Artificial intelligence can be hard to understand, but with the right breakdowns and visualizations, it doesn’t have to be.
As a part of my COGS 260 class, I combined the inspiration of a well designed textbook, online platforms as a method for teaching and learning, and my love for computer science to create a tool that would help people understand the mechanics of artificial intelligence. Artificial intelligence is made up of mathematical algorithms that can be hard to understand, but just like learning any complex subject, having right breakdowns and visualizations can help make a complex subject manageable.
I started by breaking down one specific aspect of artificial intelligence known as the perceptron. In AI, the perceptron is one of the fundamental building blocks for more advanced concepts such as neural networks. While it sounds complex, the perceptron is simply a type of algorithm that finds a line that separates data with a certain label from data with a different label.
Algorithms such as the perceptron are often expressed through linear algebra functions such as the dot product, which can be hard to intuitively grasp. Just like a graph and animation can make the equation of the trajectory of a basketball shot much more intuitive, charting out and being able to interact with the functions of a machine learning algorithm can make the mathematical manipulations more intuitive.
With my learning tool, students are able to manipulate lines, data points, and vectors on a plane to see how the algorithm behaves. By having the ability to interact with the algorithm, students are better able to intuitively identify where their knowledge is not completely correct, and metacognitively assess how to learn.