AutoTutor: an intelligent tutoring system with mixed-initiative dialogue
IEEE Transactions on Education
Hi-index | 0.00 |
We explored the impact on learning of interactive simulations that were coordinated with AutoTutor, a learning environment that helps students by holding a conversation in natural language. We randomly assigned 132 college students to one of three conditions: AutoTutor without simulations, AutoTutor with simulations, and a Monte Carlo AutoTutor that randomly generated dialogue moves. A pretest-posttest design was used to measure learning gains, as measured by objective multiple choice questions. All versions of AutoTutor were successful in promoting learning. The Monte Carlo AutoTutor produced significantly lower gains than the interactive simulation version for higher knowledge learners, and the direction of the three means were in the predicted direction. Improved simulation dialogues, modeling of good simulation manipulation strategies, and faster display of simulations are expected to enhance learning in future versions of AutoTutor.