Engelmore Award Article
1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016
Figure 5. Mix of Technologies Deployed and Emerging in IAAI Articles, 1989–2016.
The dominant technologies include expert systems, machine learning, agents, natural language, and statistical learning, and are included
in the figure legend.
bines, steam turbines, and generators. Ed Thompson
and Ben Bassford celebrated the 30th anniversary of
the system with the IAAI community when they presented an update at the 2015 conference in Austin
(Thompson et al. 2015). Their paper summarizes the
many changes that have been incorporated into the
system over its lifetime, to deal with change in
requirements, the customer business organization,
and underlying computer technologies.
1989: Authorizer’s Assistant
A knowledge-based credit-authorization system for
American Express, the Authorizer’s Assistant
(Dzierzanowski et al. 1989) was the forerunner of
now standard credit card transaction analysis. It created a capability that we all take for granted today —
and complain about every time we are called to verify a charge — until at least we ourselves are the victims of fraud.
Expansion, improvement, and testing were
planned from the start to ensure consistency as the
knowledge base changed as well as ensure general
system performance. The team found that consistency, audit tracking, and evaluation were key to acceptance and return on investment (ROI). They observed,
“the [Authorizer’s Assistant] proved to be better than
all but the most expert credit card authorizers ... and
that translated directly into huge ROI.” The system’s
internal expert system incorporated 890 rules and ran
on rack-mounted Symbolics Lisp machines connected to an IBM mainframe.
Phil Klahr generously provided these retrospective
1989: Applications of Artificial Intelligence
to Space Shuttle Mission Control
This NASA application originated in the Mission
Control Center for STS-26 as a rule-based real-time