A.I. musings
[1/5] from: carl:rebol at: 9-Jul-2001 13:13
Interesting...
You're uncovering my true intention for creating REBOL... You see,
it was neurons that got me into computers in the first place.
Unfortunately 4K on a PDP-8 didn't cut it for simulating reciprocal
synapses.
I felt I needed a tool that could help me get back to ganglia...
ultimate parallelism.
-Carl
At 7/9/01 04:49 AM -0700, you wrote:
[2/5] from: agem:crosswinds at: 9-Jul-2001 23:12
RE: [REBOL] Re: A.I. musings
[carl--rebol--com] wrote:
> Interesting...
>
Interesting..
i didn't see a demo, sure you have some neuron-simulator in rebol
somewhere?!
i see here a chance to finally understand how that stuff works :-)
(some input-fields, some lines for connections,
and a great explanation. hey!)
> You're uncovering my true intention for creating REBOL... You see,
> it was neurons that got me into computers in the first place.
> Unfortunately 4K on a PDP-8 didn't cut it for simulating reciprocal
> synapses.
>
> I felt I needed a tool that could help me get back to ganglia...
> ultimate parallelism.
>
so you thought if our neurons can be programmed in english you need
something similar to program computers? ;-)
> -Carl
-Volker
[3/5] from: kenneth::nwinet::com at: 9-Jul-2001 17:20
Had trouble with 256 bytes (no K) on my Cosmac Elf too... :-) I'm jealous
you had a PDP!
But seriously, to achieve parallelism, would you implement a message
blocking scheme something like is done by QNX?
Also, if I were to run multiple instances of REBOL on a single machine,
what's the simplest way to pass messages between the instances? I could of
course use a file as an intermediary, but it seemed there should be a better
way.
Thanks,
Ken.
> You're uncovering my true intention for creating REBOL... You see,
> it was neurons that got me into computers in the first place.
<<quoted lines omitted: 3>>
> ultimate parallelism.
> -Carl
ps: 500khz RCA 1802... sweet thang, those were the days...
[4/5] from: kenneth:nwinet at: 8-Jul-2001 22:35
Hi all,
I was musing on the use of Rebol to write a network seed A.I. when I came
across this link...
http://sysopmind.com/singularity.html
<Beware there's a big ordinal in the middle> ;-)
or the short version...
"You know, I don't understand why humans evolved as such thoughtless,
shortsighted creatures."
"Well, it can't stay that way forever."
"You think we'll get smarter?"
"That's one of the two possibilities."
- Calvin and Hobbes
At first blush I'd say the guy is having a bad trip..., but if you can
suspend disbelief it did have some food for thought.
For 28 years (which I know because I vividly remember the day at the age of
14 the question was put to me) I've felt that we are underestimating the
difficulties associated with A.I. (not meaning expert systems or chess
players but true A.I. that could pass the turing test.) My scoutmaster
(yeah, I was) who happened to be an airforce psychologist was discussing the
topic with an airman that felt emphatically that computer intelligence was
less than ten years away (1973 - 1983 uh huh.;-) I didn't know it was
called the mind-body problem at the time, but that was the issue that cause
me to say I didn't expect to see an intelligent computer. Since then I've
read "The Emperor's New Mind" - Roger Penrose, who says it better than I
ever could. However, that doesn't mean I'm not interested in the research
that could lead to byproducts (like expert systems) even if it never arrives
at the ultimate goal.
Anyway, is anyone out there using rebol in A.I. research?
Ken.
...or perhaps seen two many Terminator movies...
...or read "The adolescence of P1" -- which I'm wondering how silly this
book sounds since I haven't read it in twenty years (considering how amusing
a 5 year old computer magazine seems today.)
final note (has anyone noticed that IBM and Intel may both have broken
Moore's law recently along two independant lines of development.)
[5/5] from: kenneth:nwinet at: 9-Jul-2001 4:49
So here's what I was thinking might be a killer ap for Rebol...
A search engine that uses an adaptive distributed neural net to process
queries. I imagine it being written to provide a limit of say 5 hits and
the user would rank the hits as a feedback mechanism to the neurons.
Imagine how fast it might adapt with millions of people using it?
Frankly, I haven't a clue as to how to design the thing, but I find the idea
intriguing.
Ken.
Notes
- Quoted lines have been omitted from some messages.
View the message alone to see the lines that have been omitted