• Log in with Facebook Log in with Twitter Log In with Google      Sign In    
  • Create Account
  LongeCity
              Advocacy & Research for Unlimited Lifespans


Adverts help to support the work of this non-profit organisation. To go ad-free join as a Member.


Photo

Is the Singularity really near ?


  • Please log in to reply
65 replies to this topic

#61 PWAIN

  • Guest
  • 1,288 posts
  • 241
  • Location:Melbourne

Posted 12 December 2008 - 12:13 AM

Silicon/computer simulation runs a lot faster than evolution, so I don't think we can bring evolution into the argument on equal footing and say it will take eons to reverse engineer the brain.



Entirely dependent on the complexity of the simulation. Sufficiently complex simulations (ie. Blue Brain) are slow and expensive. And it is only simulating a small chunk of neocortex.


Slow and expensive right now, much slower and much more expensive 5 or 10 years ago, much faster and cheaper in the future. Ultimately the job for the whole brain will be done - it must be as the brain is finite. Then it will be down to the computing power to run the simulation. This doesn't look like it will be a problem in the not too distant future.


Your opinions are interesting, more so because you are both neuroscience majors. I never said that the singularity is definitely going to happen in my lifetime; i'm optimistic about it, though: no blind faith, just hope. But you seem to be certain that we won't achieve it.

I wanted to hear, out of curiosity, the reasons you think it is so impossible to reverse engineer the brain completely in this century (and reach a singularity), mygheus, i still haven't heard it from you, but polyfractal gave a bit of a more satisfactory answer.


The brain can't be so complex that someone can say that we won't manage to completely understand it in this century. 100 years is a lot of time, considering the current state of our technology/knowledge and how fast it's advancing.


I think there are several reasons

  • Biological research is just damn slow. I work in a lab and the progress of discovery is glacial at best. Even with the combined output of all the labs around the world, research is a slow, accumulative process.

But technology has helped accelerate it, no? Thinking of the decoding of the human genome as an example.

  • Progress in biology is not the same as progress in technology. With technology, most of the progress is inventive, relying on the cleverness of humans to create something new. Biology is about discovery, trying to figure out how the blackbox works. This is a lot slower. Furthermore, it is a living blackbox. If you mess around with the wrong portion, you may kill it. For instance, the protein we study is absolutely required for life, knocking it out of mice kills them at birth. This makes it a very hard protein to study.

  • Again, technological tools can aid in speeding up these discoveries.

  • Biology is really, really complicated. See this wall poster to give you an idea. And that is just a small snapshot of what is really happening in your cells. Imagine being tasked to reverse engineer the Large Hadron Collider, except you have to stay 15 miles back and use a telescope to observe it.

  • But like everything, it can be broken down and componentised and the load shared amongst many.

  • Theoretically assume you could make some of those fancy "nanobots" that read the state of every neuron. Even with the activity of every neuron in your brain recorded, you are no closer to understanding how cognition works. This is generally the issue with neuroscience research. We have literally mountains of data about the mechanics of neurons (ie. how they fire, what proteins are involved, etc etc). We also have mountains of data regarding the behavior or animals and humans. However, there is a great divide between the two. We still can't link how neuron X, Y and Z's firing affects your behavior. And this is fundamental to any type of "singularity" event.
  • Even assuming that we have figured out how the brain works in its entirety, it is conceivable this is a one-way road. You can probably "download" your brain and run in on a simulation. But I doubt you'll be able to "upload" into your brain. Memories, for instance, are likely highly distributed patterns of firing that is spread throughout your entire brain. And these memories are layered over the top of distributed firing patterns of other memories. How do you tease these apart? And how do you "record" new ones? You can't just stick a probe in and stimulate some neurons to add new memories. You'd have to alter the firing patterns of a whole subset of neurons, which would mean altering their protein expression patterns selectively.
  • I'm not saying the task is impossible, merely monumental. I do not doubt that humans will become more mechanical very quickly. I also do not doubt we have brain-computer interfaces soon. But these are a far cry from understanding how the brain works in its entirety.


    Do we really need to understand how the human brain works or merely how to emulate it? If Blue Brain can create a good enough emulation of the human brain, and we can play with a few variables, it seems fairly reasonable that we could make an emulated brain that is both faster and more creative - since we already see variability in the human population. This "smarter" brain could then be harnessed to help us improve itself. This would then form a positive feedback loop as each superior brain is used to create an even better brain. We may still not understand how the triggering of neurons translated to an emotion but we will still get the benefit of the AI. As the enabling technology costs come down, we will be able to run these brains in large numbers.

    The only real issue is how do we deal with intelligence that is greater than our own and has free will.

    #62 forever freedom

    • Guest
    • 2,362 posts
    • 67

    Posted 12 December 2008 - 02:19 AM

    I'm troubled with statements like:

    "... whole brain emulation isn't really necessary for the singularity to occur..."

    Suggests that some *magical* technological Big Bang is going to occur in time.



    No, the statement merely means that we can find other ways to create human-level AIs.


    I know what it means. :) And it remains magical unless you can explain HOW.


    I'm really sorry i have only one link (maybe someone can help with that), but there are several projects running on that intend on creating and developing AIs using various methods other than brain simulation.

    This is just one of such projects:

    http://www.itnews.co...econd-life.aspx

    sponsored ad

    • Advert

    #63 Prometheus

    • Guest
    • 592 posts
    • -3
    • Location:right behind you

    Posted 12 December 2008 - 03:55 AM

    I'm troubled with statements like:

    "... whole brain emulation isn't really necessary for the singularity to occur..."

    Suggests that some *magical* technological Big Bang is going to occur in time.



    No, the statement merely means that we can find other ways to create human-level AIs.


    I know what it means. :) And it remains magical unless you can explain HOW.


    I'm really sorry i have only one link (maybe someone can help with that), but there are several projects running on that intend on creating and developing AIs using various methods other than brain simulation.

    This is just one of such projects:

    http://www.itnews.co...econd-life.aspx



    Fundamental problem with the AI/computer science approach is that they think they can treat consciousness as a black box without having a clue as to the underlying processes that brought it about. It's like modeling a car by simulating the noises it makes and the exhaust fumes it produces when it is driven..

    #64 j0lt_c0la

    • Guest, F@H
    • 43 posts
    • 0

    Posted 24 February 2009 - 07:13 AM

    Fundamental problem with the AI/computer science approach is that they think they can treat consciousness as a black box without having a clue as to the underlying processes that brought it about. It's like modeling a car by simulating the noises it makes and the exhaust fumes it produces when it is driven..


    That's pretty off the mark. We know what consciousness does, while your example is reminiscent of cargo-cults, who knew nothing of the technologies they emulated but the appearances. It's more like an emulator of a proprietary OS: we know what we want it to do, even if we have to implement it in a totally different manner because we don't have the source code.

    Edited by j0lt_c0la, 24 February 2009 - 07:14 AM.


    #65 Prometheus

    • Guest
    • 592 posts
    • -3
    • Location:right behind you

    Posted 24 February 2009 - 07:17 AM

    We know what consciousness does

    We do? Could you tell me please? Seriously.

    sponsored ad

    • Advert

    #66 valkyrie_ice

    • Guest
    • 837 posts
    • 142
    • Location:Monteagle, TN

    Posted 27 February 2009 - 09:41 AM

    2. "By 2009, we'll have full immersion visual and auditory virtual reality"



    In the year 2001 I was working for Sony computers. I presented to them the concept of a full immersion VR rig that utilized the Micropower Impulse Radar Chip developed by Larwence Livermore Labratorys. This device consisted of a small platform with four radar sensors to map the human body at a speed of 60 frames per second, without need for gloves bodysuits or other devices. By linking this device to a pair of video goggles with stereoscopic lenses and a pair of headphone, it, in theory, should have allowed Sony to make a competative cost VR rig that could have competed with the best of the incredibly expensive VR setups in existance at the time.

    This chip, who's stated capabilities in the articles I had read, possessed sufficent sensitivity to distinguish individual soft tissues in the human body, and was being used to detect the level of fluid in beer vats, studs in walls, and was projected to be in use soon as a disposible medical diagnostic unit for the military. It was developed for particle tracking in a cyclotron, and could distinguish it's own patterns of signals from another chips, so that multiple chips could operate together without cross interference. It's total radiation output under continuous use was less stated to be less than a florescent light's.

    But, Livermore screwed up the licenseing, and Sony wouldn't touch it with a 300 foot legal cattleprod.

    BTW if anyone here is working in an industry where they may be able to utilize this concept to see if it can indeed be used to create inexpensive VR, feel free to use the concept. Sony only owns the rights to the design I submitted. Whatever you come up with is unlikely to be an exact duplicate. I'd rather see VR become a reality than allow it languish in Sony's junk box.


    Kurzweil wasn't wrong. He simply failed to take into account human stupidity.

    Edited by valkyrie_ice, 27 February 2009 - 09:57 AM.





    1 user(s) are reading this topic

    0 members, 1 guests, 0 anonymous users