The New England Patriots’ win over the Atlanta Falcons was nothing short of amazing. The Pats rallied back from a 25-point deficit to tie the game in the final minutes of regulation and secured the win with a decisive touchdown drive in overtime. You may still be reeling from the comeback, but here’s something else that will blow your mind: Even before the first ball was snapped, an artificial intelligence platform accurately predicted the outcome of the game, right down to the 34-28 win by the Patriots.
Created by Unanimous, Swarm AI is a prediction engine that combines swarming algorithms with human input. The company’s AI software allows real, live human users to gather in artificial swarms, The software monitored the conversations in these swarms and collected group intelligence data that is used to make predictions. The company has a string of success accurately predicting the top four winning horses in a recent Kentucky Derby, the last two Stanley Cup winners, and nine out of 10 NFL playoff games.
Being able to predict the final score of the Super Bowl is not an easy task, though. Of the 1,641 Super Bowl final score predictions published by Scripps Howard over the past 19 years, only two have been correct regarding Super Bowl final score predictions. For the Super Bowl, Unanimous took on this challenge by creating wars with 40 football fans and connected them online. The platform then used the swarm’s group intelligence to make its own predictions about how many points each team would score.
So when the Patriots ended up winning the Super Bowl with the predicted 34-28 score, the folks at Unanimous (and everywhere else) were blown away by this uncanny prediction. Now that the Super Bowl is out of the way, Unanimous is eyeing the NHL’s Stanley Cup and the NCAA March Madness tournament.
- A.I. perfectly predicted last year’s Super Bowl score. What happens to betting?
- Spoiler alert: The Patriots win. ‘Madden NFL 18’ already called Super Bowl 52
- Science fiction’s 5 most haunting A.I. villains, ranked
- Deep learning vs. machine learning: what's the difference between the two?
- The History of the Madden Curse