My most impactful programming experience has been a MATLAB-integrated research project for a class titled EK125 ("Introduction to Programming for Engineers"). The project was group-based, and I was responsible for the entirety of the coding.
I've also explored the uses of Artificial Intelligence through a project about the creative problem-solving abilities of ChatGPT.
Studying the Global Correlation Between Income and Obesity
Below are some figures from the EK125 project.
Figure 1: Sample of Original Data for Obesity
Figure 3: Code Scrubbing the Data
Figure 9: Improved Graphs for Low and High HDI and Comparison Graph
Figure 10: Examining the HDI Data Set and Plotting in Decreasing Order
Access the full project by clicking on the PDF logo below.
Applications of Artificial Intelligence in Creative Problem Solving:
Testing ChatGPT's Ability to Solve Riddles
I conducted a study to test ChatGPT's creative problem-solving abilities and presented the findings at the LSC 2023 Conference in Institut des Hautes Études Scientifiques (IHES) in Bures-sur-Yvette, France.
Abstract:
My goal in this project was to determine ChatGPT’s effectiveness at solving riddles. In particular I focused on 100 logical and mathematical riddles, from an obscure old book and scored ChatGPT’s solutions. Overall, it was striking that ChatGPT only solved 28% of the riddles. ChatGPT was better at solving mathematical rather than logical riddles (42% versus 14%, respectively). I further tested whether ChatGPT’s success rate improved when asked to produce code to solve a riddle, as this would allow the AI to algorithmically detect patterns and create a systematic way of finding a solution. In this case, ChatGPT only solved 22% of the riddles. The project illustrates that ChatGPT – though an immensely useful instrument – may at present feature limitations in regard to creative problem solving.
Access the full project by clicking on the PDF logo below.