AI Journalist Helps You Talk Trash During March Madness
Sure, you claim that your team is better than everyone else's, but wouldn't you like to have the hard numbers to back that up? Enter the StatSheet website and its new StatSmack application. With an enormous database of sports statistics (over a billion and counting), Stat Smack can find the numbers that show how any team is superior to another. A few selections in a drop down window, a click of a button, and you'll get a detailed list of facts that show why your team would dominate your friend's team if the two ever met for a game. The application lets you quickly share these lists via Twitter, Facebook, or email with the idiots who dare deny the glory of your franchise. Beware, though, StatSmack works both ways, a single click is all it takes to see the stats that explain why your team is worse than the one it just smacked. Turnabout is fun, isn't it? StatSheet is debuting StatSmack in time for March Madness, and currently allows you to compare any NCAA Division I men's basketball teams. That's 2.5 million different matchups for you to enjoy. While it's all fun and games, this little stat compiling application is a sign of how narrow artificial intelligence will be able to gather large stores of relevant data on any topic...and then distort it to suit our ends.
For those who missed our previous coverage of StatSheet, the website has made a name for itself by applying narrow AI to its ever growing database of sports statistics. This allows StatSheet to do things like generate written coverage on any game played in college basketball without the need for a human journalist. Robbie Allen, founder and CEO, tells me they produce over 18,000 automated articles of different varieties every month! They can also create websites for every team, no matter how obscure, and fill that site with enough relevant facts and figures to satisfy the die-hard fan. You may then take those graphics and embed them in your own site if you like. While they've focused on college basketball, StatSheet plans on expanding into every professional and college sport, and you can already go to their network and find AI and human generated content on college football, HS basketball, NASCAR and the NBA.
StatSmack is just another fun tool that's arisen from StatSheet's mastery of stats and sports. The application compiles data on two teams, and then displays those facts it's determined reflects positively on your favorite. Not an especially grueling task, computationally speaking, but I'm sure it's going to be a hit as March Madness fans will want to find new ammunition in their trash talking bouts. Even as only a passing afficinado of college basketball I must admit it's fun to read stats on important rivalries (Duke vs UNC), teams I've rooted for in the past (Texas vs. Harvard), or why the underdog might actually succeed (how is Hampton better than Ohio State?).
What really piques my interest in this concept, however, is that it's a way in which computer compiled data can be skewed to represent any view point. Do you want to know how a tiny, unranked team could defeat one of the giants of the game? StatSmack will give you any stats it can find that support that insane view point. If we extrapolate this to areas outside of sports, things get a little scary. In almost every industry we are faced with an overload of data. Warfare, global health, medicine - we have more information than any human could hope to review. We will need to turn to computers to process these piles of data, and many fields are already reliant upon narrow AI to help them sort through the dross and find the tidbits we really need. Stat Smack is a clear example of how that dependence could be a danger if we're not careful. With enough information, an AI can select data to support almost any conclusion you want to draw. That's a fun game when talking about college basketball, but it would be an horrific nightmare if we applied it to military intelligence, cancer diagnosis, or bank loans. As we move forward with AI processing of our vast stores of data, we need to be very careful to guarantee that these programs are helping us find the truth, rather than simply finding the best way to lie to ourselves.
StatSmack, you're a fun way to augment my trash talking skills, but you're also a scary foreshadowing of things to come. ...Yay technology?
[image credits: StatSheet]