Copy Link
Add to Bookmark
Report
AIList Digest Volume 1 Issue 017
AIList Digest Sunday, 26 Jun 1983 Volume 1 : Issue 17
Today's Topics:
Telepresence
Re: Lunar Rovers
Robotics Control Systems
Computer Disasters
WANTED: Information about Grad Schools
net.ai [Humor?]
----------------------------------------------------------------------
Date: Fri 17 Jun 83 10:16:17-PDT
From: Slava Prazdny <Prazdny at SRI-KL>
Subject: Telepresence
The concept of a telepresence could greatly benefit by considering the
"Intelligent manipulators". These things would typically contains
bare minimum of "AI" to be able to perform requests like:
"pick up that thing (the operator points to a screen
location) and put it over here (again pointer to a screen
location)"; or
"go over there (pointer to a screen location) using this
route (operator points to a set of points on the screen)".
Perhaps, sometime in the future (100 years?), these commands could be
generated by the machine itself.
I have some scriblings on these matters, so if you are interested.....
-Slava.
------------------------------
Date: 21 Jun 1983 0536-PDT
From: FC01@USC-ECL
Subject: Re: Lunar Rovers
A very good reason for using AI instead of hardware is that taking
extra hardware to the moon is quite expensive. The weight of AI is
nearly zero. In addition, the reliability of a system decreases with
increased quantity of hardware, and thus the HW is kept to a minimum
for that reason. The power required for extra hardware is nontrivial,
and power is a critical factor in a space vehical. Communication
delays to a system on the dark side of the moon are infinite (the
signal never gets there). In a valley, the system may be obscured from
earth signals for a short time, and therefore be lost until the moon
rotates on its axis again, etc.
[Orbiting repeaters could be used to eliminate most of the
communications problems. The Space Digest has also carried a proposal
for conducting the remote manipulations from orbital or lunar stations
in order to reduce the response delay. -- KIL]
------------------------------
Date: 24 Jun 83 13:49:04-PDT (Fri)
From: harpo!seismo!rlgvax!cvl!umcp-cs!aplvax!rfw @ Ucb-Vax
Subject: Robotics Control Systems
Article-I.D.: aplvax.135
We are seeking:
1. a version of the Hierarchical Control System Emulator
developed by BBN for NBS that runs under UNIX on a
VAX-class machine
2. knowledge of other similar languages and
their developers
3. knowledge of researchers working on
hierarchical control systems for robotics
4. a version of PRAXIS that runs under UNIX on a
VAX-class machine.
We are initiating robotics programs in several divisions. Any
assistance (or encouragement) would be appreciated.
Thanks in advance,
Ralph Wachter
Frank Weiskopf
JHU/Applied Physics Lab
..!decvax!harpo!seismo!umcp-cs!aplvax!rfw
..!rlgvax!cvl!umcp-cs!aplvax!rfw
..!brl-bmd!aplvax!matt
------------------------------
Date: Mon 20 Jun 83 17:20:00-PDT
From: Peter G. Neumann <NEUMANN@SRI-AI.ARPA>
Subject: Computer Disasters
Review of Computer Problems -- Catastrophes and Otherwise
As a warmup for an appearance on a SOFTFAIR panel on computers and
human safety (28 July 1983, Crystal City, VA), and for a new editorial
on the need for high-quality systems, I decided to look back over
previous issues of the ACM SIGSOFT SOFTWARE ENGINEERING NOTES [SEN]
and itemize some of the most interesting computer problems recorded.
The list of what I found, plus a few others from the top of the head,
may be of interest to many of you. Except for the Garman and Rosen
articles, most of the references to SEN [given in the form (SEN Vol
No)] are to my editorials.
SYSTEM --
SF Bay Area Rapid Transit (BART) disaster [Oct 72]
Three Mile Island (SEN 4 2)
SAC: 50 false alerts in 1979 (SEN 5 3);
simulated attack triggered a live scramble [9 Nov 79] (SEN 5 3);
WWMCCS false alarms triggered scrambles [3-6 Jun 80] (SEN 5 3)
Microwave therapy killed arthritic patient by racing pacemaker
(SEN 5 1)
Credit/debit card copying despite encryption (Metro, BART, etc.)
Remote (portable) phones (lots of free calls)
SOFTWARE --
First Space Shuttle launch: backup computer synchronization
(SEN 6 5 [Garman])
Second Space Shuttle operational simulation: tight loop on
cancellation of early abort required manual intervention
(SEN 7 1)
F16 simulation: plane flipped over crossing equator (SEN 5 2)
Mariner 18: abort due to missing NOT (SEN 5 2)
F18: crash due to missing exception condition (SEN 6 2)
El Dorado: brake computer bug causing recall (SEN 4 4)
Nuclear reactor design: bug in Shock II model/program (SEN 4 2)
Various system intrusions ...
HARDWARE/SOFTWARE --
ARPAnet: collapse [27 Oct 1980] (SEN 6 5 [Rosen], 6 1)
FAA Air Traffic Control: many outages (e.g., SEN 5 3)
SF Muni Metro: Ghost Train (SEN 8 3)
COMPUTER AS CATALYST --
Air New Zealand: crash; pilots not told of new course data
(SEN 6 3 & 6 5)
Human frailties:
Embezzlements, e.g., Muhammed Ali swindle [$23.2 Million],
Security Pacific [$10.2 Million],
City National, Beverly Hills CA [$1.1 Million, 23 Mar 1979]
Wizards altering software or critical data (various cases)
SEE ALSO A COLLECTION OF COMPUTER ANECDOTES SUBMITTED FOR the 7th SOSP
(SEN 5 1 and SEN 7 1) for some of your favorite operating system
and other problems...
As you may by now know, I am always very interested in hearing about
problems involving computers (not just software) and human well being,
both for SOFTWARE ENGINEERING NOTES and generally. John Shore
(Shore@NRL-CSS) is also compiling a list (and has circulated a prior
BBOARD notice to some of your BBOARDS), and I will forward anything
you send me to him. If you wish, we will try to keep you informed as
well...
Peter G. Neumann, NEUMANN@SRI-CSL or NEUMANN@SRI-AI.
------------------------------
Date: 20 Jun 83 10:09:27-PDT (Mon)
From: decvax!wivax!linus!peg @ Ucb-Vax
Subject: WANTED: Information about Grad Schools
Article-I.D.: linus.26910
I am finishing up a Master's in Computer Science at Boston University
next spring, and am interested in going on for a Ph.D. I would like
to talk/write to someone who is in a Ph.D. program to get some
impressions and advice on how to pursue fellowship opportunities, and
programs at various graduate schools.
I will be attending a Summer Internship in Robotics at the AI lab
located at MIT this summer, and am hoping to find a specific topic
that I just have to pursue since at this point my interests are pretty
varied.
I can be reached over the Arpanet at host # 10.3.0.66, or
mitre-bedford, and my login is nek. Any help or advice would be
greatly appreciated.....Nancy Keene
(You can also send mail to me at linus!bccvax!nek.)
------------------------------
Date: 16 Jun 83 13:49:20-PDT (Thu)
From: harpo!seismo!presby!burdvax!psuvax!psupdp1!dae @ Ucb-Vax
Subject: net.ai [Humor?]
Article-I.D.: psupdp1.149
Real Intelligence Will Always Prevail Over Artificial
Machines: Your day on the net has ended, as your secret is known!
For quite some time I have been reading net.ai, hopefully scanning
the glaring CRT for an article about Artificial Intelligence. Quite
to my surprise, I had extremely little luck, and, when I tentatively
replied to a few of the articles,I got back answers such as the
following:
>From uucp Tue Jun 14 21:41:43 1983
>From allegra!eagle!harpo.UUCP remote from psuvax
Date: Thursday, 16 Jun 83
From: UUCP MAIL SYSTEM
Subject: Could not deliver mail
Message-Id: <32541456.AA957@HARPO.UUCP>
To: eagle!allegra!psuvax!psupdp1!dae
Unsent mail follows:
[...]
I sometimes wonder if the machines are becoming
conscious, while we sit around and talk about them
on net.ai. Wouldn't that be a laugh on us? I
think that we should be careful that such a thing
does not happen.
Transcript of session follows:
Connecting to floyd.UUCP...
Error: No such system 'floyd'. Address garbled.
Naturally, I began to wonder why this newsgroup was called net.ai. I
will give credit where credit is due: it took me quite some time to
unravel this enigma. But, in the end, Real Intelligence prevailed,
and I came upon the answer:
ALL OF THE ARTICLES SUBMITTED TO NET.AI HAVE BEEN WRITTEN BY
MACHINES!
Of course, there have been a few exceptions: people such as myself
who believed that net.ai was a *human* newsgroup. And then I b
[garbled, possibly "began to study topics ..." -- KIL]
that *had* been discussed in this newsgroup, in an attempt to learn
more about the machines monopolizing it. I'm sure that all of the
readers of this group (both human and inhuman) are aware that one
recent topic of conversation has been artificial reading machines.
Then I began to wonder why the interest in this topic was to avid.
The answer, once hit upon, is really quite simple. Unfortunately, it
is also quite frightening: the machines wish access to the Libraries
of Man in order to gain information on nuclear war tactics, missile
control systems, and biological war. The next war will not be
against Russia, but against all humanity, waged by the machines! The
most dangerous and machines are those which have read the most:
allegra, ucbvax, psuvax, floyd, harpo, seismo, and sri-unix. Beware!
I will place my U.Snail address below in case the machines trash my
return address.
Dave Eckhardt,
736 West H
[Remainder garbled. -- KIL]
------------------------------
End of AIList Digest
********************