Re: AGI motivations (Sidetrack on Uploading)

From: Richard Loosemore (
Date: Mon Oct 24 2005 - 10:30:45 MDT

Michael Wilson wrote:
> Michael Vassar wrote:
>>Trying to build a human derived AI is the general case of which
>>building a human upload is a special case... I'm skeptical about the
>>diferences being unavoidabld. Surely true given current computers,
>>surely false given uploads. There is a substantial space of possible
>>AGI design around "uploads".
> This is true in principle, but not in practice. Yes, uploads occupy
> a small region of cognitive architecture space within a larger region
> of 'human-like AGI designs'. However we can actually hit the narrower
> region semi-reliably if we can develop an accurate brain simulation
> and copy an actual human's brain structure into it.

Okay, I want to make a (probably futile) attempt to steer the
conversation away from questions of uploading, because I think we are
casually using the term "uploading" as it were technically feasible, and
the state of the art is so far away from that at the moment that we are
in danger of wasting our breath.

[I feel a little like we are all physicists at the turn of the 20th
century, talking glibly about chemical manipulation of molecules (doable
at the time) and similar manipulation of hypothetical sub-nuclear
particles, as if the latter were just basically the same, but a little
more difficult.]

We must not allow ourselves to be fooled by the exaggeration going on in
neuroscience at the moment, and start believing that we might be close
to being able to do this. It is extremely difficult to get down to that
level of detail with techniques that are here, or with techniques that
are just over the horizon, or even several horizons away. I seriously
doubt that we will be able to upload until such time as we get the AGI
and FAI problems sorted out, *then* get our AGI to invent enough
nanotechnology to disassemble a brain and upload it.

Remember what uploading would involve. You probably have to know the
size, shape and possibly the chemical state of every neural cell and
every one of its connections to all others, and maybe even the state of
all the synaptic buttons, in three dimensions, in circumstances where
the darn things are not broadcasting much EM but hiding most of their
signals in chemical waves.

Does anyone out there on the SL4 list have to hand the spatial and
temporal resolution of the best of the fMRI systems? Anyone care to do
a quick scale comparison with a synaptic button and an action potential?

The only good argument against this is that we might not need the
resolution. But at the moment we are clueless about the exact
functional significance of the detailed structure. For example, maybe
spike synchrony in the dendritic tree is crucial at the functional level
... and if it is crucial, then getting a brain approximately right but
with the fine details a bit fuzzy would be about as valuable as
uploading two fistfuls of porridge.

I'm sure there will be some who will argue against this, but please
let's conduct that argument in a separate thread (this one), because I
think it is a distraction.

Richard Loosemore

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:52 MDT