Thought Leaders in Online Education: Dror Ben Naim, CEO of Smart Sparrow (Part 3)

Dror Ben Naim: We can base adaptation on many factors. We can base it on how students are doing right now on the problems they are working on. We can adapt in a more complex way not only by looking at what they’re doing now but also by what kind of mastery level they have on the learning objective that they are currently dealing with.

For example, I’m giving you a quadratic problem. If you don’t know how to solve the quadratic problem, I can choose to give you an easier one. If I’ve got statistical information that shows that, overall, you’re doing poorly on quadratic, I may choose to take you down a different path which starts explaining in simpler ways how to solve quadratic. In this example, I talked about adapting the sequence of learning experiences.

In order to adapt the feedback, I need to know exactly what mistake you’re making while you’re solving the quadratic. I need to say, “You got the wrong sign here.” A lot of the time, we also personalize on demographic information or even stuff that learners provide to us. They tell us their interest, age, or whatever makes sense in the context of learning. Based on that, they may be offered different types of experiences.

Sramana Mitra: One question that comes to mind as I’m listening to you is, who determines the flow. If somebody doesn’t have quadratic, what is the next thing that students should be offered? Does your LMS automatically figure out that flow or are you expecting the instructors who are designing that flow to come up with those paths of learning?

Dror Ben Naim: Fantastic question. Remember we said, what adapts and based on what? There are two flavors here. One is what we call algorithmic adaptivity. The other one is what we call designer-led adaptivity or designed adaptivity. Designed adaptivity is easy to understand. You will have a learning designer who will create a lesson. The lesson would have rules in it. It’s one big bunch of if-then rules but easy to use for non-developers.

That is a way to say, “If the student has not mastered the following, do this. If they have mastered, you can go down this route. Sometimes, it’s done using gating rules. The instructional designer sets certain rules that your system then operates on the basis of. That gives them agency, control, and pedagogical ownership. Most of the market appreciates these technologies although a lot of the investment and hype is done on the algorithmic adaptivity.

There’s this interesting phenomena where everybody wants to talk about AI and machine learning beside the actual customers. The customers actually don’t like to hear about black-box machine learning and AI-driven systems that decide for them. They would much rather have decision agency and design agency. The second flavor of algorithmic adaptivity has suitability in different cases. This lends itself well to disciplines and types of experiences where the learner is expected to do a lot of problems of the same type in order to demonstrate mastery. Of course, we’re talking about math. There’s a lot of chemistry and physics.

These types of learning environment are easier to create in such a way that an algorithm will be dynamically providing the learners the next problem item based on their performance. Our platform gives tooling and features that enable courseware developers and designers to create both algorithmically-driven adaptive experiences as well as design-driven adaptive experiences. There’re actually three types because there’s a hybrid mode.

This segment is part 3 in the series : Thought Leaders in Online Education: Dror Ben Naim, CEO of Smart Sparrow
1 2 3 4 5