On Fri, Jan 4, 2013 at 12:28 PM, Logan Streondj <[email protected]> wrote:

> I agree with Aaron,
> keep it simple to the point,
> or no one will bother reading it.
>
>

OK.  I think people who think that 'prediction' is at the center of
their ideas about AGI should try using it in real life just to see
how it works.  For instance they could use their 'predictions' about their
own projects as a way to get a better grasp on reality.

Using my own work as an example I will say:
I think I can get a limited AGI program (what I call AGi) working in 1
year.  Now, how can I use this prediction as a better way to understand
what is really happening.
*If I can't get it working within a year or two (we need to be given some
leeway) then that will be a good indication that I do not have it all
figured out.
*Suppose that after 5 months I haven't started working on the AI part.  My
prediction does not tell me anything about that.  However, I can create
auxiliary theories that predict that possible outcome from other insights
that I have.  For example, if I knew exactly what I wanted to do and I knew
that it was going to work I would be very eager to get going on the
project.  So if 5 months goes past and I haven't gotten the preparatory
work done (and nothing life-changing happened to me that would interfere
with my working on the project) this shows that I did not have all the
answers and I did not have much confidence that my ideas would work.

So by generating auxiliary theories that could be used to critically
predict that the prediction of my central theory would not come true, I can
actually use my predictions in a way that would give me a better grasp of
what is really going on.

For the most part I would say that using prediction in real life shows that
it helps to seed the crystallization of a lot of other explanatory
theories.  That is a contribution but that can be done in a lot of other
ways.  And when we expect a sub-program to work some way and it does so it
does stand as confirmatory evidence.  If the sub-program does not work the
way we expected it to it shows that we need to work on the programming some
more.  However, it does not generate the next level of creative problem
solving. It does not show us, for example, where the malfunction in our
programming might exist.  Even though 'prediction' would be used as a part
of that process, it is almost as an after thought of creative problem
solving.

So although I am skeptical of the exaggeration of the role of
'prediction' in intelligence, I am able to find some novel ways to use it
to make a better analysis of reality.  Like other AGI methods, it works
really well in a few situations but it a key feature of intelligence only
in that it is implicit in any application of understanding.

Jim Bromer



-------------------------------------------
AGI
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657
Powered by Listbox: http://www.listbox.com

Reply via email to