The rise of Intelligent adaptive learning (IAL) systems in recent years is evident: A search of the term “adaptive learning” in the Chronicle of Higher Education results in 1760 articles. Adaptive learning systems personalize and tailor content to individual students’ needs based on performance and diagnostic information captured through the each student’s interaction with the platform. Notions of adaptive learning can be traced back as early as the 1920s, when Sidney Pressey patented the idea of an automated tutor and designed several machines that test intelligence (B.F. Skinner, Teaching Machines, Science, Vol. 128 ,1958). In 1958, in an article entitled “Teaching Machines,” B.F. Skinner wrote of teaching machines that could “permit each student to proceed at his own pace” (B.F. Skinner, 1958). Skinner envisioned teaching machines as antidotes to the passive learning encouraged by audio-visual aids that he felt merely present material rather than actively involving students. He conceived of teaching machines that could encourage active learning by providing immediate feedback and supporting each student’s pace and proficiency level (B.F. Skinner, 1958). In the 1970s, the rise of the artificial intelligence movement foresaw the realization of Skinner’s vision, but cost prohibition stalled the development of IAL platforms.
In recent years, two types of IAL platforms have emerged, namely adaptive hypermedia, which is likened to smart search (online libraries or amazon.ca are two examples), and intelligent tutoring systems, which can be likened to personalized online tutoring (Brusilovsky, 2001). Intelligent tutoring systems, or intelligent adaptive learning, align difficulty and content to student’s learning goals and performances (Wauters, 2010, 550).
One of the dominant theories underlying adaptive learning platforms is Item Response Theory (IRT), which enables platforms to generate personalized content based not only on responses to questions but also on student’s response speed, which questions they skipped, and in which order they answer the questions (Kenny & Pahl, 2009).
According to Ohene-Djan and Sen, “Adaptive learning can be used to support users in learning tasks by tailoring content and presentation, customizing user interfaces and providing additional guidance and support. (IAL platforms) have proven useful in learning tasks where users have differing learning requirements, disabilities, histories and preferences.” (Ohen-Djan/Sen,,2007, 1). Adaptive learning reflects the values of inclusive education and user-centered learning by continually generating individualized learning paths, with scaffolding media, in real-time.
According to Brown and Green, instructional design is a process, discipline, science, and a practice (Brown & Green, 2011, p. 7). When considering instructional design in relation to adaptive learning platforms, it may be most useful to consider the process definition of instructional design, which Brown and Green define as the “systematic development of instructional specifications using learning and instructional theory to ensure the quality of instruction. It is the entire process of analysis of learning needs and goals and the development of a delivery system to meet those needs” (Brown & Green, 2011, p. 5). With adaptive learning, one of the underlying assumptions is that distinct students have distinct needs and adaptive learning systems should provide a multiplicity of personalized trajectories that anticipate and meet the unique needs of each student. Intelligent adaptive learning (IAL) systems use a modular instructional design that allows curriculum to be chosen and presented based on a student’s results from previous modules. The more data collected per module, the more modules and student encounters, and the number of possible branches extending from each module control the number of unique learning paths that are possible within a particular IAL system. Instructional designers who work on adaptive learning systems must have a deep understanding of the students for which the adaptive learning system is built so as to anticipate a wide variety of proficiency levels and student responses. Adaptive learning systems reflect the underlying educational values of individualized learning, efficient learning, transparency and performance tracking, arguably to the exclusion of collaborative learning, privileging of the inner life world and creative process, and privacy.
Does it matter that we can’t measure the effects of educational technologies and do we collectively agree that the effects are universally positive? Companies such as Knewton claim that they can, in fact, measure the effects of their platforms on student learning, stating that the “infrastructure platform can pinpoint student proficiency measurement, content efficacy measurement, student engagement optimization, and concept-level analytics.” The broader question we would like to pose is whether the student interaction and performance data captured by platforms such as those of Dreambox and Knewton can reflect deep, meaningful learning, which can be defined, with an appeal to Revised Bloom’s Taxonomy, as the ability to apply, evaluate, and analyze concepts and far beyond the end of the 14-week semester or academic term.
In the following pages, we present two adaptive learning platforms Dreambox Learning and Knewton Adaptive Learning, and describe two use cases, one from each company, after which we present critical questions for discussions. The questions posed focus on the design of adaptive learning environments, and yet in keeping with an intersectional, multidisciplinary focus, we also expand the scope of our questions to include ethical and sociocultural questions.