Abstract
It has been proposed that operators sometimes refuse to follow the guidance of a decision aid that they know is offering correct or useful advice (intent errors). To test this premise, some participants received a “recommendation” from a computerized decision aid before each of 100 target detection trials. These operators were told that the aid always offered correct advice. A control group performed the same task without the decision aid. Most of the aided participants (82%) made one or more errors because they were unwilling to follow the aid's recommendations. Furthermore, the number of errors made by the experimental and control groups was not statistically different. Extrapolation of the results suggests that efforts to reduce automation disuse should include training designed to mitigate intent as well as appraisal errors.
Get full access to this article
View all access options for this article.
