Copy Link
Add to Bookmark
Report
AIList Digest Volume 2 Issue 078
AIList Digest Sunday, 24 Jun 1984 Volume 2 : Issue 78
Today's Topics:
AI Programming - Characteristics,
Commonsense Reasoning - Hypothetical Math,
Cognition - Humor & Memory & Intuition,
Seminar - Full Abstraction and Semantic Equivalence
----------------------------------------------------------------------
Date: 20 Jun 84 12:14:49-PDT (Wed)
From: hplabs!hpda!fortune!amd70!intelca!glen @ Ucb-Vax.arpa
Subject: Re: Definition of an AI program
Article-I.D.: intelca.317
As a half-serious/half humorous suggestion:
Consider the fact that most of man's machines are built to do the same
thing over and over and do it very well. Some random examples:
- washing machine
- automobile hood fastner in production line
- pacman video game
AI programs (hopefully) don't fit the mold, they don't spend their lives
performing the same routine but change as they go.
^ ^ Glen Shires, Intel, Santa Clara, Ca.
O O Usenet: {ucbvax!amd70,pur-ee,hplabs}!intelca!glen
> ARPA: "amd70!intelca!glen"@BERKELEY
\-/ --- stay mellow
------------------------------
Date: Fri 22 Jun 84 11:28:46-PDT
From: Richard Treitel <TREITEL@SUMEX-AIM.ARPA>
Subject: a third of ten
Please. Everyone knows that 2*2=5 for sufficiently large values of 2.
More to the point, if you take the square root of 5 and round to the nearest
integer, you get 2. Again, if you take half of 5 and round to nearest using
accepted method, get 3. A third of ten now becomes 3 as well. How many AI
people does it take to change a lightbulb?
- Richard
[One graduate student, but it takes eight years. -- KIL (from John
Hartman, CS.Hartman@UTexas-20) ]
------------------------------
Date: 21 Jun 84 10:51:26-PDT (Thu)
From: decvax!decwrl!dec-rhea!dec-rayna!swart @ Ucb-Vax.arpa
Subject: Re: Commonsense Reasoning?
Article-I.D.: decwrl.1845
I am reminded of an old children's riddle:
Q. If you call a tail a leg, how many legs does a horse have?
A. Four. Calling a tail a leg doesn't make it so.
Mark Swartwout
UUCP {allegra,decvax,ihnp4,ucbvax}!decwrl!rhea!rayna!swart
ARPA MSWART@DEC-MARLBORO
------------------------------
Date: 21 Jun 84 22:07 PDT
From: Shrager.pa@XEROX.ARPA
Subject: Memory
This might amuse. Authorship credit to Dave Touretzky@CMU.
From: Dave Touretzky (DT50)@CMU-CS-A
To: Jeff Shrager <shrager.PA>
Subject: Q-registers in the brain
ENGRAM (en'-gram) n. 1. The physical manifestation of human memory -- "the
engram." 2. A particular memory in physical form. [Usage note: this term is
no longer in common use. Prior to Wilson & Magruder's historic discovery,
the nature of the engram was a topic of intense speculation among
neuroscientists, psychologists, and even computer scientists. In 1994
Professors M. R. Wilson and W. V. Magruder, both of Mount St. Coax University
in Palo Alto, proved conclusively that the mammalian brain is hardwired to
interpret a set of thirty seven genetically-transmitted cooperating TECO
macros. Human memory was shown to reside in 1 million Q-registers as
Huffman-coded uppercase-only ASCII strings. Interest in the engram has
declined substantially since that time.]
--- from the New Century Unabridged English Dictionary,
3rd edition, A.D. 2007. David S. Touretzky (Ed.)
------------------------------
Date: 19 Jun 84 16:02:49-PDT (Tue)
From: ihnp4!houxm!mhuxl!ulysses!gamma!pyuxww!pyuxn!rlr @ Ucb-Vax.arpa
Subject: Re: A Quick Question - Mind and Brain
Article-I.D.: pyuxn.769
> (2) Intuition - by this I mean huge leaps into discovery
> that have nothing to do with the application of logical
> association or sensual observation. This kind of stuff
> happens to all of us and cannot easily be explained by
> the physical/mechanical model of the human mind.
>
> I agree that if you could build a computer big enough and fast
> enough and taught it all the "right stuff", you could duplicate
> the human brain, but not the human mind.
Intuition is nothing more than one's subconscious employing logical thought
faster than the conscious brain can understand or realize it. What's all the
fuss about? And where's the difference between the "brain" and the "mind"?
What can this "mind" do that the physical brain doesn't?
A good dose of Hofstadterisms and Smullyanisms ("The Mind's 'I'" provides
good examples) puts to rest some of those notions of mind and brain.
"I take your opinions and multiply them by -1."
Rich Rosen pyuxn!rlr
------------------------------
Date: 19 Jun 84 13:55:43-PDT (Tue)
From: hplabs!hao!seismo!ut-sally!utastro!bill @ Ucb-Vax.arpa
Subject: Re: A Quick Question - Mind and Brain
Article-I.D.: utastro.114
> (1) Subconscious memory - a person can be enabled (through
> hypnosis or by asking him the right way) to remember
> infinite details of any experience of this or prior life
> times. Does the mind selectively block out trivia in order
> focus on what's important currently?
One of the reasons that evidence obtained under hypnosis is
inadmissable in many courts is that hypnotically induced
memories are notoriously unreliable, and can often be completely
false, even though they can seem extremely vivid. In some states,
the mere fact that a witness has been under hypnosis is enough to
disqualify the individual's testimony in the case.
I have personal, tragic experience with this phenomenon in my own
family. I don't intend to burden the net with this, but if anyone
doubts what I say, I will be glad to discuss it by E-mail.
Bill Jefferys 8-%
Astronomy Dept, University of Texas, Austin TX 78712 (USnail)
{allegra,ihnp4}!{ut-sally,noao}!utastro!bill (uucp)
utastro!bill@ut-ngp (ARPANET)
------------------------------
Date: 20 Jun 84 9:22:50-PDT (Wed)
From: hplabs!hao!seismo!ut-sally!riddle @ Ucb-Vax.arpa
Subject: Re: A Quick Question - Mind and Brain
Article-I.D.: ut-sally.2301
Now that Chuqui's obligingly created net.sci, why don't we move this
discussion there? Is there any reason for it to go on in five
newsgroups simultaneously? If interest continues, perhaps this topic
will form the basis for net.sci.psych.
Followups to net.sci, please.
--- Prentiss Riddle ("Aprendiz de todo, maestro de nada.")
--- {ihnp4,harvard,seismo,gatech,ctvax}!ut-sally!riddle
------------------------------
Date: Thu, 21 Jun 84 15:47 CST
From: Nichael Cramer <cramer%ti-csl.csnet@csnet-relay.arpa>
Subject: Memory
>
>From: hplabs!hpda!fortune!crane @ Ucb-Vax.arpa
>
> (1) Subconscious memory - a person can be [...]
But, brain is mind is brain is mind is brain is mind is brain...
[what else have you got to work with?]
So long and thanks for all the fish,
NLC
------------------------------
Date: 22 Jun 1984 1825-PDT (Friday)
From: gd@sri-spam (Greg DesBrisay)
Subject: Re: A Quick Question - Mind and Brain
Article-I.D.: aplvax.663
>The other thing to note is that while each 'memory cell' in a computer
>has ~2 connections, each 'memory cell' in the brain has ~100. Since
>processing power is relative to (cells * connections), a measure of
>relative capacities is not sufficient for comparison between the brain
>and the CRAY. -Lloyd W. Taylor
In addition, many connections in the human brain are analog in
character, so any comparison with a binary digital computer must
multiply the number of connections by the number of bits necessary to
digitize the analog range of each synapse. To do that, one would have
to know what analog resolution is required to accurately model the
behavior of a synapse. I'm not sure if any one has figured that one out
yet.
Greg DesBrisay
SRI
------------------------------
Date: 20 Jun 84 9:20:43-PDT (Wed)
From: decvax!mcnc!unc!ulysses!allegra!princeton!eosp1!robison @ Ucb-Vax.arpa
Subject: Re: Mind and Brain
Article-I.D.: eosp1.954
I'm not comfortable with Rich Rosen's assertion that intuition
is just the mind's unconscious LOGICAL reasoning that happens
too fast for the conscious to track. If intuition is simply
ordinary logical reasoning, we should be just as able to
simulate it as we can other tyes of reasoning. In fact, attempts
to simulate intuition account for some rather noteworthy successes
and failures, and seem to require a number of discoveries before
we can make much real progress. E.g.:
I think it is fair to claim that chess players use
intuition to evaluate chess positions. We acknowledge that
computers have failed to be intuitive in playing chess in at
least two ways that are easy for people:
- knowing what kinds of tactical shots to look for
in a position
- knowing how to plan longterm strategy in a position
In backgammon, Hans Berliner has a very successful program that
seems to have overcome the comparable backgammon problem.
His program has a way of deciding, in a smooth, continuous fashion,
when to shift from one set of assumptions to another while
analyzing. I am not aware of whether other people have been able
to develop his techniques to other kinds of analysis, or whether
this is one flash of success. Berliner has not been comparably
successful applying this idea to a chess program.
(The backgammon program defeated then world champion in a short
match, in which the doubling cube was used.)
[There was general agreement that the program's play was inferior,
however. Another point: while smooth transitioning between strategies
is more "human" and easier to follow or explain (and thus to debug
or improve), I can't see that it is inherently as powerful as
switching to a new optimal strategy at each turn. -- KIL]
Artists and composers use intuition as part of the process of
creating art. It is likely that one of the benefits they gain
from intuition is that a good work of art has many more internal
relationships among its parts than the creator could have planned.
It is hard to see how this result can be derived from "logical"
reasoning of any ordinary deductive or inductive kind. It is
easier to see how artists obtain this result by making various
kinds of intuitive decisions to limit their scope of free choice
in the creative process.
Computer-generated art has come closest to emulating this process
by using f-numbers rather than random numbers to generate
artistic decisions. It is unlikely that the artist's intuition
is working as "simply" as deriving decision from f-numbers.
It remains a likely possibility that a type of reasoning that we
know little about is involved.
We are still pretty bad at programming pattern recognition, which
intuitive thinking does spectacularly well. If one wishes to assert
that the pattern recognition is done by well-known logical processes,
I would like to see some substantiation.
- Toby Robison (not Robinson!)
allegra!eosp1!robison
decvax!ittvax!eosp1!robison
princeton!eosp1!robison
------------------------------
Date: 20 Jun 84 18:14:17-PDT (Wed)
From: decvax!linus!utzoo!henry @ Ucb-Vax.arpa
Subject: Re: A Quick Question - Mind and Brain
Article-I.D.: utzoo.3971
John Crane cites, as evidence for the human mind being impossible to
duplicate by computer, two phenomena.
(1) Subconscious memory - a person can be enabled (through
hypnosis or by asking him the right way) to remember
infinite details of any experience of this or prior life
times. Does the mind selectively block out trivia in order
focus on what's important currently?
As far as I know, there's no evidence of this that will stand up to
critical examination. Even disregarding the "prior life times" part,
for which the reliable evidence is, roughly speaking, nonexistent,
the accuracy of recall under hypnosis is very doubtful. True, the
subject can describe things in great detail, but it's not at all proven
that this detail represents *memory*, as opposed to imagination. In
fact, although it's quite likely that hypnosis can help bring out things
that have been mostly forgotten, there is serious doubt that the memories
can be disentangled from the imagination well enough for, say, testimony
in court to be reliable when hypnosis is used.
(2) Intuition - by this I mean huge leaps into discovery
that have nothing to do with the application of logical
association or sensual observation. This kind of stuff
happens to all of us and cannot easily be explained by
the physical/mechanical model of the human mind.
The trouble here is that "...have nothing to do with the application
of logical association or sensual observation..." is an assumption,
not a verified fact. There is (weak) evidence suggesting that intuition
may be nothing more remarkable than reasoning and observation on a
subconscious level. The human mind actually seems to be much more of
a pattern-matching engine than a reasoning engine, and it's not really
surprising if pattern-matching proceeds in a haphazard way that can
sometimes produce unexpected leaps.
Henry Spencer @ U of Toronto Zoology
{allegra,ihnp4,linus,decvax}!utzoo!henry
------------------------------
Date: 20 Jun 84 17:14:58-PDT (Wed)
From: ucbcad!tektronix!orca!shark!hutch @ Ucb-Vax.arpa
Subject: Re: A Quick Question - Mind and Brain
Article-I.D.: shark.838
| Intuition is nothing more than one's subconscious employing logical
| thought faster than the conscious brain can understand or realize it.
| What's all the fuss about? And where's the difference between the
| "brain" and the "mind"? What can this "mind" do that the physical brain
| doesn't?
| Rich Rosen pyuxn!rlr
Thank you, Rich, for so succinctly laying to rest all the questions
mankind has ever had about self and mind and consciousness.
Now, how about proving it. Oh, and by the way, what is a "subconscious"
and how do you differentiate between a "conscious" brain and a "subconscious"
in any meaningful way?
And once you have told us exactly what a physical brain can do, then we
can tell you what a mind could do that it doesn't.
Hutch
------------------------------
Date: 21 June 1984 0802-EDT
From: Lydia Defilippo at CMU-CS-A
Subject: Seminar - Full Abstraction and Semantic Equivalence
[Forwarded from the CMU bboard by Laws@SRI-AI.]
Speaker: Ketan Mulmuley
Date: Friday, June 22
Time: 11:00
Place: 5409
Title: Full Abstraction and Semantic Equivalence
The Denotational Approach of Scott-Strachey in giving semantics
to programming languages is well known. In this approach each
construct of the programming language is given a meaning in a
domain which has nice mathematical properties.
Semantic equivalence is the problem of showing that this map --
the denotational semantics -- is faithful to the operational semantics.
Because known methods showing such equivalences were too complicated,
very few such proofs have been carried out.
Many authors had expressed a need for mechanization of these proofs.
But it remained unclear whether such proofs could be mechanized at all.
We shall give in this thesis a general theory to prove such equivalences
which has a distinct advantage of being mechanizable. A mechanized tool
was acually built on top of LCF to aid the proofs of semantic
equivalence.
Other central problem of denotational semantics is the problem of
full abstraction, i.e., determining whether the meanings given to two different
language constructs by the denotational semantics are equal whenever
they are operationally equivalent. This has been known to be a hard
problem and the only known general method of constructing such models
was the {\it syntactic } method of Milner. But whether such
models could be constructed semantically remained an important open
problem. In this thesis we show that this is indeed the
case.
------------------------------
End of AIList Digest
********************
