While there are many things that fill me with wonder, the first that comes to mind is artificial intelligence (AI). I distinctly remember being a high school senior, watching a video of an AI that learned to play Super Mario World in only 24 hours, using remarkably simple code. This relatively new system, called Neuroevolution of Augmenting Topologies, or NEAT, is modeled after the human brain. Immediately fascinated by this idea, I began trying to find out more about NEAT, watching videos and reading research papers to learn how it worked. After I felt like I had a firm grasp of the concept, I tried finding a program online that would allow me to try it for myself on my computer. None of the programs I found online worked, however, so I began working on my own implementation. I have spent nearly three years building and refining this program.
My wonder was indescribable when I first saw the program that I had built from the ground up learning to solve problems. My amazement only grew when I told the program to learn to balance a ball on a board. In only a few seconds, the program was finished learning, and I gaped open-mouthed at my computer as I watched a simulation of the ball, perfectly balanced. Not satisfied with simply knowing that the program had taught itself how to achieve a given task, I wrote another program that would show visually all the nodes and connections that make the AI work. I was immediately struck by a beautifully complex system of twisting and weaving nodes and connections. I realized that even though I could see the “brain” of my creation, I could never hope to understand how it worked. In that moment, I had been surpassed by the thing I created: I knew how each individual part worked, but the combination of those parts was far too complex for any human to understand. Now, three years after I began my project, I am still far from finished. I will keep pushing forward, striving for a program that will solve harder problems, learn faster, and perform better. Artificial intelligence will never cease to be a source of curiosity and wonder for me.
Grason Humphrey, North Carolina
Student
My wonder was indescribable when I first saw the program that I had built from the ground up learning to solve problems. My amazement only grew when I told the program to learn to balance a ball on a board. In only a few seconds, the program was finished learning, and I gaped open-mouthed at my computer as I watched a simulation of the ball, perfectly balanced. Not satisfied with simply knowing that the program had taught itself how to achieve a given task, I wrote another program that would show visually all the nodes and connections that make the AI work. I was immediately struck by a beautifully complex system of twisting and weaving nodes and connections. I realized that even though I could see the “brain” of my creation, I could never hope to understand how it worked. In that moment, I had been surpassed by the thing I created: I knew how each individual part worked, but the combination of those parts was far too complex for any human to understand. Now, three years after I began my project, I am still far from finished. I will keep pushing forward, striving for a program that will solve harder problems, learn faster, and perform better. Artificial intelligence will never cease to be a source of curiosity and wonder for me.
Grason Humphrey, North Carolina
Student