AI Tries to Improve Impossible Odds of Perfect March Madness Bracket, But Still Can't Beat Human Unpredictability
-
March Madness brackets have 1 in 9 quintillion odds of predicting all games correctly. Google and AI models try to improve the odds using data analysis.
-
ChatGPT bot "March Madness Bracket Buster" made predictions, but was busted after the first day when its BYU pick lost.
-
Other AI models also made predictions, with some agreement on outcomes. But models can't account for human unpredictability.
-
A Kaggle machine learning model analyzed 153MB of data and thousands of stats to make its bracket picks, with UConn winning it all.
-
In the end, the human element of sports is what makes it so compelling - the underdogs and upsets that defy the odds.