Pleasant@Rutgers (11/28/82)
HUMAN-NETS Digest Saturday, 27 Nov 1982 Volume 5 : Issue 107
Today's Topics:
Query - English Interface,
Programming - Unix,
Computers and People - Cable TV and the First Amendment (3 msgs),
Technology - Supercomputer Project
----------------------------------------------------------------------
Date: 25 November 1982 04:02-EST
From: Robert Elton Maas <REM at MIT-MC>
Subject: English interface
A few months ago I sent out a query for parsers for English. Now I
have a slightly different query. Has anybody developed a unified
parser and un-parser (generator) system that runs from a single
grammar instead of a separate grammar in each direction?
------------------------------
Date: 24 Nov 1982 at 0912-PST
Subject: The trouble with Unix
From: zaumen at SRI-TSC
... or how the TAB key deleted all my files (in one directory).
Some time ago there was some discussion about the Unix
user-interface. Here is an example of what can happen, even if you
are reasonable proficient with Unix. The example given below is not
typical of my use of this operating system, but...
Recently, I tried to type the following sequence of Unix shell
commands:
rm ^U for i in *
do
echo $i
cp $i $i.s
done
^U (control U) is what my tty driver uses to cancel the line, so the
first line in this command should be just "for i in *". What, one
might ask, does the TAB key have to do with this? Well, the tab key
is also ^I. My typing is a bit klutzy, and I hit ^I instead of ^U:
these keys are next to each other. I also wasn't paying attention
as well as I should have been. The Shell interpreted what I
actually typed as
rm for i in *
and responded with
rm: for not found
rm: i not found
rm: in not found
Alas, * expands to all files in the current directory, and rm
removed all of them.
Moral of the story: EVERYONE makes mistakes. Anyone reading a Unix
command that starts with
rm for i in *
would guess that there was a typo, and command parsers, shells, etc.
should have a syntax that allows such errors to be detected before
the commands are executed.
__________
/ \
| - - |
(| /\ |)
| |
\ /\/\/\ / <==== ( arghhhhhhhh )
\ /
------
Bill
------------------------------
Date: 23 Nov 82 11:52:32 EST (Tue)
From: Mark Weiser <mark.umcp-cs@UDel-Relay>
Subject: Children
Being a parent is a soul-shattering experience which changes anyone
who tries it. I do not say it is bad (I happen to think it is
wonderful)--but one cannot stay the same afterwards. I always read
comments by people without children about child-raising with
amusement. If they only knew what they would really be like as
parents... It is like a non-programmer talking trying to talk to a
hacker about computers.
(~= flame on:)
Children are our investment in the future. If everything else was
lost, but we still had children, there would be some hope. But with
all the luxuries in the world, life would be pointless with no
children to continue the mysterious journey of intelligence through
the universe. There is NOTHING more important than the care and
feeding of children (which doesn't mean that everyone should do it,
any more than we all depend on farmers but not everyone must
farm--but only people with children should be allowed to vote).
(.= flame off.)
------------------------------
Date: 25 Nov 82 2:30:01-EDT (Thu)
From: Randall Gellens <randall.CC@UDel-Relay>
Subject: Re: HUMAN-NETS Digest V5 #106
Regarding your comments [HN V5 #106] about children and tv in
general, and censoring TV for kids in particular...Wow! I couldn't
agree more or have put it better.
(Of course, the fact that don't have kids and see very little TV
[Sitcom: "Facts of Life?" Wazzat?] might have something to do with
it.)
--randall
------------------------------
Date: 26 November 1982 06:49-EST
From: Jerry E. Pournelle <POURNE at MIT-MC>
Subject: Re: TV and censorship
Think of it as evolution in action.
------------------------------
Date: Thursday, 25 Nov 1982 19:17-PST
Subject: news clipping: "Supercomputer" project
From: lauren at RAND-UNIX
n043 1152 24 Nov 82
BC-SUPERCOMPUTER
(Newhouse 002)
BUILDING THE MIND OF MAN INTO A MACHINE
By PATRICK YOUNG
Newhouse News Service
BOSTON - Stephen Grossberg envisions nothing less than building
the mind of man into a machine.
He is not alone in dreaming of creating super-smart computers
with humanlike powers of thought and reason. But Grossberg is
approaching the task from a different prospective than most others
who labor in the high-tech world of artificial intelligence.
In nearly 25 years of research into the intricate physical
workings of the mind, Grossberg has evolved a series of mathematical
equations to explain brain activity. These equations, he believes,
can be used to write programs that will enable computers to think.
The eventual results, he hopes, will include robots with the
impressive powers of R2-D2 and C-3PO, the robotic duo of ''Star
Wars'' fame.
''An intelligent robot must be able to adjust it parameters and
adapt,'' says Grossberg, professor of mathematics, psychology and
biomedical engineering at Boston University. ''That means making the
computer more flexible so it can reprogram itself.''
Robots run by such advanced computers could carry out tasks too
dirty or dangerous for humans.
''These jobs might include working in poisonous atmospheres,
inside nuclear reactors or mining precious metals from the Asteroid
Belt between Mars and Jupiter,'' Grossberg says. ''The first
intelligent creature from Earth to explore another planet might not
be human at all. It might be one of our electronic progeny.''
But first there is the matter of money to develop the computer,
which Grossberg is trying to raise, and the Japanese.
Backed by funds from their government, two Japanese research
teams have embarked on a 10-year effort to build a computer with
''man-level intelligence'' that will be able to learn, think, read,
write and speak.
There is no comparable effort in the United States, although a
number of researchers in universities and industry continue to
pursue the goal of creating a computer with artificial intelligence.
The Japanese have said such an machine will require new computer
mathematics. And their description sounds to Grossberg suspiciously
like the type of equations he has developed and refined with the
aide of colleagues at Boston University's Center for Adaptive
Systems.
Today's computers, though capable of lightning-fast calculations
and storing vast amounts of information, are stupid. They don't
think; they only follow the instructions programmed into them.
This is true even of the most sophisticated units, the so-called
expert systems.
By drawing on certain scientific principles and the collected
experiences and learning of experts stored in their memories, expert
systems can do such things as diagnose medical problems and predict
the location of mineral deposits. But they don't learn from
experience and they don't reason in a human sense.
Grossberg's goal is to build a computer that will monitor its
environment and change its behavior to adapt to changing conditions.
He didn't start out to create to artificial intelligence or
build a super-smart computer.
His work has concentrated on understanding the interaction of
mind and matter - how the physical brain, through the chemical and
electrical activity of its nerve cells, produces behavior.
''Our work is to try to understand and predict neural events, to
give a unified view of behavior and its underlying neural action,''
Grossberg says.
Memory, learning, thinking and the brain's other activities
depend on the complex interactions of a series of nerve cells. A
single nerve cell can be studied in detail to determine its
chemical, electrical and physiological changes.
But ''in terms of understanding behavior, the functional level
is not the single cell,'' Grossberg says.
Yet determining how the brain integrates the actions of
individual cells to produce a specific behavior has proved more
difficult.
''A pattern of activity across cells forms a pattern of
information,'' Grossberg says. ''The question is how does a whole
field of cells compute things that no single cell will ever know.''
What he has found, he says, ''is a few different principles that
occur over and over.'' And these can be reduced to mathematical
equations.
''That mathematical model is the bridge to going to a new
machine,'' Grossberg says. ''It's not just that you've got a wiring
diagram - the architecture of the brain - it's really the dynamics
of the system, the pharmacology and physiology.
''The question is what is the best and cheapest way to implant
this in hardware. It is a major problem that we haven't dealt with
yet.''
That effort - and perhaps a competitive race with the Japanese -
will come if Grossberg secures funds to develop his super-smart
computer.
------------------------------
End of HUMAN-NETS Digest
************************