Chapter 9: An Intelligent Cat that can Deal with Disengaged Test Taking

As applications of artificial intelligence (AI) become increasingly common in everyday life, it is natural to explore its potential role in achievement measurement. This idea is reinforced by the ubiquitous presence of computers in our schools and organizations, and in the increased role they have played in the development, administration, and scoring of tests. The general notion of administering “smarter” achievement tests is attractive to educators who desire an efficient collection of valid information about what test takers know and can do.

However, the application of AI in achievement testing is not a particularly new idea. It started nearly 50 years ago, as the principles of tailored testing were introduced and investigated (Green, 1970; Linn, Rock, & Cleary, 1969; Lord, 1970; Owen, 1969; Weiss & Betz, 1973). This type of test eventually became known as a computerized adaptive test (CAT). Around that time, two major developments were influencing the field of measurement. First, item response theory (IRT) was emerging and just beginning its transformative impact on the field of measurement. Second, the possibility of computerbased testing was rapidly becoming a reality, as computers were becoming smaller and less expensive, and could support laboratories containing multiple computer terminals.