I read with interest today that some leading computer scientists believe that AI innovations might well transform basic educational practices, such as what goes on in the classroom. While I agree and believe these changes are most likely to be quite positive, let me suggest that the sequence is somewhat the opposite of what the pundits say.
Hannah Devlin (2023) writes in The Guardian that ‘[r]ecent advances in AI are likely to spell the end of the traditional school classroom’. Her article is inspired by Professor Stuart Russell, a British computer scientist based on UC Berkeley, who she reports to have said that ‘personalized ChatGPT-style tutors had the potential to hugely enrich education and widen global access by providing personalised tuition [tutoring] to every household with a smartphone.’ I am generally in agreement with this potential.
However, this and related forecasts of the dramatic implications of technical change tend to dismiss or forget the history of these innovations and under-estimate the human aspects of education. With respect to the technology, for example, the idea of using technology to enable more personalized education was promoted decades ago. It is only now that the technology is beginning to catch-up with the pedagogical potentials. With respect to the human side, the personalized tutorial involved more than personalization.
Take the late-Professor Alfred M. Bork, who taught in the Donald Bren School of Information and Computer Sciences at UC Irvine, who died in December 2007. He was a major pioneer of tutorial learning systems, using computer-based systems to provide ‘highly interactive’ online tutorials that approached the ideal of an Oxford one-to-one tutorial. While he was still an Emeritus Professor, he wrote a chapter for a book that Brian Loader and I edited, entitled Digital Academe (2002). In the chapter, Bork himself dedicates this chapter to Bertrand Ibrahim, who died in 2001, for playing a central role with Bork in developing the Irvine-Geneva system, on which his work has been anchored.
In 2002, about the time our book came out, I moved to the University of Oxford as the first Professor of Internet Studies and founding Director of the Oxford Internet Institute. It was there that I learned to appreciate the Oxford tutorial, which in its ideal form is based on one-to-one interaction between the tutor and the student that enables the personalization that Bork and his colleagues sought to achieve. However, the Oxford tutorial goes beyond personalization – addressing the specific needs of the student pursuing their chosen course of studies. It also moved away from what had become an American model of the teacher as the so-called ‘Sage on the stage’ in being the authority in the classroom, distilling their wisdom to the students. The sage model is most often contrasted to a ‘guide on the side’, but the Oxford tutor is somewhat more than a guide.
Traditionally, when Oxford students are examined, the students are not judged by their teachers or tutors, but by an exam board. If a student passes with flying colors, then the tutor can take some of the credit. The tutor’s job is to help the students learn what they need to know to excel in their examinations. They are almost a partner of the student and have a personal stake in the student’s success. In the US, in contrast, the teacher sets and marks the exam, making the teacher the judge rather than the partner of the student.
Of course, these ideals are often missed in practice. Oxford has moved towards more small group tutorials, for example. And some American teachers are true mentors to students. But two points are useful to make.
First, the idea of a more personalized tutorial education being enabled by computer-based systems pre-dated the technology that could deliver it. Over two decades since Bork wrote his chapter for us, the technology, such as in generative AI, is approaching the level required to deliver on his model of tutorial learning.
Secondly, personalization is only part of the tutorial model. Equally if not more important is the interpersonal relationships of the tutor and the student. Tutors want to help the student pass a tuff examination process. Students want to demonstrate their progress week by week, reading by reading, to their tutor, and to the exam board. AI is unlikely to approach these very human aspects of the tutorial model, even as it enables more personalized instruction.
I put these points together to suggest that AI will support more personalized education, but as a complement to educators. Perhaps AI will enable more educators to move away from the Sage model (more easily delivered online) to be the coach, guide, supporter of the student’s learning process. In this model, teaching will require smart, well educated, and empathetic teachers to work more one-to-one with their students. This does not mean that students will not be able to increasingly learn on their own, such as by informally using search, chatbots, social media, and books, but their formal education in institutional settings could be more productive and successful with the support of AI. That said, the ideas of educators preceded the technical advances of AI for education.
Bork, Alfred M. (2001), ‘Tutorial Learning for the New Century’, Journal of Science, Education, and Technology, 10 (1): 57-71.
Bork, Alfred M. (2002), ‘Distance Learning through Highly-Interactive Tutorials’, pp. 128-134 in Dutton and Loader (2002).
Bork, Alfred M. and Gunnarsdottir, S. (2001), Tutorial Learning – Rebuilding Our Educational System. New York: Kluwer Academic Systems.
Devlin, Hannah (2023), Education: AI will end traditional classrooms, expert says’, The Guardian, 7 July: p. 13.
Dutton, W. H., and Loader, B. D. (2002), Digital Academe: The New Media and Institutions of Higher Education and Learning. London and New York: Routledge.