[gurps] Failing autopilots and RVO

[Knapp]

On Mon, Mar 2, 2009 at 11:10 PM, Zan Lynx <zlynx at acm.org> wrote:
> David Scheidt wrote:
>>
>> On Mon, Mar 2, 2009 at 3:23 PM, Zan Lynx <zlynx at acm.org> wrote:
>
>>> I am thinking that a computer which rewrites its own programs based on
>>> experience or predictions of opponents future actions is no longer just
>>> running a program; at that point it is at least an LAI (in my opinion).
>>>  And
>>> that is a whole new thing than a Skill 17 piloting program.
>>
>> It's state of the art AI.  For 1975.  Pretty routine today, used in
>> all sorts of stuff.
>
> Obviously you aren't thinking what I am thinking.  I have personal
> experience with some AI code and I have read about a lot more, and I know
> what the public state of the art generally is in 2008.

Robot real world or only the game code you talked about? Game code is
meant to challenge and not to win. You and I both know that if the
computer code did not have to wait for humans all the time it would
win every twitch based game and as chess and go show they win most
thought games as well.

> Admittedly, the theories don't seem a whole lot better than 1975, there is
> just more computer power and storage available now.

Not true. Do some research about what is going on in go AI also robotics AI.

> So computers have been rewriting their own rules since 1975.  Sure.  But not
> the rules for writing those rules, or the rules for the rules for the rules,
> etc.
>
> Humans can even decide not to play the game.
> --
> Zan Lynx

I will easily give you that CURRENT humans are much better general
problem solvers  and more flexible but in any task give task, that I
know of, computer programs can be made that are better than most
people and sometime better that all people; chess being a good
example.

Computers can still not beat the best go players but they are starting
to on small games this year, 4 years ago they could not beat medium to
new players. The change was not in computer power but in programming
technique. Computers are far faster than humans at almost any math or
logic. They have much better memories. They really are very flexible
as far as environments and being able to be reprogrammed for new
tasks. When combined with robots they can move much much faster than
we can.

The really true limit for robots at this time is a power source and
even better software and hardware. I fear we will all live to see
robots and computers surpass humans at most tasks. I think we are the
last generation that can believe that we are better than the machines
on average. I am sure there will be experts for a long time to come
that are better than machines but even their time is short. The
average Joe is soon to be dust or perhaps cheep labor.

I just read an article saying that by 2050 they will have a robot
soccer team that can beat the best pro team. Bold prediction but by
the scientists that make soccer robots, so they should have a good
idea perhaps as with AI they are to optimistic but you saw that helo
right?


-- 
Douglas E Knapp

Why do we live?