Copy Link
Add to Bookmark
Report
AIList Digest Volume 2 Issue 074
AIList Digest Sunday, 17 Jun 1984 Volume 2 : Issue 74
Today's Topics:
AI Tools - Q'NIAL Request,
Knowledge-Based Systems - Spelling Correctors,
Metaphysics - Relevance of "Souls" to AI,
Scientific Method - Mathematics,
Linguistics - Commonsense Reasoning,
Brain Theory - Processing Power,
Conference - Hardware Design Verification
----------------------------------------------------------------------
Date: 15 Jun 1984 23:33-EST
From: Todd.Kueny@CMU-CS-G.ARPA
Subject: Q'NIAL
I have recently heard of a language developed in Canada
(Queens University?) called Q'Nial or Nial. These folks have been at
some shows (USENIX) and have a demo system which looks alot like a lisp
with algol syntax. Does anyone know about these guys?
Are there any technical papers?
(I think NIAL stands for Nested Inter???? Array Language.)
-Todd K.
------------------------------
Date: Fri, 15 Jun 84 11:58:54 PDT
From: Michael Pazzani <pazzani@AEROSPACE>
Subject: Spelling Correctors
I disagree with the statement that spelling correctors are knowledge-based
programs capable of outperforming even intelligent humans.
There are two basic parts to spelling correction: detection and correction.
In the case where there is more than one possible correction for a misspelled
word, people can of course use the context to find the correct spelling.
Selecting the proper choice is in many ways like selecting the intended
sense of a word.
Computers, of course, can be much better at detection of spelling errors
except when the misspelling is another word in the vocabulary. I would
not call checking a word against a prestored vocabulary knowledge based
even with a complex root stripping capability.
------------------------------
Date: 13 Jun 84 19:51:25-PDT (Wed)
From: hplabs!hao!seismo!cmcl2!floyd!clyde!burl!ulysses!allegra!princet
on!eosp1!robison @ Ucb-Vax.arpa
Subject: Re: elevance of "souls" to AI
Article-I.D.: eosp1.927
Philip Kahn, in his discussion of souls and essences, writes:
>> A "soul," like the concept of "essence," is undefinable.
>> The problem of "cognition" is far more relevent to the study of AI because
>> it can be defined within some domain... Whether "cognition"
>> constitutes a "soul" is again not relevent..."
I submit that the concept of "soul" is irrelevant only if AI is doomed
to utter failure. Use your imagination and consider a computer program
that exhibits many of the characteristics of a human being in
its ability to reason, to converse, and to be creative and unexpected in
its actions. How will you AI-ers defend yourself if a distinguished
theologian asserts that G-d has granted to your computer program a soul?
If he might be right, the program, and its hardware must not be destroyed.
Perhaps it should not be altered either, lest its soul be lost.
The casual destruction, recreation and development of computer programs
containing souls will horrify many people. You will face demonstrations,
destruction of laboratories, and government interference of the worst kind.
Start saving up now, for a defense fund for the first AI-er accused by
a district attorney of soul-murder.
On second thought, you have nothing to fear; no one in AI is really trying
to make computers act like humans, right?
- Toby Robison (not Robinson!)
allegra!eosp1!robison
decvax!ittvax!eosp1!robison
princeton!eosp1!robison
------------------------------
Date: Fri, 15 Jun 84 14:36:00 pdt
From: Harlan Sexton <hbs%BUGS@Nosc>
Subject: Mathematical Methods
It is true that most mathematics papers contain
little of the sort of informal, sloppy, and confused thinking that
always accompanies any of the mathematical discovery that I have been
a party to, but these papers are written for and by professional
mathematicians in journals that are quite backlogged.
Also, although I have always been intrigued by the differences beween
modes of discovery among various mathematicians of my acquaintance,
I never found knowing how others thought about problems
of much use to me, and I think that most practicing mathematicians
are even less inclined to wonder about such things than I was when I
was a "real" mathematician.
However, in response to the comment by David ???, I can only say that
I, and most of my fellow graduate students to whom I talked about such things,
had no trouble recalling the processes whereby we arrived at the ideas
behind proofs (and the process of proving something given an "idea"
was just tedious provided the idea was solid).
The process used to arrive at the idea, however, was as idiosyncratic
as the process one uses to choose a spouse, and it was generally as portable.
I found it very useful to know WHAT people thought about various things,
and I learned a great deal from my advisor about valuable attitudes toward
PDE's, for example (sort of expert knowledge about what to expect from a
PDE), but HOW he thought about them was
not useful. (With the exception of the infamous Paul J. Cohen, I felt that I
appreciated HOW these other people thought; it was just that it felt like
wearing someone else's shoes to think that way. In Cohen's case we just
figured that Paul was so smart that he didn't have to think, at least like
normal people.)
In the last year or so of my graduate career, someone came to the mathematics
department and interviewed a number of graduate students, including me,
about something which had to do with how we thought about mathematical
constructs (of very simple types which they specified). Presumably this
information, and related things, would be of some interest to Bundy. I'm
sorry that I can't be more specific, but if he would contact the
School of Education at Stanford (or maybe the Psychology Dept., but I think
this had to do with some project on mathematics education), they might be
able to help him. There is also a short book by J. Hadamard, published by
Dover, and some writings by H. Poincare', but as I recall these weren't
very detailed (and he probably knows of them already anyway). Finally,
I know that for a while Paul Cohen was interested in mathematical theorem
proving, and so he might have some useful information and ideas, as well.
(I believe that he is still in the Math. Dept. at Stanford. The AMS MAA SIAM
Combined Membership List should have his address.) --Harlan Sexton
------------------------------
Date: Fri 15 Jun 84 13:25:05-PDT
From: Ken Laws <Laws@SRI-AI.ARPA>
Subject: Commonsense Reasoning?
I'm not sure whether the following probes our commonsense reasoning
ability or simply demonstrates a quirk of natural language:
"The Monday class will meet on Tuesday next week. The Wednesday
class will thus be the day after the Monday class. (We may
decide to hold the Friday class on Wednesday and the Wednesday
class on Friday if everyone can make it then.)"
Another example along the same line is:
If 3 were half of 5, what would a third of 10 be?
Although it's easy enough to finesse the problem by claiming that
this is nonsense, most people would find the answer 4 to be quite
reasonable. The answer is derived by following the chain 3:5/2
as 6:5 as 12:10 as 4:10/3, where ":" represents some unspecified
transformation that is assumed to be linear. I consider this similar
to the nonlinear Monday:Tuesday reasoning above.
-- Ken Laws
------------------------------
Date: 14 Jun 84 7:29:14-PDT (Thu)
From: ihnp4!cbosgd!rbg @ Ucb-Vax.arpa
Subject: Re: A Quick Question - Mind and Brain
Article-I.D.: cbosgd.20
> I believe there are aprroximately 10 to the 9th neurons in a human
> brain, if that's of any help. Add in the glial cells (there is some
> debate about their function) and it comes to 10 to the 10th.
> Bob Binstock
Those numbers are both wrong, but so was the number in my original posting.
Let me correct the numbers, and add the discussion to some other groups which
may or may not be interested.
Recent estimates of the number of neurons in the human brain have been
increasing, for a current estimated total of between 30x10^9 and 50x10^9.
Glial cells outnumber neurons by at least 10 to one, and occupy about half
the volume of the brain, but the ratio varies widely between brain regions,
and between species within a brain region.
To get an estimate of the computational equivalent of the brain:
Assume 5x10^10 neurons with 2x10^4 synapses each = 10^15 synapses/brain.
Each synapse, on average, adds in a quantity about 20 times/sec (it can
go much faster, but not many do at the same time). So that's 2x10^16 very
simple approximate adds per second.
Even when everything is just right, a Cray can't do better than about 10^9
simple integer adds per sec.
So, IF THE SYNAPSES ARE BEING USED WITH TOTAL EFFICIENCY FOR PERFORMING
THE TASK, a brain is worth about 10^7 Crays.
[Credit for this calculation to Terry Sejnowski (Biophysics, Johns Hopkins)
and Geoff Hinton (Computer Science, Carnegie-Mellon)].
It is not surprising that most tasks use only a small fraction of this
capacity. However, I think the computation of the amount of information
used to store sensory perceptions by hound!rfg may be misleading:
>>if I assume a visual field as 10**3 bits high by 10**4 bits wide
>>by 10 bits for color and shading of each element, we have 10**8 bits
>>per visual field. Suppose a life time of 72 years and 16 hours a day of
>>observing (neglecting "visual dreams" which may also be remembered),
>>with a new observation every 10 seconds. I multiply it all out to
>>about 1.5 x 10**16 bits. (187,500 billion bytes?)
>>Adding audio, tactile, olfactory, taste to that ought to easily run the
>>total over 200 gigabytes. That's just for remembering observations
>>(eidetically, which is a faculty some do have).
Most people do not remember every detail of every scene they ever see.
How much of your early childhood (0-4) do you even remember at all?
Emotional content of a situation can have a large impact on what and how
much you recall. Dangerous or joyful experiences stand out in memory
more than most neutral events.
The role of language is also an important issue in considering the storage
and information processing capacity of the brain. Using a word to stand
for the many features which make up an object or a concept is an incredible
data compression. This may be why the gradual increase in computational
ability across primate evolution is not a very satisfying explanation for
the quantum leap in human intellectual ability. Many of the explanations
of the origin of consciousness rely on the advantages of language for
improving analytic ability. The one I like best is Julian Jaynes idea
(The Origin of Consciousness in the Breakdown of the Bicameral Mind):
that consciousness is not just a simple consequence of language, but
that the exponential growth in knowledge fostered by language generates
self-consciousness only after certain kinds of concepts are introduced
into language. This allows him to trace the evolution of consciousness
by literary analysis!
Rich Goldschmidt -- a former brain hacker (now reformed?)
cbosgd!rbg
------------------------------
Date: Wednesday, 13 June 1984 11:41:32 EDT
From: Mario.Barbacci@cmu-cs-spice.arpa
Subject: call for papers
CALL FOR PAPERS
WORKSHOP ON HARDWARE DESIGN VERIFICATION
November 26-27, 1984
Technical University of Darmstadt, F.R. Germany
This workshop is organized by IFIP Working Groups 10.2 and 10.5. Program
will cover all aspects of verification methods for hardware systems,
including:
Correctness of hardware design,
Tools and methodologies for verification,
Verification of multilevel descriptions,
Timing verification,
Temporal logic,
Correctness by construction,
Circuit extractors,
Design rule checkers,
Language issues,
Application of AI techniques.
PARTICIPATION IS BY INVITATION ONLY. If you would like to propose a
contribution to the workshop send a short summary of the intended
presentation to the Workshop Chairman before July 31, 1984. Notices of
acceptance will be sent by September 15, 1984.
Workshop Committee:
Hans Eveking (Chairman) Stephen Crocker
Institut fuer Datentechnik Aerospace Corporation
Technical University of Darmstadt P.O. Box 92957
D-6100 Darmstadt Los Angeles
Fed. Rep. Germany California 90009
(49) (6151) 162075
George J. Milne Robert Piloty
Computer Science Department Institut fuer Datentechnik
University of Edinburgh Technical University of Darmstadt
Edinburgh, Scotland D-6100 Darmstadt
United Kingdom Fed. Rep. Germany
------------------------------
End of AIList Digest
********************
