🚀 go-pugleaf

This article originally appeared in Canadian AI Magazine, Winter 1995.
Anyone interested in AI and Canada should subscribe by joining
the Canadian Society for Computational Studies of Intelligence (CSCSI).
For more information on CSCSI including other articles,
visit 
http://ai.iit.nrc.ca/cscsi_point.html
For more IJCAI information visit:
http://ijcai.org/
 - Alan
====================================================================

IJCAI: International Jamboree for Canadian Artificial Intelligence

Alan K. Mackworth

The 1995 International Joint Conference on AI, IJCAI-95, will be held in
Montreal on August 20-25, 1995. Since  IJCAI is being held in Canada for
the second time, the editor of this esteemed organ decided, for reasons
known only to himself, but with infinite wisdom (editors always have
infinite wisdom) to ask me to write a short note on the historical
context for IJCAI and Canada. So here it is. We'll start with a little
quiz.

IJCAI conferences have been, and will be, held in 

Washington, D.C., USA (1969)    
London, England (1971)  
Stanford, California, USA (1973)  
Tbilisi, Georgia (1975)  
Cambridge, Massachusetts, USA (1977) 
Tokyo, Japan (1979)    
Vancouver, Canada (1981)   
Karlsruhe, Germany (1983)   
Los Angeles, California, USA (1985) 
Milan, Italy (1987)    
Detroit, Michigan, USA (1989)   
Sydney, Australia (1991)   
Chambery, France (1993)    
Montreal, Canada (1995)    
Yokohama, Japan (1997)   
Stockholm, Sweden (1999)   
Somewhere, USA (2001)

Now, as a test for your theory of inductive learning I would like you,
or your program, to induce the simplest rule for planning IJCAI
locations based on the historical record. What's the pattern? Don't peek
at the answer below.

Here's part of it in Prolog:

ijcai_planner(Year, Location) :-    
    Year =< 1989,
    Year mod 4 =:= 1,     
    continent(Location, north_america).

ijcai_planner(Year, Location) :-     
    Year =< 1989,     
    Year mod 4 =:= 3,     
    continent(Location, Continent),     
    Continent =\= north_america.

 et cetera

You get the idea. It was a trick question: there's no single simple
rule. As IJCAI went along it has dynamically invented its own version of
the planner. At IJCAI-77, for example, some of us assumed the rule was
that the conference alternated between the USA and the rest of the world
and made a pitch for IJCAI-79 to be at UBC in Vancouver. At that point
the declared rule was clarified to the effect that IJCAI alternated
between North America and the rest of the world, so it went to Tokyo.
That clarification entitled us to make a successful bid, at IJCAI-79 in
Tokyo for IJCAI-81 in Vancouver. Later as it became clear that IJCAI
could better serve its mandate by moving more of its conferences outside
North America the rule was changed so that only every third conference
was in North America. And now the lead time for conference planning is
six years. 

Planning the location involves many factors but the key criteria include
local support, both within the country and at the site from the AI
community, and the likelihood that the conference will trigger or
consolidate growth in the national AI community. Canada will be the
first country, besides the USA, to have the honour of hosting IJCAI
twice, which is a strong vote of confidence in our AI community. 

The Canadian AI community has played an ongoing role in the non-profit
society, International Joint Conferences on Artificial Intelligence,
Inc.(IJCAII) that co-sponsors the conferences with the local societies
(CSCSI and AAAI for IJCAI-95). Don Walker, as the guiding spirit behind
IJCAII for many years, successfully built it into a truly international
force, while minimizing the bureaucratic inertia that can plague
international efforts. IJCAII is run by a minimalist organization
consisting of the Board of Trustees, chaired by the current conference
chair, and the Secretary-Treasurer. We have declared Ron Brachman, the
current Secretary-Treasurer, and Priscilla Rasmussen, who toils behind
the scenes, honorary Canadians.

Given the Canadian proclivity for international action it's not too
surprising that we have participated in so many ways in IJCAII. Just a 
few of the connections are mentioned here. The IJCAII Computers and
Thought Award to a young AI researcher went to Hector Levesque in 1985.
He was the first researcher working outside the US to receive it. In
1987 another Canadian, Johan de Kleer (a UBC and MIT grad working at
Xerox PARC) was chosen. In 1993 Ray Reiter was honoured with the IJCAI
Award for Research Excellence. People who have served as officers of
various IJCAI's include Richard Rosenberg, Local Arrangements Chair
(1981), Alan Mackworth, Conference Chair for IJCAI-85, John Mylopoulos
and Ray Reiter, Program Chairs, IJCAI-91. And, of course, for IJCAI-95
the Canadians have taken over: Raymond Perrault is Conference Chair,
Renato de Mori is Local Arrangements Chair, Gordon McCalla is Tutorial
Chair and a passel of distinguished Canadians serve on the National
Committee. 

The impression that Canadians are over represented both on the Program
Committee and in the number of papers presented, compared to the
population as a whole and to the size of the community has been
confirmed by rigorous analysis; that is, I vaguely recall that someone
looked at the statistics once. The idea that there is any causal
connection between our presence on the PC and the proportion of papers
presented is a scurrilous rumour. The correlation is simply an artefact:
both are simple consequences of the excellence of AI in Canada. There's
also a rumour that the acronym IJCAI really stands for International
Jamboree for Canadian AI. Rumours like these are spread by those
paranoids who hold the view  that IJCAI has been taken over by `the
Canucks' - a secret society that believes in the power of the Maple
Leaf, eh. The paranoids also suggest, based on only a few shreds of
evidence such as lapel pins, that there are links between the Canucks
and the once-feared, but now toothless, Knights of the Lambda Calculus.

CSCSI is the world's oldest national AI society, founded in 1973. (Both
AISB and SIGART are, indeed, older but neither was founded as a national
AI society.) So it is only natural that CSCSI should host the world's
oldest and still most prestigious AI conference. The choices of
Vancouver and Montreal as the two host cities were only natural since
they are, as we all know, the hot spots of AI activity in the country.

Over the years many research areas have splintered off  from the IJCAI
community to form their own specialized workshops and conferences. In
1973, for example, pattern recognition was explicitly excluded causing
much sound and fury, and wailing and gnashing of teeth not to mention
IJCPR, a pale IJCAI imitation. Theorem proving, representation &
reasoning, vision, natural language, constraint programming, robotics
and fuzzy systems have all factored off. Does this mean that IJCAI is
obsolete? Absolutely not. Now we realize that carving up the field by
function institutionalizes a false view of the design of intelligent
agents. IJCAI is, as always, at the centre of the action.


---------------------------------------------------------------------------
Alan Mackworth  <mack@cs.ubc.ca>  Tel: (604) 822-4893   Fax: (604) 822-5485