History was made last night in the 2018 NCAA tournament when, for the first time ever in March Madness, a #1 seed fell to a #16 seed. The University of Maryland Baltimore County Retrievers’ win over the University of Virginia Cavaliers will bust brackets across the nation, but there were 31 other games played in Round 1. The end of the two most fun days of the college basketball season is the best time to compare the bracket optimized by Swarm AI® against the “People’s Bracket” on ESPN.
Who is smarter, 40 fans using Swarm AI or 17 million people voting?
After Round 1, the NCAA Swarm is dominating the People’s Bracket on ESPN. And it will be difficult for the People to catch up to the Swarm, since the massive online poll predicted that UVa will eventually win the national championship. After 32 games, the NCAA Swarm’s bracket is sitting the top 8% of all brackets on ESPN, a remarkable amplification of intelligence over the 50th percentile that would be expected from any average bracket. In other words, by working together as a Swarm AI system, a group of 40 fans elevated their performance from average to elite, and they did so in one of the most unpredictable Round 1’s on record.
In comparison, the bracket created by averaging the votes of 17 million fans on ESPN.com was much more pedestrian in its performance. After Round 1, the “Wisdom of the Crowds” bracket is sitting in the 61st percentile, just barely ahead of the average person in the group.
Traditionally, the “Wisdom of the Crowds” has been understood as a way to amplify the intelligence of a group above the performance of its average individual. And after Round 1 of the NCAA basketball tournament, a slight amplification of intelligence is on display, as the 17 million person poll did turn expected 50th percentile results into a 61st percentile performance. But, what does it say about Swarm AI if 40 people using this new technology can amplify intelligence into the 92nd percentile?
There is a reason that bees don’t take surveys or vote on where to put their hive. Nature has found that the most efficient way to amplify intelligence is through the creation of real-time intelligent systems, not polls or surveys. This process, known as Swarm Intelligence, is the model for Swarm AI, which empowers humans to work together as intelligent systems combined with AI algorithms. The phenomenon that powers nature’s “brain of brains” so efficiently in hives and schools and swarms is the same one that just allowed 40 average fans to dominate 17 million people on one of the most challenging opening rounds in March Madness history.
These results echo the findings of a study Unanimous performed with Oxford University which compared the ability of a swarm of football fans to predict the outcome of a set of Super Bowl proposition bets against a crowd voting on those same predictions. The full results of that study are available here: Crowds vs Swarms, a Comparison of Intelligence.
Every week, Unanimous AI publishes predictions made by its Swarm AI platform. Want to have those picks delivered to your inbox?
Want to learn more about our Swarm AI platform? Check out our TED talk…