It’s my favorite time of the year, March Madness, where I get to put my ChatGPT Bracket AI up to the task to become the ultimate bracket analyst to help me answer questions like who wins in a 3 to 4 seed matchup?.
Updates of my March Madness Bracket Buster Bot had me searching the web for team statistics. I prepared the data set the same way I would in my day to day job preparing analytics. With over 20 stats in hand for each team, I told GPT to ignore seeding. This was a fun experiment highlighting the incredible opportunity and progress we have made in the last 12 months using AI to synthesize data
Resulting AI Picked Bracket
I have no complaints complaints for this year’s winner given my love for Duke basketball! What I found going through matchup by matchup through the bracket is the team statistics drive decisions that correlated to seeding. Unfortunately while my bracket buster will flag “bracket buster alerts” it ultimately picked a boring bracket, it does highlight some of the opportunities and challenges of using AI to write code and make predictions.
