Having been a math tutor for three years, I know just how important a good tutor is for some students to succeed in a class. A good tutor does not just work on problems; they work with the student and respect the fact that every student is different. Software that tutors though is not capable of this because it has only been programmed with one way to work out a problem, and could not know when to switch to another technique a student may better understand. Researchers at the University of Notre Dame, University of Memphis, and MIT have created some software to do just that though.
Intelligent Tutoring Systems (ITS) are programs with the ability to communicate more naturally and adapt its techniques to the student. AutoTutor, one of two programs the researchers created, will actually use teaching and motivational strategies human tutors use while actively trying to learn what the student’s knowledge is and keeping them engaged. Affective AutoTutor does everything AutoTutor does in addition to monitoring the student’s emotional states. This is accomplished by examining body language, conversational cues, and expressions. The software then responds in ways appropriate to keep down frustration and boredom with its own speech intonation, word choice, and the facial expressions of the animated tutor.
One-on-one human tutoring has been shown to improve student performance, and AutoTutor, which has already been tested by over one thousands students, and has shown an increase of roughly one letter grade with students. This does not quite reach as high as the best human tutors, but is better than novice and bad tutors.