The automated tutoring of introductory computer programming

Reviewer: Robin Jeffries

This paper, part of a special issue devoted to the challenges of teaching programming, describes an Intelligent Tutor for LISP. The tutor is one of the products of a project aimed at developing a theory of skill acquisition. The authors' goal seems to be to expose the tutor to a larger audience of computer professionals and educators. The paper is a readable, relatively nontechnical overview. It focuses on the workings of the tutor, with only cursory mention of attempts to evaluate its effectiveness. The authors describe the limitations of the current system and their plans to extend it. The authors have applied state-of-the-art technology in cognitive psychology and artificial intelligence to the construction of a practical tutoring system. The result is an innovative blend of standard Computer Assisted Instruction (CAI) and Intelligent Computer Assisted Instruction (ICAI). The final product is a large, complex decision tree, quite comparable to the more ambitious CAI systems developed over the last few decades. However, the decision tree is not laboriously handcrafted; rather, for the most part, it is automatically generated via an expert-system-like program that instantiates the authors' theory of tutoring. The resulting instruction can be much more fine-grained and complete than is typically achieved with conventional CAI, and it can be generated at a much lower cost per instructional hour. Furthermore, while a high-powered workstation is needed to develop the tutoring material, it can be delivered to students with low-cost personal computers. This paper ought to be read by anyone involved in computer-based instruction. It should give developers from both the ICAI and traditional CAI camps ideas that are potentially relevant to their own work.