The idea of intelligent tutors is not new, nor is the technology. From the 1940’s, Alan Turing and other early innovators of computing systems envisioned these intelligent machines would be used to teach humans, with personalized learning being a specific pedagogical affordance of the technology (Ferster, 2014, 2017; Shute, 1994; Woolf, 2010). Historically, development and implementation of intelligent tutoring systems (ITS) has aimed towards mimicking or substituting for what has been considered the “gold standard” in education: one-on-one learner-teacher interaction (Baker, 2016; Ferster, 2014, 2017; Roll & Wylie, 2016; Woolf, 2010). This “gold standard” remains a persistent challenge within distance education (DE) contexts, though recent Internet communications technology (ICT) has helped solve some of these challenges, enabling more direct learner-teacher (and learner-learner) interaction (Garrison, Anderson & Archer, 2010; Simonson, Smaldino, Albright, & Zvacek, 2012; Woods & Baker, 2004). How might ITS build on this progress?
An important question that has shaped inquiry regarding the role of ITS and machine intelligence (MI) in education and learning is, “Can machines teach?”, followed closely by the question “Should machines teach?” (Ferster, 2014, 2017). Challenges in answering these questions may stem from the embedded assumption: “Can (should) machines replace teachers?”. If we ignore this hidden question, and deal with the explicit question alone, the answer is conceptually straightforward: Yes, machines can and do (and, within the 21st century context, should) teach. Consider the question from a basic behaviourist perspective: Working one-to-one with an ITS, a student inputs an answer to the question/problem the ITS generates, and receives immediate feedback from the ITS, which leads the student to modify his/her actions (Ferster, 2014, 2017; Laurillard, 2012) In this scenario, the teacher (and teacher intervention) is not eliminated, but rather repositioned, and how the ITS is implemented will impact to where (i.e. what point in the learner-ITS-teacher interaction cycle (Laurillard, 2012).
The important question then becomes: “HOW can/should machines (ITS) teach?” (Baker, 2016; Ferster, 2014, 2017; Roll & Wylie, 2016; Woolf, 2010). This is the question I will attempt to answer in this paper. I will pay particular attention to how ITS might solve some important problems for DE, while simultaneously challenging the idea of “distance” (historically considered to be any separation of learner and teacher) (Simonson, et al., 2012). For instance, can the separation inherent in learner-ITS-interaction actually mediate “distance” and improve immediacy and teacher presence, if the integrated technology and pedagogy afford teacher interventions that reach each learner and are tailored to their individual needs? (Buckreus, 2017; Garrison, et al., 2010; Rizzotto, 2017; Wolf & Baker, 2004). Widespread integration of ITS across learning environments may render the historic distinctions between face-to-face (f2f) and DE contexts immaterial, with “distance” utilized to enhance learning in both contexts.