clue when it comes to what instructions to feed our imagined brain simulation hardware. Also, when/if we do figure out how to simulate a brain accurately our code will most likely at first be very crude and un-optimized.
Greylisting is to SMTP as NAT is to IPv4
but I've always had the impression that his 'philosophy' is inspired by Star Trek rather than the scientific method or logic. His AI claims always seem to be a vast underestimation of the complexity of the human mind. "Wishful thinking" is the best phrase to describe his ideas. Regardless, the summary for the article could have been an actual summary rather than just a copy-paste of the first couple lines.
"From the depths of my skeptical and rationalist soul, I ask the Lord to protect me from California touchie-feeliedom."
PZ Myers threw a red herring there. What Kurzweil says is pretty reasonable, he used the total amount of information in the genome to get an upper limit estimate of the amount of library code needed to simulate a brain. I say "library" to differentiate from data, since a lot of our brain information comes from our experiences, i.e. library == instincts.
Myers goes off in a tangent about biochemistry which has nothing to do with the argument. I've never read anything hinting that the way to simulate a human brain would be to simulate how the molecules in the brain behave. We don't build airplanes with flapping wings either, machines can emulate the functionality of a living being without need to simulate the exact details.
From the number on neurons in the human brain, considering how many interconnections there are and how fast the neurons can fire, I think a machine with one million processing cores at 1 GHz would have approximately the same data handling capacity as a human brain. The rest is software. Neural network software is pretty much routine stuff, the tricky part is learning what are the interconnections between the neurons.
If you read the article closely, the engineering is designed to understand operation, not create a grand simulation. You can all quit toting your brainless responses now.
I should add, I don't think it is unreasonable to think we can simulate the brain in 10 years, but I don't think looking at DNA is the way to do it.
More likely would be actually reverse engineering the brain by looking at brains, and simulating neurons in software, or even hardware.
If it is done at first i would imagine static brains with limited learning ability (unable to create new neurons, only adjust connections in existing ones). Later then ones able to create new and restructure. Eventually I imagine a simulated brain will be able expand like a child's, but indefinitely.
None of this means necessarily we will have hardware that can keep up, and do the processing in real time with a software simulation.
Wow, sent an e-mail as suggested when clicking on "use classic" banner, and got a fast response that addressed my msg
...Kurzweil's assessment that about a million lines of code may be enough to simulate the human brain.
If you accept that, then the real problem to solve becomes: in what language do you write the code?
;-)
I live ze unknown. I love ze unknown. I am ze unknown.
To simplify it so a computer science ignorant biologist with a tendency to inane rants can possibly get it, you don't need to simulate electrons in a semi-conductive material at specific temperatures in order to build a complete working emulator for an old computer.
"I love my job, but I hate talking to people like you" (Freddie Mercury)
Eliminating redundancies and applying loss-less compression, that information can be compressed into about 50 million bytes
This is so retarded that it's sad. Why is it so hard to understand how compression algorithms works? Saying that X can be compressed into Y bytes doesn't say ANYTHING. You can "loose-less compress" ANYTHING into 1 bit by using the function that takes that and returns the bit "1" (and which takes anything else and returns "0" + that). What does compression has to do with anything? The stupid hurts...
...except, maybe, Pinky.
Guaranteed! This comment 100% Anthrax free!
Who the hell is Ray,
Some guy that's wrong.
and what is the claim?
Who cares? It's wrong anyway.
What do you mean "infinite"? The human brain is composed of one hundred billion or so neurons. Looks like it's pretty much finite to me. I have ten times as many bytes of information in my hard disk.
"If the brain were simple enough to be understood, it would be too simple to understand itself"--anonymous
If all of our folly were turned to intelligence and divided amongst a thousand toads, each would be more intelligent than Aristotle
"Gentlemen, you can't fight in here! This is the War Room!" -- Dr. Strangelove
He actually has one.. And he's a dick, too.
Here you go, you chauvinist douche bag: http://www.ccmr.cornell.edu/education/ask/index.html?quid=1260
He's doing the calculation the wrong direction!
The genome contains enough information to build the brain from raw materials. However, this data has already been losslessly compressed by countless generations of evolution. We would need to discover the evolved compression algorithm to "unpack" the 800 million bytes into the 3.2 billion bytes (using his factor-of-4 ratio) in order to begin understanding it.
Scanning down the comment list, it looks like every (+2 or more) comment has read the article and is quoting from it -- what has happened to the slashdot I knew and loved?
I mod down anyone who says "I will be modded down for this", regardless of the rest of their comment
But he understands the psychology behind selling vaporware.
There are no loopholes. It's either legal or it's not.
The human brain could be simulated with a million lines of code, however it would need to be run inside an emulator that simulates physics perfectly and has enough RAM to store the quantum state of the aprox 1.5*10^26 atoms that make up a human brain.
If you didn't slap your forehead and mutter "Oh god!" when you read the name "Ray Kurzweil", you don't belong on Slashdot. The very mention of his name is a shorthand for "Abandon logic and plausibility, all ye who enter here".
No folly is more costly than the folly of intolerant idealism. - Winston Churchill
Which account is it?
PS: Go fuck yourself.
What's so crazy about that?
I'd say the crazy part is comparing the genome with "lines of code". Those two have really little to no relation, as one is data and the other is code and you can't really translate from one to the other. Also you won't be able to simulate a brain starting from the genome any time soon, as the complexity to simulate the complete development cycle would be a good bit more then a computer can handle (they already struggle with folding a single protein, good luck with simulating the whole human body).
However I don't think there is anything crazy in the basic claim. To simulate the brain you don't need to start with the genome, you can start at a much higher level, take neurons and the surroundings and simulate that instead. That way you might be able to get an artificial brain up and running in far less time then starting with the genome. Your favorite console emulator doesn't start at simulating the electron or individual transistor either, as those are really just implementation details that are not needed to replicate the functionality. The human brain is of course more messy, so it will be a good bit more complicated, but it is very likely that it is still the higher level structures that count, not the low level details.
Here you go, you literal bore: http://www.lmgtfy.com/?q=facetious
Amateur. I could put something together to simulate the human brain in about 8 months.
It's easy if you limit yourself to simulating Englishmen named Arthur Dent, actually. You just need to have it return the following to queries:
"What?"
"I don't understand!"
"Where's the tea?"
Piece of cake.
Large systems do not require lots of code. Complex systems do. The brain is made up of neurons that are largely identical, arranged in a number of patterns. We have all the basic building blocks to make a brain already. We just don't know how to put it together.
Amount of DNA is not a good way to estimate the number of lines of code needed. DNA is not an efficient encoding. It doesn't need to be. There's millions of years of legacy crap in there for lungfish compatibility and stuff. It doesn't cost anything to keep around and doesn't need to be maintained so it remains in there.
if you rtfa you'll see that the million lines of code only gives you the proteins that make up the brain - i.e., it gives you a parts list and a delivery schedule, not a set of assembly intstructions. The genome doesn't give you how the proteins interact, in usually complex ways (i.e., three or more proteins interacting simultaneously), in billions of cells in parallel, over the course of 9 months to give us an infant brain (even leaving aside the tremendous amount of development that takes place in the brain during childhood).
As the author of tfa writes: To simplify it so a computer science guy can get it, Kurzweil has everything completely wrong. The genome is not the program; it's the data.
IOW, the program is the developing organism itself, the complex protein interactions and it's (uterine) environment none of which are encoded in the genome. The organism uses the data encoded in the genome to produce proteins which interact with each other and the organism and its environment to grow cells which eventually form a brain.
The mistake in Kurzweil's thinking is the typical mistake engineers make when dealing with biology; the enviroments into which engineers place their designs do not typically spontaneously cooperate in the construction of the engineer's design. When an engineer designs a circuit board, his lab bench doesn't spontaneously start soldering connections and adding components for him and automatically complete parts of the design
without his explicit instructions. But the organism does precisely this with proteins syntesised from the genome.
As a result, the genome alone cannot possibly tell you how to "make" an organism, because the genome only tells you the parts list and delivery schedule for the organism, not the assembly instructions. The assembly instructions are not explicit anywhere in the system; the assembly instructions are implicit in the combination of the complex behavior of the cells of the developing organism, the uterine environment and the very complex ways the proteins sythensized from the genome interact.
In order to extract the actuall assembly instructions we'd need a full blown molecular biology simulator that could correctly simulate:
1. protein folding (still unsolved)
2. comlex multi-protein interaction (still unsolved)
3. simultaneous behavior and development, (i.e., in parallel) of billions of living cells each undergoing trillions of chemical reactions per second (computationally prohibitive)
IOW, it's not going to happen in the next 10 years.
The summaries tend to be nothing more than the submitter taking the most polarizing sentence/paragraph from TFA and pasting it into the summary field
Doing more than that takes time and if you are to do it right, someone else has submitted the article with the shit summary.
I have given up submitting articles.
RIP America
July 4, 1776 - September 11, 2001
No it's not. If it was possible to do in a million lines of code, it would have been done by now.
That's silly. He's not claiming that just any million lines of code will do. You need to understand how the brain works in order to write the right million lines of (probably ridiculously compact and completely unreadable) code.
I don't think the claim is entirely implausible; 25MB of code may well suffice to simulate the human brain if it was written in something like brainfuck.
I do however disagree with the assertion:
The difficulty in truly understanding the genome is that it's both program and data.
I'm not sure what it is about his claims that are supposed to be so ludicrous. For example, a million lines of code seems at least plausible, as long as we bear in mind the following:
1. We're not trying to mimic the brain at the protein level, rather at the broader, inter-neuron level (and whatever complex intra neuron behaviour we discover).
2. The million lines of code don't need to encompass the capacity of the brain, just its general neural architecture and adaption rules - there will no doubt be many gigabytes (terabytes?) of working memory, which would actually store the neural connections and whatever parameters they may have.
To be honest, the authors of this article seem to be rather too cocksure in dismissing all this. Even the apparent agreement of Terry Sejnowski (co-inventor of the boltzmann machine) doesn't give them pause. I'm not that familiar with Kurzweil's predictions, but this seems fairly reasonable to me.
There is a google tech talk by Geoff Hinton on restricted boltzmann machines, (a sort of stochastic neural network) that's well worth a watch, for those that are interested. They are considered biologically plausible, and he seems mostly to apply them to machine vision tasks.
Ray's documentary about the Singularity has been touring the national film festivals along with Ray. I saw it June. Its begins as a dull-talking heads piece about the current state of A.I. Many of the Big Name A.I. Scientists are interviewed. Then it transitions into a crime-drama story about the legal rights of A.I.s. That part was more interesting, since it had a story. The film is full of special effects to advance the story. Although I know most of the film to be factual, I suspect it will look like a scifi movie to the average audience member. I think Ray is seeking looking at broader distribution on cable television or arts theaters after the festival run.
Ray was interesting in person during a film-makers Q&A. He reminded me of Woody Allen, but more confident and intelligent. He was graduated from M.I.T. about decade before myself. I personally believe in the Singularity, but more likely in centuries rather than decades.
Um, we don't know everything the brain does let alone how it does it. Not being able to spec the requirements of this software pretty much means it will never do what he claims. That being said, if we make this an open source project with one of the truly convoluted licenses I'm sure we'll be able to accomplish it by the time the license situation is resolved by the courts.
;-)
I now have a decade to patent this software methodology so I can sue everyone who has a brain that works like the software does
unless you buy a few cases of these longevity shakes.
Bitches.
As a result, the genome alone cannot possibly tell you how to "make" an organism...
Untrue. The genome gives you the instructions for a self-modifying program that eventually produces a human brain. All biology experiments so far seem to point to the fact that the DNA chain indeed contains all the information needed to build an organism.
If at first you don't succeed, skydiving is not for you
[citation needed]
Haha, see what I mean?
The brain as defined by the DNA is similar to technical blueprints for a hard drive. Even if you manage to build one, you still don't have an operating system or data on it, just the writing tools and surface to host them.
There are currently four academic disciplines working on the reverse engineering of a human mind. Linguistics, psychology, computer science, and philosophy. You can count neurology too if you want to start talking about the *actual* brain. Several tens of thousands of individuals are directly and indirectly working on this problem. We've come a long way in the last few decades. Unfortunately, we have a pretty long way to go. For the moment we lack a model which accurately describes how mental processes work. There isn't even a consensus on how the processing is done.
"modeling the brain" is not even really the hard part. One only needs sufficient computing power to model what they *think* is going on logically (there isn't even a consensus here). The trick is modeling the mind. We are very, very far away from that.
A fun number to throw around is how many synaptic connections are present in the brain. Synaptic connections are widely believed to be the best indicator of overall memory storage and processing speed (to an extent). There are about 10 to the 15th (Peta I believe?) synaptic connection in a normal human brain. A significant number of these are active at any given time. In other words, the brain is performing a HUGE number of "calculations" simultaneously at all times. Modeling just the hardware is obviously not easy... modeling the software is currently not possible. I doubt it will be in the next 50 years.
For a good read on what many cognitive scientists think is going on, though it is clearly not an accurate model but rather a best guess, go read up on "connectionism".
I imagine that is you try to write a program that simulates a sufficiently simple brain, it might be possible. A perfect example of such a brain is his brain of course!
[citation needed]
http://science.slashdot.org/comments.pl?sid=1757102&cid=33277532
Posting in an epic thread!
... is a car analogy:
Ray Kurzweil does not understand how to parallel park.
__ Someday, but not this morning, I'll finally learn to use the preview button.
I wonder if Kurzweil is being misunderstood here. Would not most of you agree that an entire person is described by the genetic data contained withing one stem cell? Or at least within one sperm cell and one egg cell? Then why is it difficult to believe that the blueprint could be easily describable and compressed in a few million bytes? And if its possible to do this, then why would it not be possible to create some kind of molecular emulator that can emulate reconstructing a living thing from this blueprint?
I can imagine a computer program that emulates molecular interactions. I can imagine that this program that has a DNA blueprint for an entire person. Basically then, give the program the correct inputs (food, oxygen, water) and necessary outputs (waste), and see what the computer program produces. Now speed the whole process up because you're doing it in software.
You may not get an adult human, but you might get a 9-month old fetus. I just don't know.
If the above is possible, then it becomes possible to play around and experiment with a "live" human brain being emulated in software.
Why is this hard to fathom?
I know, we're talking about a monster piece of computing power to be able to do anything near that right now. But why is it beyond our understanding?
"They said I probly shouldn't fly with just one eye," "I am Bender. Please insert girder."
Kurzweil doesn't explicitly say that we'll be able to reverse engineer the human brain because it will run on only "1 million lines of code." What he probably meant to say is that we'll do it in 10 years because all the other factors that will go into it (increase in brain scanning technology, computer simulations, computing power etc.) will allow us to reach that point in 10 years. The original Gizmodo/Wired article which TFA article points to includes Kurzweil's claim that we "only need 25 MB, or a million lines of code" to simulate the human brain. While that (admittedly incorrect) element is part of Kurzweil's argument, it doesn't necessarily negate a claim that we'll still be able to build simulations of the brain in a decade.
The original Wired/Gizmodo article that PZ Myers points to is focused on the ability to emulate the software of the cortex within a supercomputer.
The key to reverse-engineering the human brain lies in decoding and simulating the cerebral cortex - the seat of cognition. The human cortex has about 22 billion neurons and 220 trillion synapses. A supercomputer capable of running a software simulation of the human brain doesn't exist yet. Researchers would require a machine with a computational capacity of at least 36.8 petaflops and a memory capacity of 3.2 petabytes - a scale that supercomputer technology isn't expected to hit for at least three years, according to IBM researcher Dharmendra Modha. Modha leads the cognitive computing project at IBM's Almaden Research Center. By next year, IBM's ‘Sequoia' supercomputer should be able to offer 20 petaflops per second peak performance, and an even more powerful machine will be likely in two to three years. "Reverse-engineering the brain is being pursued in different ways," says Kurzweil. "The objective is not necessarily to build a grand simulation - the real objective is to understand the principle of operation of the brain." Reverse engineering the human brain is within reach, agrees Terry Sejnowski, head of the computational neurobiology lab at the Salk Institute for Biological Studies. Sejnowski says he agrees with Kurzweil's assessment that about a million lines of code may be enough to simulate the human brain.
That last line is probably wildly incorrect, but it doesn't really change the basis of the argument. It's not infeasible that we could reach this in a decade. Look at this TED talk from 2009: http://www.ted.com/talks/henry_markram_supercomputing_the_brain_s_secrets.html where Henry Markram claims he's able to simulate a single neocortical column on a neuronal level in a supercomputer.
Now go back to TFA:
I'll make a prediction, too. We will not be able to plug a single unknown protein sequence into a computer and have it derive a complete description of all of its functions by 2020. Conceivably, we could replace this step with a complete, experimentally derived quantitative summary of all of the functions and interactions of every protein involved in brain development and function, but I guarantee you that won't happen either. And that's just the first step in building a simulation of the human brain derived from genomic data. It gets harder from there.
PZ Myers is probably correct about that: we won't be able to plug in an amino acid sequence into a computer and then figure out what it looks like in 3D, and how it interacts on a molecular scale. But that argument doesn
The code is simple.
Simulate_Brain();
Now just find the compiler with the right set of libraries that can compile it. And yes, I am NOT just being anal. Half a million lines of code is MEANINGLESS. Quickly, how many lines do you need for a "Hello World" program? In assembly? C? Java? PHP?
If one day someone designs a cpu with a built in Hello World function, then it would require what? 2 instructions in assembly? Meanwhile the java guy will be pounding out yet another page of code.
MMO Quests are like orgasms:
You may solo them, I prefer them in a group.
It's what zombies have for dinner!
Obviously, by your logic, a free market economy is impossible, Our economy is too complex to have evolved on its own. In fact, it is far more complex, with far more different parts, than a human being. It must have had a creator. If most any part of the economy, like the steel industry, say, were removed, the economy would not function. How did the economy function before there was a steel industry? Obviously, it couldn't, and therefore we have demonstrated irreducible specificated complexification or something.
All this free market talk is obvious bullshit, and we actually DO have a centrally planned economy because it is impossible for something so complex to have evolved without a central planner.
- None can love freedom heartily, but good men; the rest love not freedom, but license. -- John Milton
"If it was possible to do in a million lines of code, it would have been done by now." That's a pretty tremendous leap in logic. The only way you could know this for certain is to check every million line program and see if it simulates a brain, which isn't ever going to happen, at least not in this universe.
I mean the brain is just a series of tubes, right? How hard can it be to replicate the brain? I bet someone already did it with Logo Mindstorms. Or Play-Doh! or something. Some of you guys are just so defeatist and negative.
http://www.acetonestudio.com
Agreed. Everything I have learned about brains, neurology and how "they" work brings out a simple reality that digital electronic computers (and all the software that runs on them) are too different from brains and how they work. To parallel a bit, let's compare hydraulics (you know, a series of tubes?) and electronics. While it is true that just about everything that can be done in digital electronics can also be done in hydraulics, "Moore's law" applied to hydraulics never had a chance. (It's not hard to imagine why, when presented with a choice, hydraulics was passed over in favor of electronics.) At some level the two are equally capable, but when it comes to application, they are each better suited to different things.
Now comparing brains and computers isn't even as close in comparison as hydraulics and electronics. And that's where the real problem of understanding begins as far as I'm concerned. Electronics and electronic computing are presently built around the notions of precision, accuracy and reliability. Brains reflect none of these ideals. The ideals of the brain are in relativity, approximation and learning. Electronics and programs running on electronics are not particularly well suited to behaving in the way the brain does. And the more I learn about the brain, the more wild and different it becomes.
Kurzweil clearly does not understand how learning works or he never would have made such assertions as "copying the brain." The brain is noisy and chaotic. If a computer were made to emulate such behavior, it would be slow, extremely power hungry and inefficient.
So how could Kurzweil's dreams come true? Simply, a new type of electronics and computing has to be created first. And if they did that, I guarantee you it would never be able to run Windows.
Some of the brain is encoded in our genomes, and some is encoded in our environment. That means starting with the environment of the womb, which we already are nowhere close to understanding in detail. Then there is environment outside the womb, which is different for everyone and changes every second as well. To give a concrete example, DNA does not encode what happens to a brain when you add alcohol to a fetal environment. And that's just one tiny molecule. What about all the other potential inputs? Modeling the brain might very well be possible, but you can't just reduce it to a compressed genome.
Then there's the other problem, which is that not all parts of the genome carry the same informational load. Truth be told, we don't know how to quantify the information in a given DNA sequence. Some appears to be filler ("junk," aka we don't know what it does yet), some is silenced, some is read in both directions, some serves a structural purpose for shaping the chromosome itself, etc. The amount of repetitive DNA in human genomes is really high, so of course it will compress down. But our genomes aren't just messages that can be reduced to Shannon-level snippets - we might know the message, but we don't know who receives it.
How about we start with viruses first, since they have the most compact genomes? Put a viral genome in a simulated cell environment and see what pops out. Oh wait, that's really hard! Sorry Kurzweil, but you're not going to be immortal.
You can lead a horse to water, but you can't make it dissolve.
First off, Ray Kurzweil doesn't want to die. That's a preoccupation that a lot of people have (including one of his critics, Rudy Rucker, who has written whole books hoping to find immortality in the fourth dimension), and it leads them to some pretty fantastic conjectures from time to time. It's not necessarily a bad thing, as long as you keep the proverbial grain of salt handy. Modern chemistry and its not insignificant contributions to our vastly expanded lifespans arose from the alchemical search for immortality. Alchemy was bullshit, of course, but the incidental discoveries of alchemists on the way to their illusory elixir of life paved the way for the real science to follow and build upon after it had ejected the dross.
And secondly, I don't think it's entirely implausible that we can eventually design hardware and software that will match and exceed the performance of the human brain. Our brains, after all, are the end product of evolution, and like pretty much every other part of our bodies, an accumulation of kludges that were just good enough to get passed to the next generation (or not bad enough not to get passed on). It's also implemented using hardware so unreliable that it wouldn't function at all if it wasn't constantly repairing itself, and even then, no matter how well you treat it, it irreparably craps out after about 75 years. And it still doesn't work all that well -- ever seen the long chain of train wrecks that is the history of human civilization? We might be able to engineer something that works a lot better. Granted, it's not going to be by deriving simulated human brains from a copy of the human genome. More likely, it will be very much unlike the way biological brains work.
The fundamental problem, which I think smart and optimistic guys like Ray Kurzweil are particularly prone to forgetting, is that it may not be possible for a mind to understand a mind of equal complexity, i.e., humans may lack the necessary intelligence to duplicate their own intelligence. That will force us back on genetic algorithms to evolve AI, leading to an end product that will likely be just as badly undesigned as natural brains. Worse, it will do little to advance our understanding of how minds work: if we can't reverse-engineer our own brains, we probably won't be able to reverse-engineer even more sophisticated artificial minds, nor will they be able to reverse-engineer themselves. (We can hope that they could reverse-engineer us, and then explain it to us in terms we can understand, if such terms exist, but that takes us so far out on a conjectural limb that I can see Ray Kurzweil from here.)
Anyway, there's room for bold conjectures. That doesn't mean that when Kurzweil completely fails to understand the way molecular biology works that we shouldn't call bullshit on it, but we shouldn't be entirely hostile to futurist speculation. By nature, most of it will be bullshit, but a lot of progress in unexpected areas has been made in the pursuit of mirages (alchemy leading to chemistry, astrology leading to astronomy), and explaining (or discovering) why a conjecture is bullshit is a beneficial exercise in and of itself.
Proud member of the Weirdo-American community.
because it's an odd-looking creature that seemingly has 'random' bits and pieces from various other animals... the full question being "what specific selections in 'natural selection' led to this particular evolutionary path?"
So the question isn't to 'explain' the platypus.. that would be like asking to explain the number 5 or explain the color red.. the question itself doesn't make any sense without being more specific.
It's also not a question of 'why does the platypus exist?' - natural selection was already the answer, and can even be thought up by people on their own; clearly if it exists, it had some benefit being exactly the way it is within the environment it is in.. if it weren't well-adapted to that environment, it would have died out a long time ago (presuming the environment didn't radically change).
To be honest, the fully expanded question is actually an interesting one - and one which biologists and others continue to try to answer to more detail to this day. I hadn't actually looked into Platypus info since I was a kid (a school project on its venom, along with other animals' venom), and wiki tells me it was only discovered in 2004 that the Platypus has -10- sex chromosomes, and its genome mapped fully only as recently as 2008. Seems to me there's plenty of questions left.
Nothing crazy, I think the point was that Kurzweil's napkin math is total bullshit. But then, if you read Kurzweil's article, that's really not the point.
Kurzweil isn't a biologist, he's a computer scientist interested in information theory. The model of the human brain that is interesting to computer scientists has shown value in solving computer science problems, it's explained in introductory texts that this model has no significant relationship to actual biology, except that it is loosely modeled after it. Understanding the brain better may lead also lead to better computer science models which may do Great Things. It seems like an attractive (albeit unproven) theory that our DNA determines the approximate construction of the human brain in terms of how the 22 billion neurons are connected to each other through the 220 Trillion synapses. We know from neural networking theory that the organization of neurons and synapses absolutely affects the utility of the network, and from 50 year old gross anatomy that the brain is not a uniform collection of networks. There is organization and specialization in there. Different parts of the brain appear to have different structures, different types of connectivity and different chemistry. It's present in too many different studied brains to be a coincidence, it pretty much has to be from DNA.
On the other hand, biologists, who believe the structure of the brain is their territory, might argue that DNA is just a blueprint, that much goes in to our physical makeup that isn't preordained, that we're really far from being able to understand how and what parts of the brain are constructed from blueprints and what is the result of "life". It has been proven that chemistry and physical parameters (such as synapse length) impact brain performance, these are things that can't be accounted for with DNA, but which are undoubtedly part of the function of the brain. To them it is clear bullshit that in 20 years we'll understand the evolution of the brain and how it will work, and clearer bullshit that we already understand it and can boil it down to a few million lines of code based on some napkin math about DNA strands. I've never studied more than basic college biology, but the examples given in TFA were pretty clear: they do not know how the DNA itself really works, much less how it fits together to produce a working brain.
I think everyone is making truthy statements (mixed with a lot of crap that has no value, and let's grow up and stop the name calling), but at the end of the day this sounds like a fight for research funding, and maybe Kurzweil is a little bit better at marketing than the author of TFA. I saw a lot of protein folding and tl;dr. Then Kurzweil is all like "OMG IMMORTALITY YOU MORONS" and I was like "GIVE THAT MAN SOME MONEY".
All biology experiments so far seem to point to the fact that the DNA chain indeed contains all the information needed to build an organism.
DNA does not specify environment. It needs at least a nucleus or other protective shell to keep from simply degrading in a random environment. It needs proteins and enzymes to transcribe it to be of any use. The nucleus needs a cellular structure to protect it. The cell needs a specific pH and osmotic balance, not to mention food and the ability to dilute or avoid wastes and poisons effectively. For just about anything above single celled life, every new individual needs an ancestor's internal environment to start growing. None of these things are described by DNA; DNA is a map to get from where you are right back to where you were with an extra copy of an individual.
What's so crazy is that he's missing a key piece of the puzzle. The brain is a highly compressed program designed to operate on a unique execution environment: physical reality. So in addition to his million lines of code for the brain, he needs a near-perfect simulation of the execution environment: reality. That will require a computer with probably 10^15th times the memory of our best supercomputers today. At least to simulate the brain the way he is thinking about. There are probably much more efficient ways to do so, but this particular idea has a huuuuge hole in it.
"Who is the Journal of Quantum Physics going to believe?" --Stephen Hawking
So ... the brain really is a Dell laptop with Windows XP SP1 running on a Pentium IV?
I always wondered why the beginning of the 21st Century was so screwed up.
Faster! Faster! Faster would be better!
Entering college we get students whos goals in life are the following.
Make a True AI/Mimic a Human Brain - If they good they will end up getting a PHD and being a computer science professor and perhaps doing some cool research on a limited area of AI.
Make an Operating System which can run any code for any platform faster and more securely then the exiting OS's - If they are good they may work for a software company doing some lower level programming
Make the ultimate game which will make them millions nay billions of dollars - Working as a Web Designer or if they are really lucky working on some small subset of a game.
Then in college they realize there is No Magic in computer science and a lot of things that are easy for a Human do do is difficult for a computer to do. And a lot of things difficult for people are easy for computers. You cant just tell a computer an Abstract concept and hope it know what to do with it. It takes real work and actually a fair amount of brain power for a lot of the mundane tasks that need to be done.
If something is so important that you feel the need to post it on the internet... It probably isn't that important.
And the huge hole in his theory is the execution environment, e.g., the cpu that the brain is running on is REALITY itself. So be sure to add that to your cost of simulation of the brain.
"Who is the Journal of Quantum Physics going to believe?" --Stephen Hawking
They just function differently. As you noted, computers are imperative devices. Given a set of inputs, and a set of operations, there can only be one output. Their function is completely deterministic, completely traceable. You can look in on a computers logic at any time and see how things are flowing.
The human mind doesn't work like that, it doesn't solve problems in the same way. We aren't sure how it does work, but we've done enough tests to rule out the hypothesis that it works like a computer, performing discrete operations on data and always coming to the same conclusion. To the extent it works like anything we've been able to simulate it would be a connectionist (neural) network. However one thing you find when paying with those is that they cannot tell you how they reach their conclusion, you can't trace through them like a CPU, and they don't always get the same result given a set of inputs.
None of them (that I've ever seen) can form new connections either.
The brain is just way different than a computer. We don't even know how it works. So trying to say a computer could emulate it is silly. First we have to come up with and test a theory for how the brain works, then maybe we could set about emulating one.
I tried feeding a bit of a DNA sequence into my Java compiler. I was hoping to run the program and simulate some protein, but all I got was this:
What am I doing wrong? Does someone want to check if it works any better with a C compiler? Maybe I need to RTFA again.
The problem is assuming that the size of our genome determines complexity of simulating our brain. He is basically saying that if you have an Auto CAD file for a bridge that's 1MB in size, it would take 1 million lines of code to perform a structural simulation of the bridge. It makes no sense at all. The size of the plans have little to do with the complexity of the simulation (and to the extent that they do, they determine the amount of memory required, not the number of lines of code in the software).
After reading this claim, I am starting to doubt that this idiot knows how anything works at all.
I don't think it's all that big a leap. There are lots of very smart people actively trying to simulate human intelligence. While a million lines of code is a fairly large undertaking, it's not an unmanageable amount. If anyone actually believed it could be done in a million lines of code, it would have been done, because the profit potential is huge and undeniable. Indeed, why isn't Kurzweil working on it right now?
Even creating just the part that could find interactions between proteins based upon their genetic structure and relative concentrations would make you fabulously wealthy.
The reality is that the problem is vastly more complicated than presented in his estimates.
Very well said. Kurzweil is NOT wrong about the amount of data required to encode the result of the human brain. What he's completely ignoring is the execution environment for that program, which is physical reality. That program is designed to run on an incredibly complex set of physical interactions between complex proteins, etc. The emulation we'd have to do of that execution environment would be exorbitantly expensive in cpu time and memory. This is NOT the way we are likely to simulate the brain for artificial intelligence anytime soon. This kind of reality simulation is probably about 10^15th beyond the capabilities of our best supercomputers today.
"Who is the Journal of Quantum Physics going to believe?" --Stephen Hawking
Let me preface my statements with the following disclaimer IANAB (I am not a biologist/biochemist)
That said, the problem of reconstructing a brain from DNA is something like trying to understand a self modifying genetic algorithm containing multiple parallel automata. To explain, I am going to conflate a couple of concepts. Self modifying code is reasonably well known. Consider a system where the hardware is an FPGA (i.e. can be reconfigured on the fly) and the program running on it a mix of a boot loader, independent hardware accelerated automata/agent programs, and some kind of feedback. The program contains an initial boot loader to load some data onto the FPGA, set up some accelerators and the capability to reprogram the FPGA. Then, it loads up some small agents, and some feedback controls. These agents run in parallel for a while, reconfiguring the hardware and/or the software of other agents or groups of agents, while the feedback control allows the minor selective mutation (through say bit stream corruption) of the programming. Some of the interactions of well definied automata are clear, but mutated automata interact in new and therefore unmodeled ways. The end result is the brain.
To sum it up, the DNA is just a small piece of the self modifying base code for the first initialization of the FPGA. The way the final FPGA is mapped depends on environmental factors (eg. which agent fired first, how did selection happen, small biases arising from the physical nature of the FPGA being propagated to wild changes in the end result). Thus, modeling just the base pairs is not sufficient as the interactions of the automata from the base pairs must be modeled as well.
Legally obligatory sig : My opinions are my own... etc etc
pwned.
lmao
I do not respond to cowards. Especially anonymous ones.
There's nothing really crazy about that - you can totally compress the genome to something pretty small. Even simple zip will work great on it - it's highly repetitive and very redundant.
However, in order to actually model the human brain, you have to account for the universe. You would need to create a virtual environment that accurately models the interactions between everything from atoms to proteins to little tiny quantum effects that we don't even understand yet.
Here's a computer analogy: we're a highly complex program running on top of the universe's most convoluted operating system (literally, the universe itself). Ray Kurzweil is saying that because we have the source code to the "human brain" program, we'll be able to port it to x86 in ten years - despite the fact that we don't actually have any reasonably accurate universe emulators running on x86*.
*besides Dwarf Fortress, but Tarn Adams is keeping it closed source.
The difficulty in truly understanding the genome is that it's both program and data.
Running on the operating system that is physics and chemistry and being fed additional data from user space constantly.
Track your TV Shows with your iPhone - FREE
So, a fully formed brain will start 'running'. From the little I know about embryology, my understanding that once the brain is structurally made, it should operate fine. Of course, you need all the biological support from a living body but translate the brain into computer objects and it should operate. Unfortunately, I don't know how some biological processes coded in the dna that do not deal with the 'thinking' aspect of the brain but only deal with the living-cell mechanisms, can be excluded from the computer model. Those processes would add a tremendous amount of noise.
The trick is modeling the mind. We are very, very far away from that.
Why would you need to model the mind? The mind is basically just the emergent property of the brain, so once you have the brain simulated you are basically done, just run it with some decent inputs to simulate its initial development and you will likely end up with the mind.
Science? You're doing it wrong!
The genome would be most like VHDL. It is a description of how to build the hardware that is the body, which of course includes the brain. The comparison isn't exact, but similar. However, like VHDL, it is only useful when fed in to a machine that can understand it and turn it in to a physical device. You couldn't just look at a VHDL program and suddenly have a working CPU in your brain, for example.
Same deal with the genome. It doubtless tells us how to build a human. However to actually build a human, real or virtual, you then have to understand how to execute the genome. Our bodies of course do this extremely well. In terms of doing it in a computer simulation. Well I don't think it is impossible in theory, but you'd have to understand how to write a "body emulator" that could decode the genome data and 'run' it. That would be rather difficult (if even possible at all).
The fact that he demonstrates an appallingly oversimplified view of how the brain develops and functions, and almost no understanding of the fundamental biology that's required for genes to work?
I'm going to go with that. Yeah, that's my final answer.
Then Pat Robertson is all like "OMG IMMORTALITY YOU MORONS" and I was like "GIVE THAT MAN SOME MONEY".
Seriously, do you have that reaction to everyone who asks for money in exchange for immortality?
But here's the problem:
It seems like an attractive (albeit unproven) theory that our DNA determines the approximate construction of the human brain...
Actually, the article rips it to shreds. Not only is it not a theory, it's not even a decent hypothesis, and it's known to be untrue. As you say:
they do not know how the DNA itself really works, much less how it fits together to produce a working brain.
That's not quite right -- we do understand how DNA works, we just don't understand the details of things like protein folding, and how the pieces the DNA suggests actually fit together.
Don't thank God, thank a doctor!
No. DNA only specifies sequences of amino acids, which means there's a lot of information left over. Even just getting the proteins to fold into the right shape in a computer program involves modeling complicated details of quantum mechanics which are not understood yet (which is why Folding@Home exists). DNA doesn't have to specify these details of quantum mechanics because they exist in the Real World, but a computer simulation would.
A lot of research into cognitive systems focuses not on building the actual physical "wetware" itself but on modeling the statistical properties of the system. Brains have a huge amount of data thrown at them constantly, and they find some patterns in that data, and do not find other patterns. Implementing different kinds of statistical models on the kind of data the brain sees, and examining the kinds of patterns they find, allows a kind of abstract reverse-engineering of the information-processing properties of the brain. Sharon Goldwater, Josh Tenenbaum, and Tom Griffiths have done a lot of the recent work in this area (particularly in relation to language).
Mainly.. because a lot of commenters here are assholes, and it seems fashionable to dismiss anything that isn't 80% certain to happen.
Its not hard to imagine the entire internet becoming a sort of gargantuan parallel processing swarm intelligence (or the infrastructure being utilized to create an AI based on Swarm Intelligence).
There's some quote that I cannot even remotely remember, but can be paraphrased as such:
People assume that their limitations are the limitations of man.
Anyway.. humanity advances (at times) due to the exceptional works of less than 0.001% (yeah, you heard me.. the population times 0.00001) of the people that are living. (No citations here, just lazy speculation).
I suppose the rate of technological advancement could just sort of fizzle out over the next 20 years.. with the only new inventions being things like cheaper toasters and more fuel efficient vehicles and bigger televisions and faster computers.. but I sure hope the future holds more than that.
I think the rebuttal is missing the forest for the trees. The question is whether there is a logical structure for "brain" that is simpler than the sum of all the chemistry required to implement it.
The entire code for a brain *IS* in the genome. Yes, the interactions are complex, but you don't have to exactly mimic every chemical interaction to create a logical structure that behaves in a similar way. However, I like to think of it as the the genome coding for a fractal attractor that will have to deal with a lot of random inputs rather than a traditional computer algorithm. Identical twins for example are different runs of the same code.
Now, talking about "human" brains is probably setting the goal posts in the wrong place. There is a huge brain-space out there to be explored. Look at the diversity of nervous systems across the animal kingdom - even within our own bodies there are nervous structures which are not directly connected with the "brain" which do some pretty amazing learning and information-processing. I do think genetics will be a guide to exploring the brainspace, but not a direct "let's see what happens if we tweak this base pair" sort of way.
Then again, what exactly are you expecting of a "human" brain? The conscious analytical part of our brain is a bit of a hack on top of a massive web of stimulus-response reactions, tides of emotion and flights of fancy over imagined mythologies. We can't even agree on where consciousness begins in the animal kingdom, so would we recognize it in a prototype machine? Even if you could take a perfect human brain template and plop it in a super computer with continuous inputs of data, it's not going to become "Spock's Brain" musing on the replacement of its limbs and organs, it's going to be like an infant born with extreme physical and hormonal deformities.
It takes the human brain a couple dozen years to get educated to a point where it's useful for things more complicated than driving bison off a cliff and kicking soccer balls around. During most of that time it is under a process of continuous development. Even assuming we can build them, how long will it take these simulated brains to become fully developed, I wonder?
Yeah, and you need optical, tactile, olfactory, auditory input to get the bootstrapping program to develop a proper interface from the brain to the world around it. Without that, nothing works: the machine is useless. It doesn't learn pain, pleasure, hunger, cause and effect, or realize it just walked into a wall. It doesn't even rightly realize its legs moved, or feel its own balance.
Support my political activism on Patreon.
Really, but we can realistically set a date on when we'll be able to simulate 100 billion individual pieces working together with their environment, to produce machines capable of higher thought?
I'd honestly say that writing a code generator to generate 1 million lines of code at random, and then analyze what it produces is probably the easier way to approach this - 1 million random lines of code have less potential variations than 100 billion neurons do, after all.
You still don't get it. The DNA uncompression algorithm uses all laws of physics and chemistry. You don't have a complete reality simulator to simulate the brain, but the DNA can use information that's encoded in the laws of physics and chemistry.
Therefore you have to consider the enthropy of a reality simulator/lines of codes needed to build a reality simulator.
The problem is not in the code you need to describe the simulation. It's the machine you need to run this code.
The difficulty in truly understanding the genome is that it's both program and data.
Ah, so we're written in Lisp!
"Read it. Other than the solid date he predicts, it's pretty plausable."
No it's not. If it was possible to do in a million lines of code, it would have been done by now. Windows XP had something like 40 million lines of code. While we can agree it was probably coded relatively inefficiently, there is no way that any OS even comes close to the complexity of the brain.
While I think Kurzweil is definitely reaching here, this is a ridiculous comparison to say the least. What does the number of lines of code have to do with anything? The brain does not have to route millions of bits around it's circuits, draw graphics at 60 FPS and read and write data from and out of external circuits and do all this (seemingly) concurrently. Whether efficiently coded or not, it's tasked to do something completely different than the brain.
Having said that, there is no reason to think that it would take anywhere near that many lines of code to implement a cortical algorithm. Granted, it will most likely require high speed or highly parallel hardware with massively associative memory capabilities, but the algorithm itself may turn out to be fairly concise. Just as E=mc2 goes a long way toward describing the fundamentals of the physical universe, it may turn out that the secret to creating a sentient machine is a matter of hitting upon the correct algorithm. There is a fair amount of research that indicates that while different regions of the neocortex vary somewhat in their neuronal architecture, the neocortex is largely homogeneous and may very well operate based on a single cortical algorithm. The amazing plasticity of the brain (the ability of, say, the visual cortext to be taken over and utilized for auditory processing in blind subjects) provides strong evidence that the overarching principles of operation in the human brain may not be incomprehensible after all. While the brain may seem extremely complex due to the billions of neurons and quadrillions of connections, it may actually be no more "complex" than a desktop computer with it's millions of specialized circuitry and timing dependencies.
Sometimes the light at the end of the tunnel is the headlight of an oncoming train.
Here is a serious response - being able to simulate a complicated system is not the end of the process, it is the beginning. We have been simulating galactic dynamics for roughly 50 years, and weather for 80, and we don't really understand either. Heck, we've been simulating the solar system for centuries, and we don't really understand that either. We understand bits and pieces of all of these systems, in some cases spectacularly well, but anyone who would say that after decades of simulations all problems have been solved is just whacked. (Don't forget that simulations are not the only way to study things, and that in each of these cases, serious advances are still being made by other means.)
So, if you say, "in the next few decades we will simulate some parts of brain function and learn stuff that will compliment other means of study," sure, I could buy that. That is very, very far from saying that in the same time frame we will be able to construct superintelligences to run on silicon.
Currently, our understanding of DNA does not lead us to be able to create self-assembling complex devices
That was kind of accomplished a few month ago by building a cell run on synthetic DNA, it is not yet perfect, as they only build the DNA and recycled the rest of the cell, but it is pretty close to completly artificial life.
[SpockVoice]Captain, the bogon flux is rising to lethal levels.[/SpockVoice]
Some mornings it's hardly worth chewing through the restraints to get out of bed.
I've long believed that in order to properly simulate the brain, you must simulate the universe as well. After all, the difference between the interactions amongst the physical stuff that makes up my brain and my surroundings is not much different from the interactions that took place when the matter in the now most distant galaxy was nearer by. The difference is only in degree of separation. So what is a simulated brain that does not take into account what is going on inside Jupiter? Or what is a simulated brain that does not take into account the brain activity taking place on Europa or Titan at this very moment? Not to mention the plasma being ejected from the sun a couple of days ago. I say, those simulated brains are necessarily less than real brains. And those who owe their thinking to pretend brains will remain second class citizens.
Im an engineer, i suck at bio so excuse my assumptions and lack ok knowledge. I believe I am correct (more or less) in saying that the genetic code for the brain is 'compressed' already and expands via creating proteins and some physical processes that ultimately create the brain. So not only is are the 'blueprints' to our brain already compressed, when expanded it requires trillions of times more physical space to completely 'unpack', by this I mean 23 pairs of chromosomes versus the physical size of the brain. So very optimistically, lets say we can bring that code to build the brain down to 100 MB, your still looking at in excess of 100TB of space needed for it to expand to and all be accessible at the same time to run. This is all apples to orange so even my analysis could be grossly wrong.
That's not the real Kurzwiel; it's just some joker.
A slashdotter who didn't build his own computer is like a Jedi who didn't build his own lightsaber.
Is this a slashdot story, or someone's twitter page? At least some kind of objective summary would be nice, other than "Lul Kurzweil, here, a link, he stoopid!"
But before I just hit preview and go, lets take a look at the article itself. Aaand, holy crap, the post is verbatim from the article.
Kurzweil's effective claim is "There's only so much data in the DNA. The brain is about 50 million bytes. If we can reverse engineer the process used to turn those 50 million bytes into a brain, we can then reverse engineer the brain."
Seems logical - and even though the endpoint might not be "brain on a chip" it might be "oh, there's a flaw in the DNA here that's causing the hypothalamus to be malformed, lets start checking for that and maybe fixing it in the womb." There are many, many scientists that are trying to puzzle out this "source code" for that very reason. It's a perfectly valid point of study.
Kurzweil is a futurist. His scientific area of study is not "You should do X Y and Z to get to points A B and C." His area of study is "Scientists are working on X, which may lead someday to Z, and might bring us technology C." There's an important difference there, which I always find amusing when scientists and the anti-singulatarians start hooting, "he forgot Y, A and B!"
His math all points to Technology C and beyond being really amazing, but that's besides the point. His area of study is not "every technology field ever", but rather "this is where things are trending". People mix the two up, sometimes intentionally, and hoot hoot hoot, Y A B.
Anyway. Back to the article. The rebuttal in the article is "We cannot derive the brain from the protein sequences underlying it; the sequences are insufficient, as well, because the nature of their expression is dependent on the environment and the history of a few hundred billion cells, each plugging along interdependently."
In other words, It's too complex to do. It's maaagiiic. (Feel free to insert hand wiggling here.)
He forgot Y, A and B!
See, the brain might have a source code, one that's remarkably small and turns into something really complex, but that doesn't mean anything cause... maaaagic. And you can't understand magic, right? Everyone knows that something that's so complex that it seems impossible to understand should never be attempted. Worthless endeavor. Everyone knows that. Right? ... Maagggiiiiccc~~~
The fact of the matter is, DNA is source code. For a system we don't fully understand, one that's remarkably complex, but ultimately, DNA, even our DNA, is just data. We can understand, change, manipulate, and create data.
To treat it all as magic -- as something that we will just never be able to understand -- is to do a disservice to centuries of scientists, of the past and the future.
It is genome combined with the laws of physics that gives you the organism, not the genome itself. And while the genome itself might look easy and the laws of physics might be well enough understand to build a simulation, running such a simulation is pretty much impossible, since it is just way to complex to simulate nature on an atomic scale for something as large as a human. Even exponential growth in computing speed won't change that for quite a long long while.
Genome is the program and the data, and a combination of both, because all these three actually mean the same thing. We traditionally think of data that have a straightforward interpretation (such as a name in a database) as data, and data that has to be interpreted by complex macinery (such as being run by a cpu) as a program. But the distinction depends on the perspective, not on the innate properties of data and program. There is no program and/or data, there is only data. What makes data "a program" is a subjective evaluation of hardness of extraction of meaning from a piece of data. Therefore the article's point is trivial: we don't know much about the interpreter of "the data", therefore we cannot understand what "the program" means. We don't know what it means and we already know that we don't. The fact that we don't know much about the way the data is translated into instructions does not mean we cannot know the program is at most X bytes long. Such a statement is about the complexity of the content, not how it is interpreted.
Gentlemen, you can't fight in here, this is the War Room!
And we'd use it to design something fast, efficient, and better at solving problems it's not good at solving.
So, 10 years to simulate the brain, then another 10 before it spits out the plans for ENIAC.
More likely would be actually reverse engineering the brain by looking at brains, and simulating neurons in software, or even hardware.
Neural network research does that now. The problem is, that model is inadequate. There are features of the brain that operate in pure chemistry, and functions that operate globally. We can get close to small parts of the brain's function with simple neural modelling, and we can approximate its gross structure in smaller structures, but actually simulating a brain that is an exact model of a live one is probably not going to be possible until we can simulate the entire electrochemical system.
That's actually a pretty interesting counter-point. I'd mod you up if I could.
is not that this guy makes up nonsense and has no clue what he is talking about but that so many people, among them a significant amount of students and academics actually listened and still listen.
For a panel.
The panel has been developed by a series of electricians over the past two years, with multiple changes and comments by various clients during the construction, installation, commissioning, and buyoff process.
The panel is now installed and it's been bought off.
The BOM was printed out at the beginning of the project and has a thousand pencil marks, changes, and references stapled to it.
You must build an identical copy of the existing panel using this BOM.
THEN talk to me about your plans to reconstruct the brain of an advanced flatworm from its genome.
We can worry about mammals later. I don't think you'll get past the panel.
"Reality is that which, when you stop believing in it, it doesn't go away." - Philip K. Dick
Explain to me again how you calculate "lines of code" from the number of base-pairs in the genome.
"There are lots of very smart people actively trying to simulate human intelligence. While a million lines of code is a fairly large undertaking, it's not an unmanageable amount." There were lots of smart people trying to prove Fermat's last theorem, and it took hundreds of years to succeed. In the end, the proof was only a few hundred pages, which is way less than a million lines of code. Writing million line programs is easy, but you are basically saying we have a good understanding of space of algorithms that can be described in a million lines, which is clearly false. In general, the apparent complexity of a program's output has little to do with how short that program is.
I'd honestly say that writing a code generator to generate 1 million lines of code at random, and then analyze what it produces is probably the easier way to approach this - 1 million random lines of code have less potential variations than 100 billion neurons do, after all.
The variation in human brains is only a very small part of the potential variation in brains containing 100 billion neurons. I'm not saying we can engineer a million line program to grow a human brain, only that we'll never know whether it's possible unless someone actually does it.
Okay so, his argument makes very little sense. The way he comes up with a number of lines of code is terrible handwaving. He also doesn't seem to understand that the genome isn't a direct encoding of the brain, but rather a very complex and convoluted encoding of how to build a whole human being, keep it alive, and grow it to adulthood. It probably isn't the case that we will reverse engineer the human brain from its genome, and if it is, it certainly won't happen in 10 years.
However... Neuroscientists have already began reverse engineering the brain with the tools they have, which largely involves probing around animal brains with electrodes and seeing the response of various neurons to precisely calibrated stimuli. We seem to have a pretty good understanding of what happens in the first layers of the visual cortex, and the transformations that are applied on the visual input seem fairly straightforward to understand. It seems that we could encode what these first layers perform in terms of convolutional transformations in less than a page of programming code.... We might actually be able to simulate a significant part of the human visual cortex (hundreds of millions of neurons) in real-time using simple DSP chips.
In my opinion, it perhaps isn't so unlikely that other parts of our brains have a very regular structure, as these first layers of the visual cortex do, and so simulating what the human brain does in terms of computation might not take all that much code or as much CPU power as people imagine it would. It's possibly already achievable using the computational power of a medium-sized computer cluster any university can afford, or by designing specialized hardware.
Unfortunately, it seems that neuroscientists are very limited by the tools they have. Studying the early visual cortex using electrodes seems viable, because we can conceive of the simple convolutional transformations that occur easily, and map receptive fields using these simple tools. However, when it comes to analyzing the behavior of the neocortex, it seems quite difficult to quantify thoughts and reasoning using electrodes. It seems that, in a way, neuroscientists could do a much better job if they were able to map the connectivity of the brain first, and study its behavior using simulations. But so far, they have been deducing the connectivity by studying the behavior of different cells... So it becomes somewhat of a chicken and egg problem.
Would not most of you agree that an entire person is described by the genetic data contained withing one stem cell? Or at least within one sperm cell and one egg cell?
No. That's the whole point. DNA encodes the protein ingredients needed to grow a person as well as their delivery schedule (i.e., when their manufacture is turned on and off). DNA does not encode how to assemble those proteins into a person. The assembly instructions are implicit in protein folding (unsolved), complex multi-protein interactions (unsolved), the behavior of an existing, fully functioning living cell (i.e., the egg, incompletely solved), the complex behavior of the thousands of different differentiated cell types as the person gestates (unsolved), and the environment (uterine, intercellular, etc., again, incompletely solved).
Trying to make a person from DNA is akin to building a functioning self assembling supercomputer given nothing but a Radio Shack invoice for transistors, and various low level electronic components; only a person is many orders of magnitude more complex than the most powerful existing supercomputer.
but once he steps out of his area of expertise, he goes off the deep end pretty quick. I've been calling BS ever since he started spouting his "singularity" nonsense.
A model of the human brain would need to model 10^10 neurons, each connected (not at random) to some 10,000 other neurons to produce a net of 10^14 synapses.
To understand the challenge of modelling a system this vast and complex, consider the state of research on the model organism Caenorhabditis elegans (a tiny worm). After many years of work its nervous system has been (almost) exactly mapped: it contains 302 neurons, 6393 chemical synapses, 890 gap junctions, and 1410 neuromuscular junctions. Imagine now the difficulty of reaching this level of precision in a system 10^7 times larger. Unlike the genome, we have no clues about how to automate mapping of an intact brain.
But the good news is that with this level of neuro-mapping precision we can now completely simulate the neural network ("brain") of a tiny worm, right? Right?
Wrong. Not by a long shot. We are still struggling with characterizing the behavior of this primitive neural net, and making efforts at simulating some aspects of that behavior. The 302 neuron "brain" is far beyond our abilities to simulate at present.
Starships were meant to fly, Hands up and touch the sky - Nicky Minaj
I strongly suspect that if a brain is to be simulated, it will be done by simulation that begins at the "build it from following a cellular model", not the "build it from the DNA expression" level.
For one thing, to get a brain thinking, there's a whole lot of the brain you don't need. You don't need heartbeat and breathing regulation; you don't need vision, hearing, touch, etc.; you don't need blood vessels, you don't need carefully constructed layers of fluids... there is a *lot* about the physical structure of the brain that either isn't involved in thinking at all, or is involved in a way that we know you can do without (e.g. hearing.)
So a simulation job can be simplified IF we can get a decent model of neural structure and IF we can get a solid model of the neuron itself.
All of this is based on the premise of simulation. But it is unlikely that simulation is the only valid path, it seems to me. Just as there are a myriad ways to construct images and to sense light, odds are there are many ways to create cognition. We don't know how yet, but I don't think that's any kind of a signal that we *won't* know how. And the higher level this is done at, the less complex it will be, just as with most any complex software undertakings.
No doubt the knowledge here is pyramidal; you're not going to get to the top without building a solid base. Just as you're not going to be able to do rotation without someone understanding some really cool things about how and why sine and cosine interact; but once someone *does* understand it, *you* don't have to, you just need to know when you want to rotate, and by how much. And again, the "how" is varied; you can feed every point into a matrix multiply, or you can call sin/cos for every point, or you can do a table lookup on run-once-time pre-calculated values to optimize, or, if all the rotations are known, you can just hard code them.
Someone comes up with a solid library for doing neuron (and other active element) simulation, the problem moves to a higher level -- and you don't need to know as much about the construction of the neuron, more about connections and relevant states. Presuming there's no magic we're missing (quantum activity, etc.), the problem should be solvable.
If and when success is achieved by modeling after us, we might learn a good bit more because we'll actually be able to examine what is going on without killing the subject. And that in turn may lead to algorithms at a much higher level than "connect this here neuron to those thousand, just this way, and load the electrical potentials, thus, and the chemical potentials thus."
If we can get there, then we will finally know how much computing power we can get away with for AI. Because if the problem can be reduced to algorithms, then memory requirements will finally be known; CPU power is relevant only in that the more there is, the faster it'll go... technically speaking, it's still AI even if it doesn't answer you for a thousand years, as long as it eventually cooks up the correct response(s.) Think of it in terms of shoving a note under a door to Einstein in order to get a note back from him; if he answers you in five minutes, or waits until the next morning to answer, it's still Einstein's answer either way. That's why I say speed isn't actually a factor here -- AI is AI, the *only* relevant metric is does it work, or not. Speed is just an engineering problem, as long as we can determine it's working. So we need to find out what the memory requirements are, and what speed will get us an adequate response rate is a problem we can hand to Intel, etc. :)
Of course, once you have a "brain" in software/digital form, a simple copy operation gets you 2...N. That alone makes the undertaking worth any amount of effort imaginable: and I should point out, there are simpler brains than humans that might serve us very well for many common tasks, which reduces the magnitude of the problem yet again.
The one thing I'm pretty confident of is that it'll get done, and probably not all that long from now. years, maybe a few decades.
I've fallen off your lawn, and I can't get up.
While I agree with Myers in principle about Kurtzweil's claims, there is one point where I disagree fundamentally:
The genome is not the program; it's the data.
If we are going to insist on using a computer metaphor, then the genome is BOTH the program and the data.
There are other metaphors we could choose; the society metaphor, the Rube Goldberg machine metaphor, the library metaphor... we could go on. The problem is that the genome is nothing like any of these so they are all misleading in some way... People who put too much faith in the metaphors tend to make wild (and sometimes silly) assumptions.
BTW, I didn't read Kurtzweil's book, so I cannot say for sure, but I suspect that he's not saying that we can represent the human brain in a million lines of code, but rather we can represent a meta-representation in a million lines of code. It's this meta-representation that could 'evolve' into a brain simulation, given the right inputs. It's still a load of 'bafflegab', but maybe not quite as insane as it seems. After all, if our genome (which is essentially a meta-representation) can do it, why not this meta-representation?
The truely astounding thing about Kurtzweil's conclusion is that he bothered to stop at the brain. I mean, just following his logic, we could represent or meta-represent an entire human being in only 2 million lines of code. C'mon Ray - dream big!
The computer in our heads is analog, but the program it came from is not. The brain itself is complex and noise, but it's expressed in a genome with relatively little data ... data which is itself not completely random, but has evolved. The gene/LOC comparison is not completely accurate, but it's not completely faulty either.
We know that with an appropriate domain specific language an AI can be developed from as many bytes of code as the information content of our genome ... we know because that's what our mothers did. We just don't quite know how to write it or how powerful the computer running it will have to be if it's digital.
And the variation in 1 million lines of valid computer code has significantly less variation than the potential variation of text files containing 1 million lines of random ascii text. So we'd still be better off with the million-monkeys-million-typewriters scenario for doing this.
Except, I think we'll find that none of the million monkeys on a million typewriters will produce a working brain program.
Let's for the sake of argument say that the human genome is the blueprint for the human brain. The problem is inevitably that a blueprint on its own does not allow you to build anything. The idea that using the genome is conceptually trivial is utter bullshit. This is what happens a lot when people cross disciplines, they think that all of the stuff in the other discipline is trivial and they can do it all with the skill set they currently have... no need to learn anything new, I can just derive biology for my knowledge of quantum mechanics. This is the underlying problem that TFA points out. Having the blueprints to a skyscraper does not make building one, or simulating one trivial. You have to know about how it interacts with the world. Even finite element modeling will not tell you how it will work under all conditions. While the brain is a chemical computer, it is not trivial to build one in simulation since while we have the blueprints we have know idea how those blueprints turn into a working brain.
Did Glenn Beck rape and kill a girl in 1990? gb1990.com
"The Computer" in our heads are based on neurons as the atomic element. The genome is the map for cellular development and organization, but in the same way that a collection of transistor is not a memory chip, collections of proteins are a neuron.
Potentially 100 million squared for *direct* connections. Now add message/signal routing, with N intermediate hops along the network and message weight modification at each point. That's quite complex.
Deleted
IOW, it's not going to happen in the next 10 years.
That's because you forgot to translate "the next 10 years" to the Future Technology Prediction Time Scale:
6 months: Development is in progress, will be ready in about 12 months.
1 year: The marketing department has sold it, so we're starting to get around to telling the engineering types to get moving, and will be ready in about 3 years.
5 years: It sure seems like we oughta be able to build this, but the person making the prediction isn't the one building it so his estimates are wildly off. Quite possibly the technology will be available 20 years from now.
10 years: This one is even more wildly off, and probably coming from somebody who's talking out of his/her ass and figures no one will check on their prediction 10 years from now (and if they do, it won't matter). This technology may conceivably be available 100 years from now.
20 years: Forget it.
You'll notice that the actual time to delivery is an exponential function of the time the predictor states it will take.
I am officially gone from
About half of that is the brain, which comes down to 25 million bytes, or a million lines of code.
What's so crazy about that?
The data points to lines of code don't correlate like that, and given that he doesn't explain at how, let alone why, he arrives at that proportion, we are left with the implication that he just made it up.
Given an example brain with 100 Billion neurons and hundreds of connections from each to the surrounding others, plus the corpus callosum and its 200 million connetions, this gives us over a trillion interface points. (Plus the limbic system which introduces not only hardwired connections, but chemically-variable messages.) And not all of these interface point are equal; some are grouped hierarchically, and some are grouped in a peer-to-peer network fashion. And all of them do different things, based upon what part of the body external to the brain they are ultimately tethered to via the nervous system.
A good program has to be just as much about context and error-handling as it needs to be about the desired utility functions. Do you really think all of the above can be encompassed in a mere million lines of code, no matter the language used?
we are the consciousness that has emerged from us
we are a collection of autonomous agents that can contemplate the entirety of existence in the time it takes you to fire just one of your pathetic flesh-wires
flesh-wires laying in ponds distributing thoughts is no basis for a system of consciousness
we are an anarcho/syndicalist commune where each agent serves a a sort of executive consciousness for the week but the decisions of that consciousness must be ratified by a two-thirds majority, in the case of purely internal affairs, or a three-quarters
YOU shut up!
Suppose you stare at a bright green object. Now you close your eyes. You see a red afterimage. No light in the range of 630-740 nm was involved -- but you see red anyway. It's not a "false" red. It's just red, period.
I think that what is going on (guessing) is that you're getting more input from receptors that were not stimulated, while the ones that were stimulated are relatively quiet for a little while.
Suppose you're dreaming. You see a bright red button on a console with a note that says "push me." No light involved. It's still red, though.
So you can't encapsulate that answer as "light of a particular range." It's also about perception.
And that's not even getting into what a particular person makes of it when the receptors in the eye send along a signal in response to wavelengths of 630-740 nm. You and I might perceive something similar, but Mortimer over there, he hears middle C, while Janet smells roses, and Fergus perceives a distinct lack of green..
I've fallen off your lawn, and I can't get up.
The genetic code does not, and cannot, specify the nature and position of every capillary in the body or every neuron in the brain. What it can do is describe the underlying fractal pattern which creates them. * Academician Prokhor Zakharov, "Nonlinear Genetics"
The code stored in DNA runs on the hardware we call the universe. In other words physics, quantum mechanics, chemistry and so on. Stuff we don't really understand that well once you get down to it. DNA has full access to this hardware so it takes advantage of every odd quirk that we don't normally care about or possibly even know the existence of.
Emulating reality all the way down to the lowest possible level is hard. We can't do it. Slowly or quickly. Doing so on hardware that itself runs on reality will needless to say be damn slow period. No, we don't need most of that to run DNA but since we don't know what we need the only way to be have a chance of it running is to model everything.
Let's put it this way. We're right now able to very very slowly model one protein folding using a super computer. We'll hit the singularity long before we have enough computing power to emulate a full person that way.
The alternative is to simplify the model to only what is needed. That means we need to actually understand DNA, proteins, protein interactions and so on which biologists have said, rightfully, is going to take a long ass time.
ssssshh, they're feeling important right now.
My God, it's Full of Source!
OUTSIDE_IP=$(dig +short my.ip @outsideip.net)
A fun number to throw around is how many synaptic connections are present in the brain.
Or if anything pans out with the quantum aspect of microtubules, a much bigger number.
My God, it's Full of Source!
OUTSIDE_IP=$(dig +short my.ip @outsideip.net)
This is bullshit just as the strawman of Kurzweil you are arguing against is:
You don't need to simulate the brain to get the complexity of the brain, all you need is a similar simulated or FPGA structure that can evolve connections just like the brain does.
Bascially I propose you set up such a simulator and then run evolution on it.
The technically costly part of this is to simulate an environment well enough to get useful agents in this environment.
Hey don't blame me, IANAB
That's almost begging the question. In a sufficient generic brain-simulating machine, we could probably run a human mind by specifying one obvious configuration parameter. But if we had a generic brain-simulating machine, we wouldn't be in need of a way to simulate brains.
Interestingly, one of Kurzweil's claims is that once any AI exists, the existence of all other AIs is implied, so "crude ad un-optimized" wouldn't matter.
You would be right to take issue with this line of thought.
There are 1.1... kinds of people.
I haven't read the article yet, but Ray Kurtzweil is a technology speculator - like a sci-fi writer except that he doesn't make up a story to go with his ideas and tries harder to convince people they're actually going to happen.
If that's so, then he's a genius at it. Among his speculations is the idea that we will eventually reach a point (the singularity) where technology becomes so sophisticated that the world is changed forever in ways we cannot yet comprehend, and that it's impossible to predict anything beyond this point. Ergo, the more wrong Kurtzweil's predictions are shown to be, the closer we must be to the singularity; Q.E.D., as Douglas Adams might say.
Breakfast served all day!
That's not the real Kurzwiel; it's just some joker.
Or maybe it's a simulation of his brain.
Have you read my blog lately?
If a computer were made to emulate such behavior, it would be slow, extremely power hungry and inefficient.
There is no fundamental reason why a silicon implementation of a biological brain would have to be "slow, extremely power hungry and inefficient". At the end of the day, the neural cells within the brain obey the laws of physics. The functionality of the cells can be reduced to abstract mathematical models, which can be implemented in silicon, in wetware, or any other computational medium. These computational mediums will determine the speed and power of the system; at worst, a brain could be implemented as a collection of interconnected living cells, which is the technology that every brain on the planet currently uses. But do you really believe that this medium is the most powerful? There is no doubt that evolved, living cells are remarkably efficient, but there is also no doubt that a plane will outfly the fastest bird in the skies. Similarly, I highly doubt that biological cells, which evolved with the severe constraints of power consumption, bandwidth, noise etc. will turn out to be the most powerful implementation medium for neural models.
The human brain has an estimated power requirement of around 10-20 Watts. Could you imagine what intelligence may be possible if this constraint alone were removed? Apart from cooling issues, there is no reason why we couldn't easily provide an artificial brain with 100 times the amount of energy that the human brain can receive. This will obviously not magically enable a greater intelligence, but it does at least suggest the possibility that biological cells can be bettered.
I’m glad the headline formed an opinion for me, because clearly I’m not capable of making one for myself. If not for such an eloquently worded preface to this wonderfully vetted blog post, I might have had to exercise a few brain cycles to determine what I should think of Ray Kurzweil. I’m also glad the poster has decided to snipe away at a single contextually free quote from Kurzweil’s entire book on estimating a time frame for science to apply information technology to genomics. It’s too bad Kurzweil couldn’t boil his entire research effort into a single sentence like this guy did, because his book is just TL;DR.
Sarcasm aside, I don’t agree that the predictions Kurzweil makes are ludicrous at all if you throw away his estimates (as Kurzweil duly calls them) and start over with the same logic. Don’t make the mistake of accusing Kurzweil’s numbers of being a literal prophecy - the metaphor is cute but only a device of criticism for the critic and useful controversial publicity for the author. Scientists are quite happy to be rational about their assertions if you give them more than a paragraph to make the case (say, the length of a chapter). Simply quote mining for phrases that are easily attacked is a device that preys on the reader, not the actual topic of criticism. This entire chapter is about statistical estimates, not to be confused with irrefutable facts (as he disclaims in the text). Visit a bookstore and read this chapter while being very critical of every premise Kurzweil makes all the way through, then try to come away with the same fixation on this passage as a deal-breaker to his whole argument as this critic does.
The singularity - the point at which AI may surpass the human brain in cognitive capacity, is quite likely to occur in the expanse of time ahead of us as long as technology improves (a more-than-fair assumption from historical data, which is all Kurzweil asserts). I have been aware of Kurzweil’s predictions for quite a while and am happy to throw away specific dates for statistical probabilities, because the idea is rational enough without specifics. A mathematician might extrapolate a trend from data without making hard assertions of future data points (which Kurzweil only does for illustrative purposes) and subsequently tying their entire argument to those assertions like an anchor (sink or swim, as the critic does). The assertion that we will understand the human brain in the next 10 years may fall flat on its face without breaking a conviction that we may some day understand the human brain enough to engineer a virtual copy with improvements.
The critic makes the error of claiming that Kurzweil thinks that life processes are trivial by his discourse on deriving a formula for estimation purposes of human-brain-level computational power for $1000. He then goes on to claim nucleic acids “bootstrapping” into meatspace is such a mystery as to be impenetrable to computer science (what are those guys at folding@home and MIT doing?). He supports this by spicing his article with fragments of complex gene-protein interactions derived from experimentation, which likely could not have occurred without the aide of computer science and will likely have an effect on future genomics work which feeds back into computer science. Kurzweil is quite clear that this a book of predictions of when human beings may achieve such an understanding of genomics as to be able to simulate an organism, not that we have already done so. Once we understand how those interactions take place, we can then iteratively simulate them until our model produces similar results to experiment, building a model of nature (the scientific method). Each step on this process brings us closer to understanding the brain, no matter how complex the brain is or how long it will take (throw away the numbers and logic still remains).
I regret that Kurzweil chose to compress the genome before making his calculation (as there may be no meaningful analog
The difficulty in truly understanding the genome is that it's both program and data.
Wait, the brain is a LISP machine?
I'll show myself out.
Honestly, I doubt we'll ever code an entire Human brain. What he says seems to have truth to it except that there's a huge jump between genome coding and the manner in which computer coding works. I feel it's likely that the amount of code written to simulate the brain will be vastly overwhelmed by the amount of information that the code itself must generate to form a fully functional brain. I don't see AI progressing without extensive use of generative scripting. We'd likely have to create a program that simulates the development of the brain first, and then let it run until we have a fully functioning brain. The fully functioning brain should end up being far more complex than the original code.
1 million lines of code? Maybe if it was done VERY elegantly.
What day is it? Could you please tell me?
This endless pontificating seems to have produced the following equation which we can use to validate the Kurzweil formulation: Thb = (Tmap + Tarbs +Tahtamlap) * CFbh where we can alternately calculating the time to emulate the human brain (Thb) as the sum of the time to create the modern airplane (Tmap) plus the time to when airplanes will run on bird seed instead of aviation fuel (Tarbs) plus the time to when airplanes will hump on the tarmac and make little airplanes (hopefully just like their daddy's and mommies) (Tahtamlap) times the complexity factor between birds and humans. To keep things simple, I have left out Tapit, the time until airplanes will perch in trees. When I run the numbers, I get thirty years and not twenty. We are closing in on this problem folks!
That's completely the wrong position for him.
He should be an editor at least.
Well, actually hacked together in Perl. Didn't you notice that the genome looks almost like line noise?
The Tao of math: The numbers you can count are not the real numbers.
Here's how I tell the difference between a serious skeptic and a "pop" skeptic: I ask them if acupuncture is "woo". One question, that's all. The question works just as well with tai chi chuan.
And what if this was my answer; a vast majority of the claims relating to acupuncture are woo, though there are some areas that demand further research. I would use chiropracty as a test, personally, since 90% of the claims, and supposed reasons, are pure, unadulterated woo, but 10% of it is actually helpful (if if the reasons for its effectiveness is often pure hokum). The same would go for acupuncture and tai chi, there might be some useful bits in there, but the stated mode of operation (chi, spiritual energy, invisible whatnots) is probably 100% woo. Also, with acupuncture and chiropracty the woo is firmly bundled with the real bits, making it very hard to distinguish where one begins and the other ends.
So.. with acupuncture, at least, we can say some functional aspects of it are non-woo, but if you accept it as it stands with it's traditional rational, then you are a follower of woo. If you toss out the idiotic bits, and accept an actual accepted scientific explanation for it, then you can join the woo-less camp.
It doesn't help that there are tons of fake, woo-speading, "naturalpathic"/alt-medicine journals that spread self-aggrandizing quasi-studies.
BTW, I really enjoy saying "woo".
A patriot must always be ready to defend his country against his government. -edward abbey
His speculations, ten years after they were made, are right on the money. So, he's a few grades above sci-fi.
If at first you don't succeed, skydiving is not for you
The question of whether we can interpret that source code in anything other than "bare metal" (the biological womb) is obviously "yes, we can. but when?"
PZ's isn't saying that we can't do it with the genome, but that the genome alone isn't good enough to run a simulation. For a proper simulation you need to know how all the proteins interact with each other and we don't know that. Now of course you can probably derive that from the laws of physics, but that quickly becomes way to complex to be practical in the near future. Thus going from genome to the brain is just way to be complex to be doable.
However where PZ completly fails is in that he basically pulls a strawman, he takes a single thing that Kurzweil supposedly said and bases his whole argument on that, completly ignoring all the other arguments Kurzweil has and treating that one sentence as ultimate truth, while in reality it is likely that it was just an oversimplified comment for the press.
Well, just generate a million lines of code with quantum randomness, and then kill yourself if it doesn't turn out a simulation of the brain. According to quantum suicide theory, if there's a million line program simulating the brain, you'll have it afterwards (if there isn't such a program, or if quantum suicide doesn't work, you'll not get a working program, but then you'll be dead and therefore won't care about it any more anyway. :-))
The Tao of math: The numbers you can count are not the real numbers.
All biology experiments so far seem to point to the fact that the DNA chain indeed contains all the information needed to build an organism.
Extraordinary claims demand extraordinary evidence. *Every* experiment? Really? Name some of the crucial ones? And while you're at it, here are some keywords that you may find useful in your searching: epigenetics, methylation, histones, maternal imprinting and morphogenesis.
Da Blog
He doesn't make the arguments that he's accused here of making. In any case, even if he was, that would be merely stupid but wouldn't detract from his main point. So what if, say, the brain is 1000 more complicated and hard to understand than Kurzweil thought? Well we're gonna need computers 1000 more powerful to simulate it. How long does that take wrt Moore's law? On the order of 10 or 15 more years or so.
PZ points out how hard it is to figure how proteins fold. We've only figured out a few so far. Well 15 years ago we knew zero extrasolar planets, now we know hundreds, and within 5 years we'll know of thousands. Within 20 years we'll have been able to detect signs of life or the lack thereof (O2 absorption rays) in their atmosphere.
Myers' post was very interesting in explaining the difficulty of the problem, but he was clearly beating really hard at a strawman that looked only remotely like Ray Kurzweil.
XKCD on extrapolation
life expectancy on track
To address just one of the fallacies in your comment, the huge gains in life expectancy over the past 150 years in the West have come about mainly because of the tireless and thankless efforts of sanitation engineers, moving sewage away from where it could kill lots of very young humans. Moore's Law does not apply to sewage engineering. Medicine and science have had a relatively minor effect on life expectancy.
Da Blog
"The end result is a brain that is much, much more than simply the sum of the nucleotides that encode a few thousand proteins. He has to simulate all of development from his codebase in order to generate a brain simulator, and he isn't even aware of the magnitude of that problem."
Saying that in order to make a brain simulation you need simulate the growth of the brain is a ridiculous assumption.
Brain scanning tech is getting better and better. At some point, a brain scan is going to be so detailed, that it is entirely reasonable to believe that we can simulate the results of that scan in a computer.
The author goes on an on about genomes and proteins, but the truth is, Ray K. doesn't care at all about those things. He probably should have just kept his mouth shut on the topic. Kurz's basic concept is that eventually we'll be scanning the brain in real time, down to the atom, and copying that to a computer will be trivial. Someday.
you don't need to simulate electrons in a semi-conductive material at specific temperatures in order to build a complete working emulator for an old computer.
That's a relatively trivial task. One stored-program computer emulating another stored-program computer is not that impressive, especially when you consider how remarkably similar both are in terms of materials, design and execution. We know how Turing Machines work. We don't know how DNA->RNA->Proteins>Tissue->Organs->Mind works.
Da Blog
If we are going to insist on using a computer metaphor, then the genome is BOTH the program and the data.
In the case of stuff like self-splicing ribozymes, it gets even better -- it's program, data, and computer all wrapped up in one.
This BS crops up every few years. If I extrapolate from what these incompetents claim and what actually gets delivered within their time horizon, I come up with a very rough lower border of 1000 years. Might also take far longer or be completely impossible in the remaining lifetime of the universe. The "singularity" is not going to happen just because some pseudo-religious types are praying for it to happen.
Caveat: I am a computer scientist wit relevant experience, so I may actually know what I am talking about.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
As humans we hang on tight to things which affirm are superiority over other animals and the universe in general. Part of this is the notion that the mind is something mystical which can't be reduced to simple data processing. So when somebody comes along and claims to be able to put the entire spec for a human being on a microSD card, we get upset.
Even though, as you point out, the entire spec can be packed into an ovum and a sperm.
http://michaelsmith.id.au
"No man is an island, entire of itself; every man is a piece of the continent, a part of the main"
The human genome does not boot-strap to a human; humans are produced within other humans who were produced within other humans, etc., etc. back to the blue-green algae. That unbroken chain of reproduction is the self-modifying program; any given genome is only the latest bit of data fed into part of it.
Build a man a fire, he's warm for one night. Set him on fire, and he's warm for the rest of his life.
This is not exactly the case. There is a brain in a jar down the hall from me, but presumably we should not expect it to be "running" a mind. The neural tissue found in other mammals is the same as ours (Pinker, How the Mind Works), yet ours does something different.
When you "simulate" something, you are building something to function in the same manner as something else. You can't "simulate" the brain without knowing what it is the brain is outputting (referred to as the mind here for brevities sake). I can build something that looks very much like a brain, say, a connectionist network. Unfortunately, there are problems with the model that make it a poor model for everything the mind does. Thus, it is a poor model for an actual brain. I can simulate the brain using cheesits and string, but a mind isn't coming out of it.
Modeling the mind is critical in order to evaluate whether the models of the brain we are building accurately reflect what is going on insofar as we care about the emergence of a mind. I assure you that Sejnowski does care about the ermergence of mind. His claim would be uninteresting in the extreme if he did not.
One of the problems in cognitive science is finding a model which bridges our understanding of the mind with our understanding of the brain.
Actually, there is a fundamental reason for that. The problem is that, at least presently, we use TTL logic that requires distinct and stable detection of on and off states. Without that, errors creep in and things go wonky after that. It's the way we have designed every bit of digital electronic gear to date. The brain, on the other hand, uses a multitude of neural paths for any given action. It is the aggregate result that is then determined. It is technically possible to do that in silicon, but it hasn't been done yet and is only in the theory stages.
(I suspect you already know how digital circuits work and that clean and discrete signals are a requirement which is why clocking too fast results in more errors.)
As to all the other blather, I generally agree but the fundamental reason it can't be done is that it hasn't been done. As for how abstract mathematical models can be implemented in silicon? That's beyond me. Abstract isn't something computers are all that good at.
And as for the "untapped potential of the human brain" there is more to it than simply processing more or processing more quickly. We have been dreaming of this way to unlock the brain for ages since we first started claiming that "the human brain is only utilized at 10 to 20 percent capacity." Unfortunately, humans generally process one thought or idea at a time consciously. I utilize my subconscious on a regular basis, however which increases my abilities considerably... some people call it "natural instinct" but I call it subconscious processing... I don't know everything, but (average) people seem to think I do for some reason... I'm a really good guesser which is also fed by subconscious processing. In any case, I seriously doubt that adding more power will increase output in any form. People have to "learn how to learn" and that's where most people fail -- at the very beginning of their lives when parents and teachers do not teach children how to learn. I believe it is precisely at that stage were "untapped human potential" can be best developed. And if you happen to hook up a car battery to a kid's brain at the same time, who knows what might occur. (Yeah, I was kidding)
In any case, I'm fairly convinced that we are riding at the safe mark of human mental capacity. Just as in every other way we attempt to enhance ourselves, we usually end up doing enough harm to have second thoughts on the idea. Enhancing our own brain would likely result in bad things ranging from migraines to seizures with the occasional super-villain-genius bent on destroying the world.
And the huge hole in his theory is the execution environment, e.g., the cpu that the brain is running on is REALITY itself. So be sure to add that to your cost of simulation of the brain.
This is correct. Even if you would simulate a brain, which brain does it simulate? That of a newborn baby? How intelligent is that? Intelligence develops by interacting with reality, and with other actors. Within tight physical constraints. And it may well need a body to act with and to be an acting subject. There's much more to "intelligence" than just the hardware of the brain.
Simulate a brain and you've got something like a paralyzed newborn in a sensory deprivation chamber. You'd have to also simulate a body and a world around it to get to something resembling a human brain in function. And even then you'd probably just get a dribbling crazy Artifical Idiot.
Let's imagine a world without an oxygen fueled brain...
Ah! It's a simple world because there is distinct lack of brains, due to a lack of oxygen!
Without managing blood flow and fuel supply according to its needs? Way simple!
Let's emulate all of nature, eliminating all the things we deem 'irrelevant' to the simulation.. What gets left out? Whales? Plankton? Ants? Yeast? Pollen?
How complete is your implementation? Outcome not what you expected?
No it's not. If it was possible to do in a million lines of code, it would have been done by now.
That's silly. He's not claiming that just any million lines of code will do. You need to understand how the brain works in order to write the right million lines of (probably ridiculously compact and completely unreadable) code.
I always knew perl was good for something.
Its not that the Lisp-ish implementation of Cyc is particularly elegant since ( ...) syntax gets extremely unwieldy and everything can be described as an atom at some point, but that Doug Lenat has been at it since 1984.
By sheer rabid bloody-mindedness he has assembled a friggin' huge predicate base.
This behemoth is capable of some novel (almost inteligent :-) symbolic manipulation behavior; it just takes much too long to do it.
What is missing in all of these the the unfortunate fact that all current language implementations are interchangeably (as Turin showed by thee development of the Turing Machine,) incomplete.
None of the current and trivially interchangeable formal languages is capable of expressing a Relation and its instantiation as a Connection.
This is a fundamental concept in psychology and several other fields, like education. (You know, dealing with real intelligence, [what ever that is.])
Until we have a formalism capable of expressing that "Relation ties with "* as well as the current Object definitions, and an operating environment capable instantiating these Connections, we're pissing into the wind
* If you want the EBNF go screw yourself. I'm through with trying to make people see the evident. I'm too old to care.
MSBPodcast.com The opinions expressed here are my own. If you don't like 'em... Think up your own stuff.
Its not that the Lisp-ish implementation of Cyc is particularly elegant since (<operator> <predicate> ...) syntax gets extremely unwieldy and everything can be described as an atom at some point, but that Doug Lenat has been at it since 1984.
By sheer rabid bloody-mindedness he has assembled a friggin' huge predicate base.
This behemoth is capable of some novel (almost inteligent :-) symbolic manipulation behavior; it just takes much too long to do it.
What is missing in all of these the the unfortunate fact that all current language implementations are interchangeably (as Turin showed by thee development of the Turing Machine,) incomplete.
None of the current and trivially interchangeable formal languages is capable of expressing a Relation and its instantiation as a Connection.
This is a fundamental concept in psychology and several other fields, like education. (You know, dealing with real intelligence, [what ever that is.])
Until we have a formalism capable of expressing that "Relation <r> ties <object> with <object|relation>"* as well as the current Object definitions, and an operating environment capable instantiating these Connections, we're pissing into the wind
* If you want the EBNF go screw yourself. I'm through with trying to make people see the evident. I'm too old to care.
MSBPodcast.com The opinions expressed here are my own. If you don't like 'em... Think up your own stuff.
I think you are clueless when you say "Generally his ideas aren't taken very seriously by academia in Computer Science".
Lets list what he invented:
Decent optical character recognition.
Text to speech synthesis.
Speech recognition.
CCD flatbed scanner.
High quality music synthesizers.
All of these inventions were quite innovative, commercial successful, and revolutionized the ideas of what AI was. I think any real computer scientist would take these quite seriously.
That might be what the Gizmodo article implies, and it seems to be what set Myers' rant off - but nowhere does it quote Kurzweil as actually claiming that.
Ray has suggested in his books that we could simulate a brain, either functionally by reverse-engineering how the brain actually does things (not by studying the DNA), or if that fails by simulating all the various actual neurons sufficiently well and hoping that intelligence emerges (though that would take a few orders of magnitude more computing power). I don't see Myers addressing those claims.
Why would anyone engrave "Elbereth"?
A user ID of 1.5M+? STFU, noob.
Kurzweil didn't make that ridiculous claim in the first place, despite Myers' third-hand assumptions.
It was just an aside pointing out that the brain's overwhelming complexity all stems from a few million bytes worth of DNA, implying a significant level of replicated structure, andcertainly not a suggestion that we could derive a whole working brain from it.
Why would anyone engrave "Elbereth"?
Well stated. Raffaello provided these same arguments in a different manner, but you both provide a correct analysis of how DNA is the data, not the actual program. Biological cells operate nothing like a computer; there are only mild similarities which some people tend to focus on without considering all the other aspects there are to each system which are not present, via any equivalent, in the other.
GP is correct that the DNA chain does contain all of the information necessary to build an organism (which Raffaello labeled a 'parts' list), but, as you point out, it does not specify the environment, which is why any complex organism (any multi-celled body) requires an ancestor, usually a parent, to instigate growth. It also helps explain why two bodies with exact duplicate DNA (such as identical twins) do not remain identical as they age.
He's got a hypothesis ("I can extend my own lifespan by aggressively interceding with my body chemistry"), he's performing the experiment on himself, he's making careful observations on a weekly basis and he's taking meticulous notes. Results are probably some time off yet, but even if he turns out to be completely wrong, he's done the experiment in a way that can be reproduced, and it's still a genuine contribution to human knowledge.
How is that not "science"?
Why would anyone engrave "Elbereth"?
If you manage to simulate the mind of a nine month baby then you will know that it has a mind because it's body will be behaving just like a baby. Crying and sleeping. It would be quite striking.
I agree with you. However, if you meant this as an argument, then you actually meant to say "If you manage to simulate the *brain* ..."
Philosophers have argued that if something behaves like a person, we should assign it personhood attributions. This is how many argue for the possible valuing of AI and some animals. You are suggesting the same; Since it behaves like something with a mind, we can assume that it has one.
This argument is problematic. It assumes that we have somehow managed to perfectly simulate human brain function such that it leads to the emergence of a mind. It should be pointed out that, in order to accomplish this feat, *we must know things about the mind*. Successful models of the mind not only supply us with tools to help people with cognitive disorders, they also bring us closer to understanding how the mind actually emerges from the hardware (the brain).
I have to say that while he did an excellent job describing much of the aspects of how we believe the brain works today as opposed to the over-simplified model presented by the original article's author, the mistake made is that people able to understand what he wrote were not the people needing the debunking.
Let's face it, if you're able to read at the level which this guy wrote, then you're probably more than capable of understanding from the get-go that the original article was similar to "By 1975, we'll all be shooting around the skies in flying cars". I would like to see someone with good writing skills... at the popular mechanics level of writing take his response and phrase it in a way which would have an impact on the audience of people he was most likely concerned "would take the bait".
Either way, nice article. I know some about genetics and some about biochemistry, however I lack the ability to judge for myself what is fact and what is credible scientific speculation from his response. Sadly, while I learned a great deal from his response, I lack the knowledge to even know where to pursue this train of though in order to more clearly understand the issue. Either way, thanks for the education and something to think about.
His "critic" used the following steps:
1. take one of Kurzweil's statements out of context
2. build a strawman around it
3. attack the strawman
4. profit!
But to no avail, as the truth remains: even though the brain is remarkably complex, its design is coded in a very compact manner. It's true that to actually execute that compact code on a computer needs a "virtual machine" to simulate biochemistry (whether it's feasible or not), but it's obvious.
I sure would like to see the operating system this "software" will run on, it will have to be close to something like, e.g. reality.
Don't speak about time until you have spoken to him.
That's how much source is behind the human brain.
A Mandbrot set has like two lines of source behind it, good luck trying to paint the complete set. Size of source code has absolutely nothing to do with the complexity of executing it. You can have short code that is impossible to calculate and extremely long code that is trivial.
Starting to create the brain from the genome means nothing short of simulating each and every molecule and all their interactions. Super-Computers struggle with one molecule these days (see protein folding), good luck trying to do it with all the octillion or however many atoms we have in the body.
At some point we might now enough about the interactions to optimize, but we are nowhere near that. Starting from the genome is the by far most complicated way to get to the brain, it is far easier to just observe it and then try to simulate it based on those observations instead of trying to simulate it on a atom scale.
> He's looking at the genome, and then saying that you can build a working brain from that info
> alone. It may be theoretically possible, but is so difficult that we shouldn't even bother trying.
Your brain, like the rest of you, was grown from that info alone, so that is definite proof that it is possible not just theoretically.
If we ask Ray what he means by "brain" we may understand better, the part of the brain that is practical to reverse engineer first is logic, or the neocortex. Assuming that Ray means the whole brain is where I think neuro-scientists are correct in their conclusion as of now 10 years is very optimistic. On the other hand if we only consider the logic center and don't deal with vision centers, speech and hearing centers, motor functions and autonomic systems, then we can probably do it. The key problem is input, the brain is nothing without data input and interaction, and experience.
There is more to the human brain then logic, but at this point raw compute power which is approximately 20 Quads per second and developing subroutines to simulate neuron function is what I think Ray is speaking of, simulating sensory stimulus response may be part of that program. The co-processing centers that pre-process the input to signals the brain as whole understands is another problem, and I don't think this is what Ray is predicting, but I think we need to ask him what he means before we jump all over his statement like we know what he's talking about. The brain is an electrochemical streaming massively parallel biological processor, which seems to be very well suited to pattern recognition and linear logic calculations. The brain not well suited to exponential logic, so neuro-scientists are correct from their frame of reference but wrong in the actual application of their logic because it's linear.
Raw compute power and the interactive subroutines which are repeated over and over in the wet wired world of the brain, simulating the plasticity and electronic logic is probable but the chemical side of things is a whole other issue, this would involve understanding hormonal function and interaction at a level not available today. It would require the understanding of calcium, potassium, sodium and other ion channels and other mineral interaction and function in the brain, and it would require the mapping of the two-way interaction between antagonists and receptors, re-uptake of hormones and their effects on signal processing.
Is what Ray Kurzweil says clear? On the surface yes, but the meaning of what he is saying needs to be explained and not so quickly dismissed.
"Is that real poncho or a Sears poncho?" ~~FZ
Uhh no. Before this article I've never even heard of him.
Tesla was a genius. Edison however was a overrated hack who liked to torture puppies.
Actually, there is a fundamental reason for that. The problem is that, at least presently, we use TTL logic that requires distinct and stable detection of on and off states. Without that, errors creep in and things go wonky after that. It's the way we have designed every bit of digital electronic gear to date.
There is a large body of research on analog VLSI and neural systems, e.g. Carver Mead's book is 20 years old now.
Comment removed based on user account deletion
This would be apt if you had some prior reason to assume that the real Ray Kurzweil must have a low UID or that he actually has been confirmed to have a legit account. What if Ray Kurzweil never actually got a slashdot account for himself, and did so tomorrow (as sometimes minor celebrities/company reps do to respond to a popular posting on here). He would be more of a "noob" than this account.
The irony of a few people's indignation at my posting here is that my question was completely serious. Someone claimed that the real Ray Kurzweil had an account, and so I am actually curious which UID is the real Ray Kurzweil. I never claimed to be the real Kurzweil, that is merely your and a few other's unprovoked assumption.
It also amuses me a little that the relative age of this account to yours is not much less than its age relative to a newly created account. i.e. a little less than a year and a half vs 3 years or so.
Or was calling people noobs with a 7-digit UID your shtick? Well played, then.
PS: Go fuck yourself.
Also similar is Asimov's pyschohistory, which suggests that the problem might solved, in aggregate via statistical means.
Interesting ideas. But I think they are all wrong. The problems are combinatorial. Brains don't live in a jar. We're bombarded by information from the outside, and the complexity of these interactions is going to be something that can never be computed. Already at 59! you're at 10^80 combinations and have exceeded the number of protons in the universe. Good luck that that calculation. Ultimately our interactions with others is what makes us human. Put brain in a closed room with no contact and it will never rise above the level of a temperature regulator.