Copy Link
Add to Bookmark
Report
AIList Digest Volume 3 Issue 176
AIList Digest Friday, 22 Nov 1985 Volume 3 : Issue 176
Today's Topics:
Queries - Ordering ICOT Reports & MTSLISP on VM/CMS &
Expert System Shells for IBM PC & Knowledge Ownership,
Intelligence - Man/Machine IQ Tests,
Programming Languages - Hard Typing,
Literature - Spang Robinson Report,
Expert Systems - COBOL Restructuring
----------------------------------------------------------------------
Date: Fri, 22 Nov 85 12:52:35 EST
From: Frank Ritter <ritter@BBN-LABS-B.ARPA>
Subject: Ordering ICOT Reports
How does one order ICOT tech reports?
[I'm haven't had time to really go though the list of reports that I
had copied, but I'm sure many will want the address to order reports.]
Thanks for posting the list,
Frank
------------------------------
Date: Wed, 20 Nov 1985 18:42 PLT
From: George Cross <FACCROSS%WSUVM1.BITNET@WISCVM.ARPA>
Subject: MTSLISP on VM/CMS
Hi,
Does anyone have MTSLISP running on VM/CMS?
---- George
George R. Cross cross@wsu.CSNET
Computer Science Department cross%wsu@csnet-relay.ARPA
Washington State University faccross@wsuvm1.BITNET
Pullman, WA 99164-1210 (509)-335-6319/6636
------------------------------
Date: 20 Nov 85 15:03 PST
From: kwhite.pasa@Xerox.ARPA
Reply-to: kwhite.pasa@Xerox.ARPA
Subject: Expert System Shells for IBM PC
Does anyone out there have some experience on the IBM PC (or its
compatible clones) using Expert System Shells? I have been giving
consideration to KDS, EXSYS, and the Personal Consultant. I would like
to have the system be capable of running DOS commands which could invoke
other programs (in another language) and return. I am also particularly
interested in shells that support the user when s/he is lost and says "I
don't know" or "I don't understand". The system should be capable of
making suggestions to the user in this situation.
Thanks in advance for your help.
...Kendall...
------------------------------
Date: Wed, 20 Nov 85 12:02:13 est
From: mayerk%UPenn-GradEd%upenn.csnet@CSNET-RELAY.ARPA
Subject: Issues Concerning Expert Systems -- Who Owns What
This article was clipped from the Wall Street Journal,
Thursday, November 14, 1985.
* * *
AFTER 15 YEARS of medical training in Pittsburgh, Caduceus will soon
be ready for its first patients.
Caduceus is a computer program that tries to copy the skill of
72-year-old internist Jack Myers and other physicians at diagnosing about
600 illnesses. The computer will never be as good as the best doctor
because it lacks an imagination, Dr. Myers says. Diseases that haven't
been described or categorized will stump Caduceus, he explains, whereas
"a brilliant human being will realize he's facing something new." Unlike
a human being, however, Caduceus never forgets to order a test or overlook a
symptom, so doctors could use the program to check their diagonoses.
University of Pittsburgh computer engineer Harry Pople says the
long time spent developing Caduceus is partly a result of the system's
complexity: The software had to be designed to rapidly diagnose several
illnesses at once. But the system had some human problems, too.
An early version, call Internist I, was rather authoritarian,
making diagnoses without telling doctors how it had arrived at its
conclusions. That sort of bedside manner invariably stepped on fragile
medical egos.
Caduceus, named for the coiled-snake medical emblem, now reasons
along with a doctor, suggests alternative diagnoses and explains its
"reasoning." The system will be field-tested late next year.
* * *
I now would like to ask for opinions on the following issue;
One reason experts (human) are valuable is because they are scarce. This
scarcity is one of the driving forces behind implementing an expert
system.
My question then, Is knowledge proprietary, if so, who owns it?
The domain expert, or the organization (university, corporation, etc.),
that the domain expert worked for when the special knowledge was acquired?
Is the knowledge "stored" within a domain experts' brain copyrightable?
Obviously an expert system is, the same as any piece of software. But
since, as so many people have said, "In the knowledge lies the power," an
expert system is only as valuable as the domain knowledge it contains.
I think that in the next few years, serious questions like these
will affect the way we think about computer systems. Another hard question
is who is liable? Let's say some expert system in medical diagnosis was
fed data, which is later found to be faulty, or downright wrong. And
a serious injury was related to a physician's use of this information.
If the writers of the expert system could be shown to be negligent in
the verification of their data, could it be possible that they are liable,
just as if it were a case of malpractice?
Kenneth Mayer (mayerk@UPenn-Graded)
------------------------------
Date: 21 Nov 85 07:35:42 EST (Thursday)
From: MJackson.Wbst@Xerox.ARPA
Subject: Re: IQ test for AI
>From Laursen.PA@Xerox in V3 #174:
". . .has anyone considered building an AI system that boosts a human
being's performance on IQ tests?"
Of course, *any* computer system which is more than a toy is intended to
boost human performance to some degree (answer questions one could not
answer so quickly, or at all, unaided). Closer to the point, the
unifying principle of Doug Engelbart's work over the years has been
"augmentation of the human intellect."
Mark
------------------------------
Date: Thu, 21 Nov 85 9:13:49 PST
From: Laursen.pa@Xerox.ARPA
Subject: Re: IQ test for AI
"MJackson.Wbst's message of Thu, 21 Nov 85 7:35:42 EST"
"...Of course, *any* computer system which is more than a toy is
intended to boost human performance to some degree (answer questions one
could not answer so quickly, or at all, unaided). "
True, but spreadsheets and inventory programs would be of little
immediate help in sitting down to an IQ test. In the particular domain
of IQ tests, what partitioning of human/machine skills would make sense?
For me it comes down to the question of artificial intelligence as a
replacement or an assistant to human intelligence. Who would you hire
for the job, the computer system with an IQ of 200, a person with an IQ
of 150, or a person with an IQ of 125 who knows how to work with a
computer system with an IQ of 200 :-)?
------------------------------
Date: Fri, 22 Nov 85 08:10 EST
From: D E Stevenson <dsteven%clemson.csnet@CSNET-RELAY.ARPA>
Subject: Hard Typing
I asked our local software engineer about investigations using
hard typing. His response was that such investigations have
been going on since about 1977. Much of this was done by Gannon.
Anyone interested in such work can contact David Hutchens on
Hutch@clemson.
Steve
------------------------------
Date: Thu 21 Nov 85 22:00:11-PST
From: Ken Laws <Laws@SRI-AI.ARPA>
Subject: Spang Robinson Report
Newsletter news: The TARGET newsletter has merged with The Artificial
Intelligence Report to form the Spang Robinson Report. Sara Spang
is the editor, Louis G. Robinson the publisher. The report is
available on-line through NewsNet and Mead Data Central's Nexis.
For subscription info contact Spang Robinson, 3600 West Bayshore
Road, Palo Alto, CA 94303. (It's $295 U.S. and Canada, $345 elsewhere.)
The contents of the first issue were listed in a recent AIList message
from Laurence Leff.
------------------------------
Date: Thu, 21 Nov 85 13:48:10 EST
From: Norman Haas <nhaas.yktvmz@ibm-sj.csnet>
Subject: A.I. to the rescue!
[Forwarded from the SRI bboard by Laws@SRI-AI.]
The following is a slightly abbreviated version of a recent press release. It's
interesting what applications AI is being put to these days...
Norm
Corporate News Bulletin
November 19, 1985
IBM ANNOUNCES NEW PROGRAM TO STRUCTURE COBOL CODE
IBM today announced a software program that uses "artificial
intelligence" techniques to structure existing programs written
in the COBOL language so they can be maintained and modified more
easily.
COBOL Structuring Facility (COBOL/SF) analyzes complex programs,
structures them into a top-down hierarchy of components and
identifies components for possible re-use in new program
development. This can extend the life of current programs,
reduce the time and cost to develop new ones and improve the
productivity and quality of program maintenance.
The "artificial intelligence" techniques used to organize the
unstructured program into a hierarchy of components resemble the
way a person would analyze and reorganize the program.
Since COBOL (Common Business Oriented Language) was one of the
first and most widely used computer programming languages, many
businesses and organizations depend on COBOL-developed programs
that have been modified repeatedly. This can result in programs
that are more and more complex and difficult to maintain and can
increase the risk of making changes that have hidden
consequences.
Programmers can use it to structure old code -- helping
make programs easier to understand, maintain and modify -- and
keep programs in structured form despite continued modifications.
It has two modes of operation: Analysis Mode and
Generation Mode. In Analysis Mode before structuring, it
analyzes the program and produces a report that identifies
unreachable code and endless loops it has detected. The report
also lists especially complex parts of the program where minor
modifications by a programmer could improve the structuring
process.
In Generation Mode, it structures the program and organizes
it in a hierarchy of individually structured procedures. This
process -- which resembles the way a person would analyze and
reorganize the program -- is an example of what many computer
scientists refer to as "artificial intelligence."
After a program is structured, a report summarizes the changes
by it and provides a structured table of program contents.
The structured program then must be recompiled.
------------------------------
End of AIList Digest
********************
