Slashdot Mirror
Kurzweil on the Future
dwrugh writes "With these new tools, [Kurzweil] says, by the 2020s we'll be adding computers to our brains and building machines as smart as ourselves.
This serene confidence is not shared by neuroscientists like Vilayanur S. Ramachandran, who discussed future brains with Dr. Kurzweil at the festival. It might be possible to create a thinking, empathetic machine, Dr. Ramachandran said, but it might prove too difficult to reverse-engineer the brain's circuitry because it evolved so haphazardly. 'My colleague Francis Crick used to say that God is a hacker, not an engineer,' Dr. Ramachandran said. 'You can do reverse engineering, but you can't do reverse hacking.'"
How is haphazardly hacked together code any harder to reverse engineer than intentionally obfuscated code? We know the latter isn't a problem for a determined hacker....
Only three things are certain; death, taxes, and apocryphal quotations - Ben Franklin.
Taking into consideration computer security issues, I think I'll pass.
Proverbs 21:19
Furthermore, looking at the broader picture, I was reading an artificial intelligence textbook that was available online (sorry, can't recall it offhand) which said that current AI researchers have shifted their focus from "how do humans think?" to "how would an optimally intelligent being think?" and therefore "what is the optimal rational inference you can make, given data?" In that paradigm, the brain is only important insofar as it tracks the optimal inference algorithm, and even if we want to understand the brain, that gives us another clue about it: to improve fitness, it must move closer to the what the optimal algorithm would get from that environment.
Information theory is life. The rest is just the KL divergence.
I attended a talk by Kurzweil a couple of weeks ago at the Broad Institute in Cambrige, MA. Absolutely fascinating what he foresees in the near future (~20 years). I believe it is 2028 when he believes a machine will pass the Turing Test. Even sooner, he predicts that we will have nanobots roaming around inside our bodies, fixing things and improving on our inherent deficiencies. Very cool. He also addressed a similar complaint about being able to reverse-engineer the brain, but it was of the nature that we may not be smart enough to do so. I (and he of course) doubt that that is the case. Kurzweil thinks of the brain as a massively parallel system, one that has very low signaling rate (neuron firing) compared to a CPU which it overcomes by the massive number of interconnections. It will definitely be a big problem to solve, but he is confident that it will be.
Kurzweil's predictions will come to pass, by not on the time-scale he envisions. probably centuries. He has been hoping for personal immortality through technology and takes over 200 anti-aging pills a day.
You can't reverse-hack? Who says?
You can reverse engineer anything. Whether it has a well-thought out design or not, its functions can be analyzed and documented and re-implemented and/or tweaked.
If anything, the timetable may be in question, but not the question of whether or not it can be done. I have no doubt it can be done, it's just a matter of how long it'll take given the right resources, the right talent, the right technology, and the right knowledge.
Granted, I'm just an idiot posting on slashdot, and not an inventor or neuroscientist, but I still think I'm right on this.
You see? You see? Your stupid minds! Stupid! Stupid!
AI is our generation's flying car. It's what we see in the future, not what will be. Instead of the flying car, we got the internet. It isn't very picturesque (especially over at goatse.cx), but it is cool.
The future will be like that: something people aren't really predicting. Something neat, but not flashy.
Alternatively, the future will be the "inverse singularity" -- you know, instead of the Vinge godlike AI future singularity of knowledge, there could be a singular event that wipes out civilization. We certainly have the tools to do that today.
Use the Firehose to mod down Second Life stories!
Where's my motherf'ing flying car?
With current technology I can very easily move some 20-30gb in my stomach (2gb microsd cards in some small pill-like protective case). But download is then a little shitty...
Extreme Programming - Redundant Array of Inexpensive Developers
IANANS (I am not a Neuroscientist), but as with other approaches of interfacing the human brain with periphery it seems to work really well to let the brain do the hard interfacing work.
;)
So, as haphazardly as the brain structures, memory storage, sensory input, etc. might have evolved, it might still be flexible enough to figure out a sufficiently simple interface with anything you might connect to it. Given a smart training of finding the newly connected "hardware", it might be possible to interface with some really interesting brain extensions.
The complexity and the abstractness of the extension might be limited by the very pragmatic learning approach of the brain, making it more and more difficult to learn the interface if the learning progress is too abstract/far away for the brain to "stay interested". Though maybe with sufficiently long or "intense" sensory deprivation that could be extended a bit.
My problem with the argument of the "haphazard" structure of the brain is that it could have been used to deny the possibility of artificial limbs or artificial eyes, which both seem to work pretty well. Sure, these make use of already pre-existing interfaces in the brain, but as far as I know (not very far) the brain is incredibly malleable at least during the first 3 years of childhood.
So, as ethically questionable as that may sound to us, it might make sense to implant such extensions in newborn babies and let them interface to them in the same way they learn to use their eyes, coordinate their limbs and acquire language.
Good times
He not only makes predictions about technology (which is a feasible endeavor, though fraught with difficulties), but also about the universe that the technology will interact with. Predicting that brain scan technology will improve is (pardon the pun) a no-brainer. Predicting that we will map out hundreds of specialized areas within the brain is a prediction that is completely off the wall, because we don't know enough about brain function to know if all areas are specialized.
If you haven't been down-modded lately, you aren't trying.
Sacred cows make the best hamburger.
by the 2020s we'll be adding computers to our brains and building machines as smart as ourselves
As a cyborg myself, I don't see any sane person adding a computer to his brain for non-medical uses.
I was going to say that sane people don't undergo surgery for trivial reasons, then I thought of liposuction and botox for rich morons, and LASIK for baseball players without myopia. I don't see any ethical surgeons doing something as dangerous as brain surgery for anything but the most profound medical reasons, like blindness or deafness.
As to the "as smart as ourselves", the word "smart" has so many meanings that you could say they already are and have been since at least the 1940s: "1. to be a source of sharp, local, and usually superficial pain, as a wound." Drop ENIAC on your foot ans see how it smarts. "7. quick or prompt in action, as persons." By that definition a pocket calculater is smarter than a human.
Kurtzwiel has been saying this since the 1970s, only then it was "by the year 2000".
We don't even know what consciousness is. How can you build a machine that can produce something you don't understand?
mcgrew's razor: Never attribute to stupidity that which can be explained by greedy self-interest
The brain is an adaptive system. Provide it with a stimulus, and it will reprogram itself. How do you think the monkey learned how to use the robotic arm? Did they hack into the neurons and input code to work a third arm?
No, the monkey's brain spontaneously created the neural network to control it. Sentient beings aren't computers, at least not in the conventional sense, because they reformat themselves to process new data (learning), and even to process new types of inputs. One might be able to build a computer advanced enough to handle this level of functionality, but once it is built, you won't be programming it with code. Instead, you'll be teaching it just like you do a child.
The future Conan???
PS Anyone having trouble getting their rightful Karma bonuses despite still having 'excellent' Karma?
If you can read this... 01110101 01110010 00100000 01100001 00100000 01100111 01100101 01100101 01101011
Without the ability to install properly open code, I suggest a good security patch, like zen, or some other semi-mystical skepticism.
Damn those pesky terrorists
...that means, in 12 years we will be able to write code as complex as our brains without any catastrophic bugs that crashes it frequently or leads to totally useless results?
I don't think so. Computers are now faster and "bigger" (not physically) than 30 years before. Programs have more functions than 30 years before.
But essentially, they do exactly the same thing as 30 years before, just MORE of the same thing. And they don't do it BETTER, they still have the SAME BUGS, and the SAME NUMBER of bugs per lines of code.
Computer programs are made by humans, and I don't think that we - the humans, the creators of these computers - will evolve faster in the next 12 years than we did in the last 30 years.
If so, let's hope there's a grammar check on the data beforehand.
Kurzweil is one seriously messed up scientist. This guy extended the Moore's law (look up Kurzweil's law of accelerating returns) to predict that civilization as we know it will cease to exist and will effectively become the civilization of super/trans-humans or artificial intelligent beings by 2020.
Never mind all the scientific or technical obstacles that even non-scientific person could think of, let alone once we get into philosophical issues (for things we don't even have words to talk about yet).
Yet there is still a very simple reason why the prediction will not happen. Does he know how long it takes for FDA to approve a brain implant of the kind he is suggesting (even if we had one)?
I've said it before and I will say it again. This is nothing more than a religion posing as pseudo-science from a guy who takes 200 anti-aging pills hoping to reach immortality though technology.
But one thing is for sure, Kurzweil will die just like every other "prophet" before him.
As the island of our knowledge grows, so does the shore of our ignorance.
It might be less low hanging than most people think. Most predictions I've seen for, basically, "OMGWTFBBQ, computers are gonna be as intelligent as humans" are based on, basically, "OMGWTFBBQ, we'll soon have as many transistors on a chip as there are neurons in a human brain." Especially marketing depts love to hint that way now and then, but they're not the only culprits.
Unfortunately,
1. A neuron isn't a transistor. Even the inputs alone would need a lot more transistors to implement at our current technology level.
An average brain neuron takes its inputs from an _average_ of 7000 other neurons, with the max being somewhere around 10k, IIRC. The vast majority of synapses are one-way, so an input coming through input 6999 can't flow back through inputs 0 to 6998. So even just to implement that kind of insulation between inputs, you'd need an average of 7000 transistors per "silicon neuron" just for the inputs.
Let's say we build our silicon transistor to allow for 8k inputs, so we have only one modul repeated ad nauseam, instead of custom-designing different ones for each number of inputs between 5000 and 10000. Especially since, we'll see soon, that number of inputs doesn't even stay constant during the life of a neuron. It must accomodate a bit of variation. That's 2^13 transistors per neuron just for the inputs, or enough to push those optimistic predictions back by 13 whole Moore cycles. Even if you believe that they're still only 1.5 years each, that pushes back the predictions by almost 20 years. Just for the inputs.
2. Here's the fun part: neurons form new connections and give up old ones all the time. Your brain is essentially one giant FPGA, that gets rewired all the time.
Biological neurons do it by physically growing dendrites which connect to an axon terminal. A "silicon neuron" can't physically modify traces on the chip. You have to include the gates and busses that switch an input to another nearby source from thousands available outputs of another "neuron". _Somehow_. E.g., a crossbar kind of architecture. For each of those thousands of inputs.
Now granted, we'll probably figure out something smarter out, and save some transistor for that reconfiguration, but even that only goes so far.
There go a few more Moore cycles.
4. And that was before we even get to the neuron body. That thing must be able to do something with that many inputs, plus stuff like deciding by itself to rewire its inputs, or even (yep we have documented cases) one area of the brain decides to move to a whole other "module" of the brain or take over its function. It's like an ALU deciding to become a pipeline element instead in a CPU, because that element broke. In the FPGA analogy, each logic block there is complex enough to also decide by itself how it wants to rewire its inputs, and what it wants to be a part of.
There are some pretty complex proteins at work there.
So frankly even for the neuron body itself, imagining that one single transistor is enough to approximate it, is plain old dumb.
5. And that's before we even get to the waste we do with transistors nowadays. It's not like old transistor radios, where you thought twice how many you need, and what else you could use instead. Transistors on microchips are routinely used instead of resistors, capacitors, or whatever else someone needed there.
And then there are a bunch wasted because, frankly, noone ever designs a 100 million transistor chip by lovingly drawing and connecting each one by hand. We use libraries of whole blocks and software which calculates how to interconnect them.
So basically look at any chip you want, and it's not a case of 1 transistor = 1 neuron. It's more like a whole block of them would be equivalent to one neuron.
I.e., we're far from approaching a human brain in silicon. We're more like approaching the point where we could simulate the semi-autonomous ganglion of an insect's leg in silicon. Maybe.
6. And that's before we get to the probl
A polar bear is a cartesian bear after a coordinate transform.
Its often dangerous to extrapolate an exponential trend, much less a linear trend because they can have the nasty effect of flattening out or even turning over (bubble investing). Who knows whether we are at the middle, base or top of the curve for computing or biotechnology?
Ok, let's consider it.
Which is basically a non-factor, since the wiring gets it back to the right orientation anyway.
And it does saccades that not only allow it to see in any direction anyway, but also greatly increase resolution.
It's got a low light mode, unlike most modern cameras which become 100% useless in low light. Most cameras you can buy need a flashlight even in relatively well artificially lit rooms, and become freaking useless at the light levels where the eye becomes predominantly B/W. So, hmm, between going monochrome and going blind, it seems to me that the eye wins, hands down.
Only in as much as any other piece of biology is. Even so, it can withstand a lot of things which would render a cheap camera useless. And it can self-heal from most things.
But it's wired to something which can do a reasonable job even with an unfocused image. Try an OCR or, better yet, image recognition in the same conditions, and you'll see some epic fail.
But let's talk about some other advantages:
- better resolution than almost any digital camera
- saccades help increase the effective resolution even more
- some image processing and compression is built right into the retina, so it needs _far_ less bandwidth on the optic nerve than a modern camera would
- takes up less space than a camera able to focus over the same range of distances, and get similar image quality. (Hint: it doesn't need to move the lens waay forward and back to focus.)
- can deal with a wider range of brightness in the same image (most cameras need postprocessing so if the bride looks ok, the groom doesn't look like a light-sucking black hole, or viceversa)
- it can even rewire itself to deal with stuff it wasn't designed to deal with. E.g., you can get a camera-style photo-receptor as an implant against blindness, and the neurons in the eye and brain will rewire themselves to work with the fundamentally different image it gives. (That's one amazing thing about neurons: they can essentially reverse-engineer almost any kind of body, and learn to use it.)
Etc.
Now I'm not saying it's _perfect_, nor "proof of creation". But it's a lot better than you seem to assume, anyway. We're not quite at the point where we can equal it. Yet. We will be eventually, but not yet. We can do better in _some_ aspects, but often at the price of doing something else worse.
A polar bear is a cartesian bear after a coordinate transform.