An AIS may be asked to work in a very sensitive domain, such as the
intensive care unit domain of Guardian. Every machine decision has drastic
consequences, and the system is accountable for all actions taken. In such
an environment, there are usually supervisors, who monitor the system's
decisions. Any behavior which is unacceptable to these human observers must
be corrected immediately. To facilitate this, AIS must be able to explain
its actions, in terms of a line of reasoning which would allow the
supervisor to pinpoint the error. The knowledge base thus uses a declarative style and a uniform representation that represents
causal information.
To return, press HOME.
To go to the next document, press NEXT.