Computer-based tutoring research is still alive

The Journal of Educational Technology & Society, Vol. 13, No. 1 has a new edition out on the topic of Intelligent Tutoring Systems.

From the editorial:

Intelligent Tutoring Systems (ITS) are meant to provide useful tutoring services for assisting the student. These services include coaching, assisting, guiding, helping, and tracking the student during problem-solving situations. To offer high-quality tutoring services, an ITS must be able to establish the correct student profile, then understand and diagnose the student cognitive as well as its affective state. This special issue of Educational Technology & Society presents recent works dealing with those matters.

I was curious to see that the topic of ITS is still alive. I remember that in the 1980s, huge sums of money were spent on research into the use of artificial intelligence techniques to develop automated educational tutoring, with almost no meaningful results (at least in terms of low cost reproducible tools). The basic problem at the time was that teachers were much more effective than computers in analysing and diagnosing learning problems and solutions, which are complex and have many possible causes and effects.

One new development now that is apparently being explored, according to this journal, is educational data-mining, to help identify intelligent tutoring strategies.

However, I still see problems in this approach to computerize the processes of learning. First there is a philosophical issue about the nature of learning. If learning is constructed, it is an intuitive and ‘fuzzy’ process whereas attempts to build computer models of human learning tend to be reductionist and by definition precise. Data-mining may provide ‘background’ or historical data about learners’ previous attempts at learning, but this does not necessarily mean they will predict future attempts at learning. Also, a fundamental premise of this approach is that there are what the computer scientists call ‘low’ and ‘high’ learners, in other words, those that get incorrect answers and those that get correct answers. But learning – even in the sciences – is not really like this (although as teachers we often try to make it so.) Learning is a complex process not easily or even properly reduced to mechanical principles. Lastly, there is the profound moral issue of whether we should be trying to replace human teachers with machines. Although I am a great believer in the value of technology for teaching and learning, I tremble at the thought of automated, machine-driven learning. (One article for instance tries to develop computer correlates for the processes that underpin social networking).

The irony here of course is that while it is almost impossible to get research funding now for qualitative studies of learning construction, computer scientists have plentiful sources of funding for this kind of research – even though it may be going (once again) into a black hole.