A brain in a supercomputer | Henry Markram
Vložit
- čas přidán 14. 10. 2009
- www.ted.com Henry Markram says the mysteries of the mind can be solved -- soon. Mental illness, memory, perception: they're made of neurons and electric signals, and he plans to find them with a supercomputer that models all the brain's 100,000,000,000,000 synapses.
TEDTalks is a daily video podcast of the best talks and performances from the TED Conference, where the world's leading thinkers and doers give the talk of their lives in 18 minutes. Featured speakers have included Al Gore on climate change, Philippe Starck on design, Jill Bolte Taylor on observing her own stroke, Nicholas Negroponte on One Laptop per Child, Jane Goodall on chimpanzees, Bill Gates on malaria and mosquitoes, Pattie Maes on the "Sixth Sense" wearable tech, and "Lost" producer JJ Abrams on the allure of mystery. TED stands for Technology, Entertainment, Design, and TEDTalks cover these topics as well as science, business, development and the arts. Closed captions and translated subtitles in a variety of languages are now available on TED.com, at www.ted.com/translate. Watch a highlight reel of the Top 10 TEDTalks at www.ted.com/index.php/talks/top10 - Věda a technologie
Watching and listening to these kind of videos always bring me back to the fundamental question I ask myself - what the hell am I?!
I can't believe how clear and concise this talk was, especially since it involved the brain!. It was so elegantly spoken and he skillfully avoided talking psycho-babel which I think we all appreciate :)
This is one of the best TED talks. Great speech.
These get better and better! Amazing what they are doing.
Astounding. I hope we see this technology evolve into something very interesting.
@Udinbak cont. It could see which connections are active during a particular thought, it could refine it's thoguht processes, edvice on improvements, or refinements. It could ask to sever a synapse in order to reduce an unwanted response, strengthen it to increase a desireable response, even request a new synaptic pathway to connect and explore disparate functions/memories/senses/etc (synesthesia). If the new region could make these adjustments, then it would be able to do it automatically.
@Flem1337 I'm a 17 year old (a dumb one at that) and I could understand most of what he said. Of course, I did need to search up a bit to have a more thorough understanding of what he was talking about, but hey, we never stop learning.
In face, this video, along with Jill Bolte Taylor's TED talk enabled me to get a 96% on my English essay about truth, reality and perception!
I love learning about the brain, and this project excites me!
just a few tips to consider when you've got the brain mapped and symulated..
basically consider giving it extra regions. by which I mean, for example, create a region which is given information on the brain ie the information the reasearchers are lookng at. not all of it, but at least the map of neural activity. then connect that region with the visual cortex. The point being to allow the simulated consciousness to "see" how it's brain functions. That way, it can give direct feedback.
Wow, that was awesome.. Also, the sentence "We can fire it up, and see what happens" is just so awesome :P
That was pretty amazing to see ! 5 stars !
I can produce a super-computer that will smoke any supercomputer in production today... it takes me about 9 months to build it, and then about 18 years to train it.
"it's blew a gasket"
Indeed. :) I wrote the same phrase earlier in this phrase and got it right. The typing fingers sometimes have a mind of their own; muscle memory, I guess. I have a real problem transposing 'think' and 'thing', do it all the time.
i haven't seen this video for a while now! BUT THIS IS AWESOME!!!! 8 years to go! lol
Once you build the circuit, how do you then model the dynamics of the morphology of the "circuitry" itself, that is, how the neurons re-wire themselves based on learning. If the circuitry is static, and the only dynamics are those of the circuit, namely, the cumulative firings between neurons, the brain is never really learning. Am I right? In other words, the brain architecture changes over time. How do you account for this?
woww this is what ive been waiting for from Ted!!
the electrical stimulation part around 13 minutes was incredible.
@Udinbak cont2. The region would have to be passive at first in order to train the new brain to understand the sensory experience, then active control gradually increased while the mind becomes accustomed to making self adjustments. safety protocols in place of course. active and passive control at the early phases would be counter productive, since the young mind would exploratorially make potentially damaging alterations which could render the mind insane. or the equivalent.. just a thought..
thanks ted, ive been waiting months to hear this talk.
This is very similar to Hierarchical Temporal Memory, and both of them are different implementations of Adaptive Resonance Theory.
That sounds very cool, what work are you referring specifically about?
This video is on the net only 1 day and look at the amount of (different) reactions ... Amazing subject.!!!
I actually looked this up by using my brain
when they activated it, it began experiencing and thinking. Learning is provided by experience. the moment they showed it a flower, it started to think about the flower and in the process learned about the shape and colour, but since it's out of context (imagine complete sensory deprevation, then suddenly you see a flower) it's ability to "learn" is limited.
This is amazing and inspiring.
Damn... My brain's just a jumble of equations?
I feel mortally insulted.
I like this video a lot -
too actually be able to map out the processes, the coding, the hard-wiring of neuronal circuitry is amazing - I can't wait to see the results.
Supercomputer can perform perfectly at Antarctica, Alaska, Greenland, Arctic, North Canada and North Russia.
This is really a great and commendable effort . It will help engineers built a living , evolving and adaptive hardware. Somthing like a standalone decision making machine . Well the flip side of it will more people of jobs .
Amazing!
@Udinbak cont3. the study of creating new regions in order to process information in new ways (like a region that analyses visual input and computes precise distances, velocities, mass, angles, weight, etc. or one that can interpret hmtl and has access to the internet) will be invaluable when you (the future) get round to addnig functionality to the human brain.virtualising a new region and integrating it into an unaltered human brain will be the key to transhuman consciousness.
Well put.
when they crack the code i hope they install firewall
i dont want flash adverts projected into my dreams lol
that's 100 billion laptops
Truly amazing
A deeply fascinating & most Intriguing subject....
Absolutely fascinating!
I'm glad there are people out there with a more... optimistic attitude towards the things we have yet to explain;)
Ofcourse it'll be solved. You think people a million years from now will still have no clue? But it IS a tough one, and I agree very much that what he is talking about here, has very little to do with the actual "feeling" of being a human being. Though, as soon as we have build the model, it'll be alot easier to approach the following topics.
the theory he is presenting here around the 13-15 minute mark is just mind-blowing. i doubt many people listening to markram for the first time realize what he is actually saying though..
A very facsinating talk, even if it's all theoretical. Though, it's nice to see that people are putting the effort into trying to figure out how to build a brain. I would love to see how these guys progress over the next decade.
This is absolutely fantastic.
The concept of being able to build a brain is both exciting and terrifying. I don't know if we'll have synthethic brains in 10 years but I might go pull Blade Runner off my dvd rack and give it a spin.
Amazing.
Such is the stable equilibrium.
Very inspiring talk.
When we approach such possibilities with sincerity, the results are a celebration of life.
It will be interesting to follow this experiment and see how far they get. I wouldn't think it takes an entire 'laptop' to simulate a neuron since neurons run much slower than computers, but who am I to say ;) I'm sure sentience is way way off though.
That's great! At least you have some sort of unseen sense of fashion.
very interesting presentation .... even the advertisement is interesting lol
Looking forward to his holographic presentation in 9 years!
the reason the moon 'looks' or 'seems' larger when it's near the horizon is due to a lens magnification effect caused by the light from it traveling through a larger section of the earths atmosphere.
another cool exercise for your brain- when you see part of an object, but the majority is hidden out of your view, your brain can finish the object and represent it in your mind. i find this to be a very important clue to how the universe around us is represented.
@Udinbak Do you think that when they activated it it began learning?
Interesting!
He didn't propose any new theory. That wasn't the point. He was filling us in on the progress of the research. Meet TED.
I don't know.. I can only comment on where I have been living for the last 30 yrs (UsA) it seems people got along with each other better in the 70 early 80s than they do today. A LOT better. These days it seems people are full of rage for some reason.. I don't know why..
As long as we continue our exponential progress in science, it seems very certain that this will eventually be solved considering how far we came from just the last century. Thing is, the final answer to this question might seem completely different or ever more complex than what we previously postulated.
Science is awesome !!!
Fascinating. If we are confined in a bubble or territorial space known as the universe and it provides at the extreme human possible extension of his potential, then it it could be seen as a deliberate design for "milking" / nursery / isolation virtual room for potentialities. In the layman's term if we keep on zooming neurons, we would be looking at a quantum representation of ourselves. Perhaps that is how we are created.. just a hypothesis among many other possibilities,
Really wish they'd discuss the actual equations they bring out in videos like this, I like having the math behind it explained.
yeah, that's what happens when people recite and not research. i've made knowing and researching the sky and the universe my life, and plenty know more than me. it doesn't bother me when someone questions me or says i'm wrong, it motivates me to double check my theories and beliefs, and i usually learn something new..
This is quite interesting because I've read that, a brain can grow infinitely big with a relative amount of energy
I agree, but it is hard to remain as open as most people when you are an "intellectual" free thinker, because one's subject of intellect often requires that person to dedicate a majority of his life to master the subject objectively. As a result, it would be expected for that person to be quite reserved to his theories.
Yes, it is possible to remain still open despite that. It takes a great person. And that's a lifelong journey, irrespective of intellect.
We always create to fill a need, with great variation on the definition of 'need'. There's no reason a sufficiently large and well-trained neural network or other brain simulation could not do the same.
@The7whoate9 in fact they would need to effectively map an entire nervous system between the arificial brain, and the skin of the in game avatar. however that could be done seperately, and could in theory be interchangable VR bodies for the minds to inhabit.The main reason you would do this, is to study how the human brain processes information from multiple sources to create the internalised holographic world experienced by the mind.
Just ten years away from when things realllly get moving...
yeah that was awesome
@Hsapienslaptopicus I don't know whether you meant that as an oblique insult, but I love HHGTTG!
Also, at the risk of losing all my credibility, I'll clarify: I meant to say "neurobiology"**. I frequently type one in place of the other. I know...I'm an idiot.
@winterstellar agree!
incredible
What about the 'hard problem of consciousness'?
They should have done a presentation on rolex watch. now thats high tech!
life creates life.
@matt7kiwi what's a spirit? how do you detect one and do you believe you posses one?
big bang - "one day there was nothing, then it went boom"
I am in LOVE with SCIENCE!!!!
Seeing as these simulations are run on computers, and thus are completely and understandably deterministic, any simulation you run on it can be perfectly recreated at any time.
Besides, I highly doubt that they are going to get anywhere near what we would recognize as consciousness in anything like the time table he laid down.
And, as we can see, Raymond Kurzweil was right: the Singularity is near !
Imagine a fusion of Henry Markram simulation of the brain and Edward Witten's M T heory, wich includes 7 extra dimensions and there correspondence with our 3 D - time world. I wonder how the picture of the rose would 'reflect"in 11 dimensions.
It makes perfect sense if you don't intend on being mystical. The 'projection' is just the brain's comprehension of the world, which is translated to motor output, which produces new input and further 'projections' ad infinitum. So yes, it's infinite in a sense, much like any other piece of non-halting computer software or other information processes for that matter, until it is forced into conclusion through an external force of nature.
AMAZING
Hmmm...there are some interesting consequences of the points that have been raised here pertaining to mental equilibrium.
I would use an analogy that nearly all psychotropic drugs (especially SSRI's) are tantamount to being nothing more than blunt instruments, like using a chain saw for brain surgery, conversely entheogens that attenuate essential neurotransmitters such as Serotonin or dopamine can be likened to being a precision "scalpel" like tool.
Way way cool.......
Knowledge is power.
@neverthat79 Fair point. I took his use of the word "evolving" in the sense of darwinian evolution, but if he simply meant "It's taken 11 billion years for brains to appear in the Universe" and "brains continued to change rapidly" then that's fair and I take back my comment!
Right, but the point of what he was saying with all the perception stuff was that you only believe you can think infinitely. If you have a single neuron and you stimulate it, all it can know or think is either "yes" or "no".
Through millions of these networked together a more complicated "infinity" emerges. There is no magic in your brain. If it can think of infinity, then there is a reason for it, and that reason can be measured / understood / reproduced.
@leostoltoy
If the universe is around 15 billion years old, than it took it around 11 billion years to create the first brain. Which also goes together with what he says later, that the universe has evolved a brain to see itself. Also reminds one of Carl Sagan: "We are the way the cosmos can know itself"
@neymoura But we don't. There is only a tiny, very short period in the evolution period of a conscious being during which the being is occupied with emulation of living creatures. Our road towards pecfection is very long, and we are almost in the beginning. In a matter of few centuries or millenia humanity will evolve to a certain level, when that happens we won't be occupying our lifes with such things.
@Udinbak I agree. Smell and taste is (just) chemical impulses. We have detectors that can identify chemicals/smells. But the blue brain will be the only machine to experience it as we do.
You have to admit, artificial intelligence is a truly fascinating topic. If there's one thing I wish I could see before I die is for someone to build a "self aware" computer.
he says a neuron per laptop, if i'm not mistaken. And 10,000 per neo-cortical-column
so consciousness is a series of decisions... that what I got from it
I agree. So the craft of appliance by powerhouses could be called a tool of evolution?
@Udinbak however once this has been created, there's not much stopping us from meeting them in their own reality. even if it's not fully emersve for us.
As live4Cha said in a response to me, regarding a factual claim I made, "You can hold on to it if you it gives you peace." As if beliefs should be dictated not by the facts, but by what makes you feel good. This, sadly, is overwhelming true for the vast majority of human beings, even for the most objective among us.
@Mood1ndigo Well Said
The world of Science and Consciousness are coming together. His great studies will revolutionize the world. This Presentation, as you view it, is one of the links that directs and propels my studies into a higher level of thought. Thank you, Dr. Angel De Jesus
@The7whoate9 technically the AI will need more sensory input than simply visual, auditory and collision detection. but it shouldn't be too difficult to include smell, taste, temperature and pressure to the game environment. We won't fully appreciate the game untill we invent a means of fully immersive VR. porting them into a normal game would be like having your skin anesthetized and your tastebuds and olfactory receptor neurons lazered off.. and is is making good sci-fi, it's called caprica :P
It happens both ways all the time.
Ethics and free will will be the questions.
There are many good things that can come from this, but the seduction of illegally controlling another that someone else has decided needs "fix" will be a big issue that will need to addressed.
@winterstellar agreed.
Where does the computer's spirit go when the lecturer turns its power off?
Isnt that a wonderful question that i wish i could answer. lol
@neymoura I'm counting on it