Users of a decision aid must make a series of decisions that
determine their level of dependence on the aid. First, they
may or may not choose to use the aid at all. Second, if they
use it, they may be able to set the mode in which the aid
will operate, including possibly the degree to which a task
is turned over to the aid or requires human intervention.
During the course of an operation, users must decide whether
or not, or how often, to monitor the aid, and whether or not
to revisit decisons about the aid's mode of operation. After
the aid has produced a recommendation, the user must decide
whether to accept it (with or without modifications) or reject
it; or in the case of some aids, whether or not to modify
the aid's parameters and wait for another recommendation.
All these decisions depend on the degree of trust of the
user in the aid under varying conditions and over varyiong
spans of time. Yet training of decision aid users has seldom
touched on this most critical issue. A key to user acceptance
and effective use of decision aids is an understanding of
both its strengths and weaknesses.
In research for the Army's Rotorcraft Pilot's Associate Program,
CTI developed a systematic framework for understanding trust
in decision aids, its parameters, and how it evolves, The
framework also describes the relationship between trust at
any time and user decisions regarding reliance on the aid.
Finally, the framework was applied in the development of training
for users of the RPA.