“This app will revolutionize the education market by bringing in an unprecedented level of customization”
“Using AI, this technology will personalize learning to the most detailed level, allowing learners to construct their own learning that fit their needs”
Some learning applications no longer in market
Checking assumptions
A couple of years ago, I was presenting before the management board of a big education system in Viet Nam about a project idea of personalization: a platform that breaks the learning content to pieces at a micro-level. The project was promised close to a million-dollar in funding. It was not a venture but rather an intrapreneurship within a very well-established school. And at that time, I truly believed that I had a good idea as to how a learning app would work. I carried the same idea with me to Harvard – and just a semester and many conversations later, I realized that was one of the stupidest ideas I have ever had.
Now it is not fair to say that all learning applications are like that, but most learning applications, especially those designed to meet learners at a scale, all have this similar trait: using mass-generated content and allowing learners to self-select their own learning journey, with or without assistance, and naturally follow through until the end. And the assumptions behind this design are scientifically valid, for quite some time.
1. Learner differences
Whether you subscribe to classical theories of Jean Piaget (Constructivism) or Seymour Papert (who popularized Constructionism) or if you are a fan of HGSE’s Professor Howard Gardner (author of Multiple Intelligences), we can all see that every learner is different because of their varying contexts and learning styles. This has been the main argument against a standardized approach to curriculum and instructions. So logically, more pathways, more choices, more whatever will allow more accommodation.
True, but not quite: People learn differently: but the degree of differences is so vast that even algorithms could not account for it. Artificial intelligence (AI) itself is a learner, and whatever the data we teach it, it spits out the pattern that it has previously learned, which is predetermined, and therefore, limited. When a personalized learning application is designed, it accounted for the majority of the learning styles and approaches that work for the majority of people – and the input of those who were fed into the learning algorithms, most of the time, are those who have already had their learning styles deconstructed and operationalized by literature and data source, leaving out those who had no access, and no digital footprints out in the cold. And remember, that human cognition and learning science is a rather new field to science (the late 50s and early 60s) and between then and now, the human minds continue to evolve and there are new insights coming into play every day about how the human minds work.
So what? In addition to building a mass content platform, learning designers should take time and continuously learn about how people learn. And to quote Professor Christopher Dede from my favorite class about technology implementation. EdTech designers should not consider themselves in an education business, but a rather life-long learning business
EdTech designers should not consider themselves in an education business, but a rather life-long learning business
Chris Dede – Implementation of Education Technologies
2. Learner autonomy
The second assumption to this is that the learners, when introduced with relevant and meaningful learning content, will take ownership of their learning and advance their cognitive journey efficiently.
False: From my experience as a former school vice principal, among students who struggled with their learning, very few actually had cognitive problems – most learners’ problems, especially young learners, come from outside the classroom. I often giggled when someone discuss the possibilities of machines replacing teachers someday. But that is based on the assumption that what teachers only do is teach. The very term “teacher” does its profession little justice as any competent school administrators or education leaders know a fact: teachers, especially great ones, have little to do with teaching but how well they understand and connect with the students. And unless machines someday magically understand how to solve problems outside of the learner environments: family, friendships, relationships issues, only then should we begin talking about replacing those whose only job is to teach.
So what: I have no doubt that someday those who only teach will certainly be replaced. Teachers who exclusively rely on their disciplinary knowledge and not the knowledge of how well they understand the students, will be the first to walk out of the schools. But until then, EdTech designers need to look deeper into the learners and their learning journeys and identify the factors impacting students beyond the cognitive realm. A good example of this is how we have progressed from the mass instructions mode to computer instructed tutor and now, a more collaborative model.
And to play the devil’s advocate, algorithmic personalization has actually and historically been rather efficient in dealing with social issues taking place among family members, romantic relationships, and peers, evidenced by the vast data we are contributing every day to Meta and Twitter. The only problem is, they seem to be heading in the wrong direction. The open question for learning designers is: whether the data concerning learners’ problems can be leveraged and solved by algorithms? And will this happen in our lifetime?
References
Chris Dede and John Richards, eds., The 60-Year Curriculum: New Models for Lifelong Learning in the Digital Economy (New York: Routledge, 2020); Rovy Branon, “Learning for a Lifetime,” Inside Higher Ed, November 16, 2018.
Hobbs, R. (2020, February 28). Propaganda in an age of algorithmic personalization: Expanding literacy research and practice. Reading Research Quarterly
Justin Reich (2020), Three genres of Learning at Scale. Failure to Disrupt: Why Technology Alone Can’t Transform Education. Harvard University Press, MA