Re: [sl4] The Jaguar Supercomputer

From: Mu In Taiwan (mu.in.taiwan@gmail.com)
Date: Mon Nov 23 2009 - 04:01:39 MST


My angle is more:

- capturing the state of an entire human brain is definitely SL3 or SL4
technology.
- may not even be possible, ever.

- capturing the learning behaviour of some neurons, etc. is very possible.
- capturing the structural setup on a general human brain is very possible.
- but those neurons/structure then need to be trained into a human starting
from zilch, much as a baby's brain (probably) does.

I suggest we build brains equivalent to baby's brains, in some sense,
because it seems easier and the other option might actually be impossible
anyway. I'm not suggesting copying a baby. I'm talking about approaches
based on rebuilding the template.

G

On Sun, Nov 22, 2009 at 9:52 PM, Luke <wlgriffiths@gmail.com> wrote:

> Aye,
>
> I was assuming a situation in which you'd measure an adult brain and then
> evolve the neural net to emulate that. So in the method I mentioned there
> would be no "baby stage".
>
> However, this brings up an interesting notion: IF the technique works (and
> we have no idea whether it would), it would be interesting to train a
> computer up to baby standards. Might give us a lot of insight into the
> information patterns within a baby's brain. Nature/nurture problem, all
> that.
>
> - Luke
>
>
>
> On Sun, Nov 22, 2009 at 4:40 PM, Mu In Taiwan <mu.in.taiwan@gmail.com>wrote:
>
>> >re: 1) When I said "hogwash", I was referring to the statement "IBM
>> simulated a cat cortex"
>>
>> Fair enough. I'm inclined to agree.
>>
>> > What about being a "useful human" might not be captured if you were able
>> to
>> > capture the behavior of that particular brain?
>>
>> I mean a 'fully trained for human society' brain. e.g. your brain vs. a
>> newborn baby's brain. This appeared to be what another poster was referring
>> to when they spoke of taking years to train brains.
>>
>> The problem of turning 'baby brain' into 'adult brain' is known to take
>> years of training.
>>
>> The problem of turning 'unconfigured computer simulation of a physical
>> brain' into 'configured computer simulation of a physical brain' appears to
>> be a different problem.
>>
>> Of course, you might find that once you've got a working simulation of a
>> physical human brain, you then also need to take it from 'baby' to 'sociable
>> human adult'. I'm just noting that these seem to be different kinds of
>> training problems.
>>
>>
>>
>>
>>
>>
>>
>>
>



This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:05 MDT