On Mon, Jun 28, 2010 at 11:28 PM, rob levy <> wrote:

though I wonder what it actually would be in terms of neurons, (and if that
>>> matters).
>> I don't see any route to the answer except via neurons.
> I agree this is true of natural intelligence, though maybe in modeling, the
> neural level can be shortcut to the topo map level without recourse to
> neural computation (use some more straightforward computation like matrix
> algebra instead).

Your suggestion involves the application of a priori knowledge of real world
structure, which may or may not be valid, and which may or may not be
persistent. Further, it addresses only one tiny stone of the mountain of
unknowns regarding GI operation.

There is a basic message here that I keep saying, but people keep missing.
Of course, in situations like this, the message sender (me) must accept
responsibility. Anyway, here goes yet another attempt to transfer my grok. I
wonder if anyone will grok it?

When neurons and their associated math is finally understood, there will
doubtless be SOME sort of symbolic math that preserves the essence of what
they do, but without the baggage of simulation. However, we do NOT presently
have any (realistic) idea what that symbolic math might be. However there is
plenty of evidence (from the slowness of evolution, the large number (~200)
of neuron types, etc.), that it is many-layered and quite complex, and
further (from our own million years on earth) that we have JUST BARELY
enough GI to advance. Hence, we seem to presently be a LONG way from being
able to program anything to equal it. Hence, I see absolutely NO prospect
for present AGI efforts to produce any sort of GI, people's intuition to the
contrary not withstanding. Certainly, not until the missing math is in hand,
and history has shown math to be VERY slow in coming.

Note my recent past posting regarding the possible use of Genetic Algorithm
(GA) based approaches to "fill in the blanks" in neuronal functionality.
This approach would be mentally difficult to actually make work, but at
least it provides SOME prospect for forward progress. Certainly, ignoring
the unknowns and writing code can never work, as we just don't have enough
monkeys, nor enough keyboards, to ever make it work.


RSS Feed:
Modify Your Subscription:
Powered by Listbox:

Reply via email to