From Osher Doctorow [EMAIL PROTECTED], Sat. Nov. 30, 2002 1005
I agree generally with Tim May on mathematics and physics vs computers and
AI. My most amusing example is something of a Jonathon Swift parody of all
four of these fields. Gulliver lands on an island inhabited by
mathematicians, physicists, computer scientists/computer engineers, and AI
people, all competing. He notices that they all rushing ahead to greater
and greater complication and complexity and so on, and it occurs to him that
this might be their weak point. Could they all have overlooked something
simple [question-mark]. He discovers, as it so happens that I discovered
some 20 or so years ago, that they are all using division and multiplication
to formulate relationships involving influence and causation [I omit
calculus limits for those unfamiliar with them], and minimally using
subtraction and addition. Gulliver then reformulates all of their theories
using subtraction and/or addition, and it turns out that all of the
resulting theories are completely different from the old ones. Not one
person among all the island's so-called geniuses had come up with the very
tiny idea of changing from division/multiplication to subtraction/addition
in all of their work. Upon presenting this fact to the gathered people of
the island, the people debate for a long time, and then decide that Gulliver
knows more, so they decide to drop their entire four fields and start all
over again much more slowly, this time not racing to greater complication
before analyzing the simple concepts that they are using. The name Socrates
is mentioned as an example. The predictions of several popular science
writers on the island such as Professor Kaku to the effect that computers
are going to almost literally swamp everything else are accordingly
considerably modified to say the least.
Osher Doctorow
- Original Message -
From: Tim May [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Saturday, November 30, 2002 2:37 PM
Subject: Alien science
On Saturday, November 30, 2002, at 01:32 PM, Ben Goertzel wrote:
...
I think this is certainly a plausible prediction of the future, but I
see it
as an unlikely one.
I think that intelligent software programs will be brought into
existence
within the next 10-50 years, and that among other effects, this will
cause a
physics revolution. Furthermore, it will be a revolution in a
direction now
wholly unanticipated.
It will be interesting and exciting if you are right, but I think the
kind of AI you describe above and below is further off than 10-30
years, though perhaps not 50 years.
Right now we analyze data about the microworld in a very crude way.
For
example, we scan Fermilab data for events -- but what about all the
other
data that isn't events but contains meaningful patterns?
Create an AI mind whose sensors and actuators are quantum level, and
allow
it to form its own hypotheses, ideas, concepts, ontologies Do you
really think it's going to come up with anything as awkward and
overcomplex
as our current physics theories?
I have no idea. True, it may come up with all sorts of weird theories.
But, absent new experimental evidence, will these new theories actually
tell us anything new?
Your point about AIs exploring physics is an interesting one. And you
are right that Egan has his AIs, his uploaded Orlandos and even his
computer-produced Yatima, looking very much like humans. Not at all
like the Entities of Vinge's Deepness, Zindell's Neverness, or
Stephenson's Hyperion series. But let us imagine that an advanced AI
were to be turned loose on a Newtonian world. I can well imagine that
such an entity, left to its own devices, might come up with weird names
for inertia, mass, friction, etc. Perhaps even synthetic combinations
of what we take to be the basic vectors of classical mechanics. Instead
of 3-space being so primal, phase spaces of 6, 18, and even many more
dimensions would perhaps be more natural to such a mind. (Needless to
say, given that today's best AI programs and computers are having a
very hard time even doing naive physics, a la ThingLab and its
descendants, I'm not expecting progress very quickly. And ThingLab is
more than 20 years old now, so expecting massive breakthroughs in the
next 10-20 years seems overly optimistic.)
More importantly, would an AI version of classical physics, complete
with incomprehensible (to us) phase spaces and n-categories and so on,
including constructs with no known analogs in our current universe of
discourse, would this version give any predictions which differ from
our own? In short, would the AI's version of physics give us any new
physics?
My hunch is no. It might be better at solving some problems, just as
the mental architecture of birds may give them much better abilities to
solve certain kinds of 3D problems than we have had to evolve, and so
