(NewsNation) — March Madness bracket fillers are tapping in an assist from AI to help predict how things will play out, but the human variable has proven to be pretty unpredictable.
With matches starting on Tuesday, fans are busy filling out their brackets with Alabama, Houston, Purdue and Kansas as this year’s number one seeds.
Mike Randle of the FTN Fantasy network says AI could potentially help decode the statistics, but human error can’t be accounted for.
“There’s a natural fear of mathematics from the public and I think now with AI it’s becoming easier to use, it’s something that can be translated to the average person better and they can comprehend those stats in a quick way,” said Randle. “There are so many variables and things that are going to play out. It’s going to be difficult to get a winner from AI prior to when the tournament starts.
This year, AI picked the Creighton Bluejays, Houston Cougars, Duke Blue Devils and the Kansas Jayhawks for the final four, with the machine picking Houston as the winner of the tournament.
But of course, AI predictions were entirely wrong last year
Last year, the Kansas Jayhawks claimed the championship after one AI model predicted they wouldn’t even make the final four. Instead, AI chose Gonzaga Bulldogs, Kentucky Wildcats, Arizona Wildcats and Auburn Tigers.
In reality, the final four came down to the Kansas Jayhawks, Villanova Wildcats, North Carolina Tar Heels and the Duke Blue Devils.
“You can consider all kinds of things right decent season statistics, what conference you’re in, how many injuries you have. The fact of the matter is, at the end of the day, you won’t be able to predict everything, which is what makes it a very difficult event to try to use AI to isolate like a single winner,” said Matt McGorrey, account executive at Atrium AI.
“It wouldn’t surprise me if we ended up you know, they find 10 years down the road. And we’re still wondering how to kind of like you know, accurately predict the tournament outcomes. And we got a lot of these other problems pretty well solved, right.”