Adaptive Learning

Adaptive learning

Adaptive learning today is increasingly contextualized by various consumer media technologies, and more recently the adaptive learning technologies. With the advent of these new technologies and the development of authoring software that facilitate adaptive learning, instruction can now be orchestrated to tailor to individual learner's strengths, weaknesses, goals, background, beliefs, learning style and engagement patterns.

It has often been demonstrated that adaptive learning results in better learning outcomes and better engagement patterns. Learning is personalized, targeted to specific needs, and every single learning moment is optimized at scale to support mastery-learning. Thus in this learner-centered model, learning as a whole becomes a satisfying process for the learner.

The field of adaptive learning draws ideas upon knowledge domains such as machine learning, cognitive science, predictive analytics, and science of learning. Adaptive learning is underpinned by the logic that content has to be backed and synchronized by data in an an adaptive system. Ongoing research and collaboration in learning and brain science is utilized to maximize the full potential of adaptive learning. Adaptive learning technology inexpensively scales the benefits of 1-on-1 mentorship, providing each learner with their own personalized learning journey, which continually adjusts in real-time to provide the right level of challenge. Again, this provides context for content, and this can lead to mastery in a personalized way in the most efficient and optimal way possible.

Adaptive learning and Role of IDs

Given this, it is imperative that IDs are agile and adept at updating themselves in a rapidly evolving world of educational technology in order to apply best practices to the design of an adaptive learning projects. Since adaptive learning can change the way educational content is authored – not just how it’s consumed in classrooms, this development of new technologies require IDs to work with those creating the technology.

It is likely that with the advances in Artificial intelligence, IDs may experience an upskilling-scenario where IDs are required to extend their skill-sets. These skill-sets can range from understanding the details of how particular software works to applying their expertise/knowledge to unravel the complexity of adult learning while incorporating adaptive learning and while developing new technology to facilitate adaptive learning .

Discussion in a scenario

Let me give an example of how my signature assignment would look if adaptive learning were incorporated. Firstly, while authoring the software, I would need to rely on assessment algorithms to detect mastery and non-mastery of learning items to adjust learning paths for each learner. I suspect ongoing assessment of learning would drive the direction and pace of learning. Having said that, I may still 'force' all learners to complete certain tasks that require reflection, peer feedback/collaboration.

Such a conceptualization raises an important question about how fair the assessment algorithm is to all learners? Well, although how learning is being designed and delivered has been central to a lot of discussions, the underlying assumption of a roughly homogeneous group of learners has not been heavily challenged in many of our course designs. To the degree that the 'diversity of learners' is acknowledged/considered in course designs, we were faced with the limitations of technology despite our best intentions. Hence, adaptive learning, I feel, should be able to address the perceived difference of learners to a greater degree compared to traditional course designs and delivery.

Clearly, assessment is the engine that propels adaptive learning forward. Hence, assessment practices need to be informed by insights from fields such as education, psychology and cognitive science. At the moment, I feel that assessments in adaptive learning are mostly data-driven. I think here lies the irony. To what extent can the learning be touted as being customized to personal preferences, beliefs, experiences etc when the underlying data (read 'assessment') does not conceptualize a learner as an individual? I guess some work needs to be done here, and can the answer be anything other than 'collecting more data'?

Now to the question of algorithms being fair to all learners: I think, as adaptive learning becomes more mainstream, I think the underlying algorithm will be further questioned. In my signature assessment, I may begin with an introduction of the course materials followed by a quick assessment. The performance in this assessment will determine the next steps for learners , as informed by underlying algorithm which takes into account mastery/non-mastery of current topic (and, if available, a bank of data about the learner about his/her previous learning, background, preferences etc). Is the assessment algorithm fair to all - maybe not. Perhaps I can take solace in the fact that such an algorithm may disproportionally benefit those who would generally struggle learning in traditional designs.