Last night I was talking with a good friend who works in a company that brings AI to educational products. One of the challenges they have is that AI became just a buzz word education companies want to use for marketing purposes but they don’t really care for the potential it can unlock in the learning of a child.
If we want to use AI to optimize the multiplication table learning process, there is only so much we can optimize for. Maybe we can make kids learn it a bit earlier, a bit faster or make them memorize all the way to the 20 x 20 table. We could track how much they are spending in each part of the process, learn what gestures expert kids do and compare them with those kids that take longer. But really who cares about that? Humans will never do it as fast and reliably as a computer.
What about using AI to give more agency to children or use it to help them achieve projects they couldn’t do otherwise? Why don’t we see more education companies exploring those options? The answer is that the money still is in getting children's scores higher on standardized tests. Important educational decisions are still based on those scores.
Here’s an idea I’m exploring. What if we could use AI to get information about children’s projects. So instead of using standardized tests to measure learning, we can use AI to gain a better understanding of what a child can actually do with all the knowledge gained at school. We don’t need to know how well a child can retrieve the algorithm for long multiplication. We need to know if children understand how to use it for real situations they care about, not for a hypothetical imposed test question.
Let me give you a real example: building a projector in the laser cutter. The children in our school, Portfolio School, had the problem of having to build something which will allow them to project colorful images on the wall. They decided to build a projector.
If we see the final artifact, the projector, it’s hard to get an idea of all the learning the children applied to create such a thing. It requires asking the children a lot of questions, reverse engineering the artifact, checking documentation, and more. If a person has several of these projects to analyze, the task is impossible.
This is a very brief description of the process of building the projector. The first part was to laser cut a box that was transformed into the core of the projector (first image on the left). Second step was to think of a mechanism that will project images. The idea was to get a circular shape around the core of the projector with the images. The children measured the length of the box and added an inch. That measurement was the diameter of the circular piece. Then using the perimeter formula: C=2πr and the algorithm of long multiplication they were able to calculate the length of that piece.
As you can see, the multiplication algorithm is just 1 component of the overall project. If I hadn't explain how the children made the project, it would be hard to see they needed to use multiplication. How many other forms of knowledge were involved in doing this project, and are also not obvious to the unaided eye?
What if AI could help teachers, evaluators and universities to analyze children’s projects. What if AI could also give that feedback directly to the children. In the next months I’ll be exploring the idea . Stay tuned!