Formal Evaluation of The Expert System DEMOTOX

Abstract

The expert system DEMOTOX has been developed to demonstrate the utility of applied artificial intelligence (AI) to aid in assessing groundwater contamination potential. A formal evaluation of DEMOTOX in terms of perceived threat, overall confidence, time savings, and “super expert” agreement is presented. Two groups of participants were asked to solve a series of hypothetical groundwater contamination problems. One group comprised nonexperts (graduate students), while the other group was composed of “experts” (professional environmental engineers and scientists). Participants alternated in using DEMOTOX to aid in assessing the hypothetical contamination problems. DEMOTOX consistently saved the nonexperts time and significantly improved their performance. While the software did not offer the experts significant time savings, it did improve their overall agreement with the designated “super‐expert.”