The Turing Test for an artificially intelligent tutoring system is that if a machine tutor and a human tutor were placed on the other side of a partition, a student who is attempting to learn something could not tell the difference on the basis of the dialogue. In 2001: A Space Odyssey, the computer named HAL might have passed the Turing Test. HAL is fictional, of course, and extant artificially intelligent tutoring systems are a distant cry from HAL in terms of their cognitive abilities (cf., Kearsley, 1987; Sleeman & Brown, 1982).
Researchers and developers of artificially intelligent tutoring systems will benefit from a better understanding of natural intelligence. Instructional theorists will benefit also. In general, little attention has been paid recently to kinds of natural intelligence. Notable exceptions are Howard Gardner's (1985) theory of multiple intelligences and George Maccia's (1987; 1988) genetic epistemology of intelligent systems.
Maccia refers to four domains of natural intelligence: qualitative, quantitative, praxic, and inventivel. Since his epistemology is not well-known, I will first explain it with examples, then list his general conditions for such intelligences, and finally discuss their imDlications for instructional theorists and intelligent tutoring systerns.