[tt] [NSG] Meeting Announcement 11/20

Mon Nov 19 21:38:34 UTC 2007

----- Forwarded message from Fred Hapgood <hapgood at pobox.com> -----

From: Fred Hapgood <hapgood at pobox.com>
Date: Mon, 19 Nov 2007 12:06:16 -0500
To: nsg at marshome.org
Subject: [NSG] Meeting Announcement 11/20
X-Mailer: MessagingEngine.com Webmail Interface
Reply-To: Nanotechnology Study Group <nsg at marshome.org>

Meeting notice: The 071120 meeting will be held at 7:30 P.M. at the
Royal East (782 Main St., Cambridge), a block down from the corner of
Main St. and Mass Ave.  If you're new and can't recognize us, ask the
manager. He'll probably know where we are. More details below.

Suggested topic:  General learning in Virtual Worlds.

(I wrote a version of this for the last meeting and suspect I forgot
to send it.)

Last Tuesday Dan Kara, an analyst specializing in the robot industry,
gave a talk at MIT on the topic of "Global Trends in the Personal,
Service and Mobile Robotics Market."

I didn't see anyone else so let me summarize: there is a ton of money
flowing into the field.  The biggest investors are the militaries.
Robotics is the second largest procurement program in the US DOD (after
the Strike fighter), but our military is far from the only one
interested.  There is in particular a lot of talk about automating the
logistics chain.  (Many militaries seem to have decided that the classic
ratio of logistical to fighting personnel has something suboptimal about
it.)  Health care is second on a dollar basis; entertainment, defined
broadly (to include education), is third.  (Kara stayed conspicuously
silent on dildonics, which history suggests will not be a small market.)
Korea and Japan consider robotics strategic industries and are funding
the most ridiculous ideas lavishly.

And all this without any real progress on the general learning problem.
You have to wonder: if there is this much interest in the field now,
what will happen when the first advance, no matter how crude -- say to
the level of a pigeon -- is made in the ability to deal with novel
landscapes?  Most of the obvious peripherals for robots -- manipulators,
vision, language recognition -- would all benefit enormously from even
weak general learning.  For many apps general learning is really the
only piece missing to make the concept economically practical.

One approach to the general learning problem posits that there is A
Right Idea out there, an algorithm that can be found, or figured out,
and executed.

But there is another perspective: that the missing ingredient is not a
single concept but the the right kind of embedding into the right
environment. The idea here (which so far as I know originated with
Rodney Brooks) is that the general learning problem is far more
tractable in environments that come with lots of feedback and a rich
texture of "teachable moments".

Some people think that virtual worlds might be such a place.  VWs are
fairly simple (compared with the physical world) and yet varied enough
to require some sort of learning.  They are almost completely knowable
(you don't need to measure anything; you just ask the simulator). They
are full of real people eager to help teach machines how to act smarter,
and the common protocols mean that all devices in virtual worlds can
learn together (and compete against each other). What one agent learns,
they all learn.

Finally, and perhaps most important, the demand within virtual worlds
for general learning machines is bottomless.  The first virtual world to
come with creatures that learn, creatures with user-adjustable
smartness, is going to have a major competitive advantage and everyone
knows it.  In such a world the wildlife would become steadily better at
eluding you when you tried to hunt it, virtual warriors would set ever-
cleverer ambushes, storekeepers would play to your tastes and interests,
and so on.

General learning is being targeted and targeted explicitly by a number
of companies interested in this space. Novamente and Electric Sheep
have announced their intention of having a trainable virtual pet for
sale by next year.  That pet will learn by observing avatars -- if you
want to train your dog to fetch you will have to show him how to do it
-- but Ben Goertzel of Novamente has said that the second generation of
this product will come with a linguistic engine, yielding a virtual
parrot that can be trained with speech. DadenLimited, a virtual worlds
consultancy, is building trainable chatbots, entities that learn to
interact with virtual world players via texting or speech. (Chatbots
are used to perform a range of inworld services, such as virtual
floor sales.)

While general learning entities have not been released into virtual
worlds yet, the companies could not be focussed more tightly on the
concept. The least we can say is that the embedded worlds theory does
not lack for testing.

It is a matter of speculation how generalizable general learning
developed in virtual worlds will be, but surely if it did radiate across
all those other tasks noone would be surprised.  The whole point of
General Learning is context independence.

<+><+><+><+><+><+><+><+><+><+><+><+><+><+>

Coolest press release of the week:

http://www.news.uiuc.edu/news/07/1109photosynthesis.html

<+><+><+><+><+><+><+><+><+><+><+><+><+><+>

In twenty years half the population of Europe will have visited the
moon.

                                -- Jules Verne, 1865

<+><+><+><+><+><+><+><+><+><+><+><+><+>

Announcement Archive: http://www.pobox.com/~fhapgood/nsgpage.html.

<+><+><+><+><+><+><+><+><+><+><+><+><+>

Legend:

"NSG" expands to Nanotechnology Study Group.  The Group meets on the
first and third Tuesdays of each month at the above address, which
refers to a restaurant located in Cambridge, Massachusetts.

The NSG mailing list carries announcements of these meetings and little
else. If you wish to subscribe to this list (perhaps having received a
sample via a forward) send the string 'subscribe nsg'  to
majordomo at polymathy.org.  Unsubs follow the same model.

Comments, petitions, and suggestions re list management to:
nsg at pobox.com. www.pobox.com/~fhapgood   www.pobox.com/~fhapgood

http://www.BostonScienceAndEngineeringLectures.com
http://www.pobox.com/~fhapgood

_______________________________________
Nanotechnology Study Group NSG discussion group
http://www.marshome.org/mailman/listinfo/nsg
To Reply to Everyone: NSG___no-spam at marshome.org
Questions for list admin: NSG-owner___no-spam at marshome.org
Archive: http://MarsHome.org/mailman/private/NSG
Unsubscribe: NSG-unsubscribe at marshome.org
Password or Options or Unsubscribe: http://MarsHome.org/mailman/options/NSG
More info and other e-lists:  http://MarsHome.org/about/discussion_groups.html
Hosted by CyberTeams.com and Mars Foundation(tm), http://MarsHome.org

----- End forwarded message -----
-- 
Eugen* Leitl <a href="http://leitl.org">leitl</a> http://leitl.org
______________________________________________________________
ICBM: 48.07100, 11.36820 http://www.ativel.com http://postbiota.org
8B29F6BE: 099D 78BA 2FD3 B014 B08A  7779 75B0 2443 8B29 F6BE