March Madness is well underway and some A.I.-driven digital assistants are getting in on the widespread bracket action with predictions of their own. However, it may not be the ones most people would expect to take advantage of the 67 game-long NCAA championship tournament – despite the opportunity this provides to have some fun and gain some publicity. In fact, only Amazon's Alexa and Microsoft's Cortana are taking part at all, as of this writing. Perhaps surprisingly, the most in-depth predictions come from what is arguably the underdog of the digital assistant world.
Microsoft's Cortana will respond to questions about who will win the tournament with not only the overall winner – which it predicts to be The University of Virginia Cavaliers – but also winners of individual games. Leveraging information pooled from the web and Microsoft's Bing, the assistant provides commentary on why it thinks its selected teams will win and a percentage representing how likely the outcome is. The Cavaliers, for example, have a 15-percent chance of seizing victory, according to Cortana. On the other hand, Amazon's Alexa has picked the Villanova University Wildcats to take the final victory of the tournament. That response comes alongside reasoning tied in with the team's strong offensive play style when users ask Alexa "Who will win March Madness? Amazon didn't go too in-depth with its A.I., though and it won't provide more information than that. Moreover, that specific wording needs to be used to elicit a response. Meanwhile, Google Assistant and Apple's Siri are staying out of the predictions entirely.
For Google, that makes some sense since the company is actually hosting its own bracket-filling contest for A.I. iterations via its Kaggle service and in conjunction with the NCAA. So, the company is simply responding to questions about predictions with information about the games themselves. It will also port users out to external sites officially associated with the tournament. Siri, for its part, only goes so far as to provide dates, times, and other information about upcoming games. It goes without saying that these predictions are, despite being made by some of the smartest A.I. on the planet, still just predictions. It would probably be a bad idea to use them to make any kind of wager, as any wins predicted by the various chatbots is likely to hinge at least partially on luck.