[Top] [All Lists]

Re: [ontolog-forum] IBM Watson's Final Jeopardy error "explanation"

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Ed Barkmeyer <edbark@xxxxxxxx>
Date: Thu, 17 Feb 2011 17:15:47 -0500
Message-id: <4D5D9E13.9090802@xxxxxxxx>

John F. Sowa wrote:    (01)

> The amount of high quality research done under the name of AI
> has been enormous, and it has been so thoroughly integrated into
> the foundations of computer science that its AI origins have often
> been forgotten:
>       (02)

I can agree with this.  Unfortunately, the amount of high-quality 
research that was done under the name AI has been confused with the 
amount of high quality research in computational technologies that was 
not done under the name AI, because they have all been integrated into 
the foundations of computer science.    (03)

>   1. Just look at LISP, which contributed the if-then-else statement,
>      recursion, lambda expressions, metalanguage, garbage collection,
>      the ability to write an interpreter or compiler of a language
>      in itself, etc.  (McCarthy, by the way, was also a member of
>      the IFIP committee that designed Algol, so his influence was
>      very direct.)
>       (04)

Note the merger here of Algol and LISP ideas being attributed to LISP.  
I believe many of the programming language pioneers of the late 1950s 
would argue that McCarthy took from their discussions as much as he 
provided, and that if-then-else and recursion, which were features of 
Algol 58, hardly originated with LISP.  Further, the well-known Algol 60 
'thunk' was arguably a lambda expression.     (05)

Writing a compiler for a language in itself is a 'bootstrap' thing -- 
you must first write a compiler in assembly code.  The reason why LISP 
got there first was that the fundamentals of the language were simple, 
and thus the bootstrap process was very short.  Doing this for a typed 
language with variables and complex syntax takes rather longer, and 
convincing a commercial organization that it is worth the investment is 
more difficult.  John Backus didn't get support for that concept until 
1963, but we finally did see IBM 360 Fortran H -- an optimizing compiler 
written in Fortran!  Whether John got the idea from McCarthy I couldn't say.    (06)

I do think that LISP was the original use of 'garbage collection', and 
it was definitely the first 'applicative' programming language -- its 
execution model is not procedural; it is composition of functions.  But 
I think Wegner and Gear and Backus and Aronson have equal right to claim 
if-then-else, recursion, and the foundations of compiler technology.  
But you see, they weren't "AI" people.    (07)

>   2. Java is basically LISP + CLOS (Common Lisp Object System)
>      written in a syntax based on C.      (08)

I wonder if Jim Gosling will agree with this characterization.  I took 
it rather that he had the benefit of 20 years worth of appoximations to 
object-oriented programming, from Ada and its precursors, Smalltalk 
(from PARC), C and C++ (from Bell Labs), LISP/CLOS (Stanford, MIT), 
SAIL, Euclid, Objective C, etc.  The byte-code idea originated with the 
GE POPS compilers of the early 60s (that I know of, there may be 
precursors), and so on.  Amazing as it might seem, a lot of intelligent 
effort in programming language design was done outside the LISP 
community, and Jim was definitely aware of most of it.  Put another way, 
I would be willing to bet that Bill Plauger and Jean Ichbiah had more 
influence on Java than John McCarthy did.  ;-)    (09)

>      But the AI community had
>      30 years of experience in using and extending that technology.
>      Sun (which designed Java) was founded by former Stanford
>      students who learned LISP and AI from McCarthy and others
>      and who built their company by selling workstations for AI.
>   3. Most of the technology for logic-based systems, theorem provers,
>      formal languages, parsers, etc., was either pioneered in AI
>      or applied and extended in AI projects.
>       (010)

Yes.  That is the trade that called itself AI, along with a lot of 
natural language processing technology.    (011)

>   4. People like Ted Codd, who founded the relational DB community,
>      were strongly influenced by AI.  Codd wrote his PhD dissertation
>      on cellular automata and participated in joint projects on AI
>      related issues.  Among them was his RENDEZVOUS system for
>      an English query language for RDBs (and, by the way, Codd's
>      group used a parser that I wrote for their front end).
>       (012)

Yes.  Ted Codd is one of the great minds in computer science.  He is 
known for his contributions to 'database science', which in the 1970s 
was very different from what was then called AI.  But he also made his 
mark on other AI and 'ai' technologies.    (013)

 From 1955 to present, a lot of programming was mathematical analysis, 
and a lot of programming was business record processing, and a lot of 
programming was something else.  And all of that something else, in 
several very different communities, might reasonably be called 
'artificial intelligence'.  In the robotics community, for example, some 
of the pioneers will talk about LISP, and others cellular automata, or 
about Basic and control languages and signal processing -- 
software/hardware that linked computational intelligence to the real 
world.  In text analysis, you have things like SNOBOL; in simulation, 
you had Simula and Simscript.  Compiler writing itself was a black art 
in the 1960s (before Dave Gries), because it was a specialized form of 
'language processing and interpretation', and much of the work in 
'computational linguistics' arose from working with formal languages and 
compilers.  Image analysis was pioneered at NBS and Bell Labs and the 
University of Maryland in the 1950s and early 60s, but not usually 
called AI, until much later, because the original work was done by 
"electrical engineers".  And then there was 'operations research' -- 
practical applications of heuristic, statistical and discrete 
mathematical algorithms to problems of planning and resource allocation, 
which made major contributions to search algorithms and rule 
stratification and so on.  Was Samuels' checker-player for the IBM 704 
an AI application or an OR application?  The amount of high quality 
research that was 'artificial intelligence' and not done under the name 
AI from 1955 to 1985 was also very important to computer science as we 
know it.    (014)

My point is only that it is now convenient for John to recall many of 
the 'something else' technologies as 'AI contributions', when "AI" was 
for 20 years a reserved designation for a handful of universities and 
other programs working directly on knowledge modeling and language 
processing.  There were many major contributions to machine intelligence 
in many other domains, and it wasn't until the 1990s that we really 
started to see those technologies come together.    (015)

Computer science as we know it owes a great deal to many great minds 
practicing disciplines with many different designations, and those 
designations have changed as the disciplines became established.  Yes, 
the great minds of AI in its first heyday were among those, but there 
are many pioneers of the 'something else' technologies who also deserve 
recognition, even though their contributions to computational 
intelligence were not always called "AI".  We are all in the business of 
developing machines that replace human intelligence in the performance 
of tasks.  Exactly what part of that discipline is "AI"?    (016)

-Ed    (017)

> PB
>> [AI] now re-emerges with respectable "semantic web" clothing.
> Please note that the Semantic Web is just a tiny subset of AI
> technology, and the primary developers came from the AI community.
> The person who developed RDF was Ramanathan Guha, who wrote his
> PhD dissertation at Stanford with McCarthy as his supervisor.
> While he was finishing his PhD, he worked on Cyc and became
> the associate director of Cyc.  He later went to Apple, where
> he developed the Meta Content Framework (MCF).  He then went
> to Netscape, where he worked with Tim Bray to rewrite MCF in
> XML to form RDF.
> Guha later collaborated with Pat Hayes and others (also from
> the AI community) to define the semantic foundations for RDF
> and OWL.  (Just check Google for "Guha Hayes RDF" and
> "Guha Hayes OWL" to find the W3C documents.)  And OWL began
> as a combination of two AI projects, DAML + OIL, and was
> further enhanced by people from the AI community.
> KJ
>> Watson is a question answering machine and a very good one. Maybe one
>> day they will deploy it on your mobile phone with a Internet connection
>> to the processing and storage unit in the cloud similar to the Wolfram
>> Alpha App. Watson is not intelligent in the sense that it does not
>> understand the answers or questions but it turns out that in many cases
>> this is not necessary. I think that as a research domain we should be
>> rather happy that Watson won and congratulate IBM -- it is a strong
>> showcase for our work.
> I strongly agree.  The people who worked on Watson had a strong
> foundation in both AI and comp. sci.  It is a respectable piece
> of research.
> Anybody who doubts these points should do some remedial studies
> in the history of AI and computer science.
> John
> _________________________________________________________________
> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
> Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
> Shared Files: http://ontolog.cim3.net/file/
> Community Wiki: http://ontolog.cim3.net/wiki/ 
> To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
> To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
>       (018)

Edward J. Barkmeyer                        Email: edbark@xxxxxxxx
National Institute of Standards & Technology
Manufacturing Systems Integration Division
100 Bureau Drive, Stop 8263                Tel: +1 301-975-3528
Gaithersburg, MD 20899-8263                Cel: +1 240-672-5800    (019)

"The opinions expressed above do not reflect consensus of NIST, 
 and have not been reviewed by any Government authority."    (020)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (021)

<Prev in Thread] Current Thread [Next in Thread>