Artificial intelligence, data mining, business intelligence; they promise a lot and can deliver stunning outcomes but not always the results we’re looking for.
An old colleague of mine from Microsoft (who will know who he is if he reads this) created some AI to predict horse racing results (ah, the holy grail). The training was based on historic almanac data which he painstakingly input into the engine. The concept was he could use future data such as the horses in a race, the going, etc. and the engine would then predict which horse would be most likely to win and which horses would place second and third.
However, the results were surprising when the system was run. Rather than predict the winning order for an upcoming race, it predicted which other horses any particular horse would prefer to run against. In effect, the world’s first horse dating agency.
So the flawed project was shelved which is a shame in my opinion. A lot of work went into the system and it could have been reconfigured. Also, with the success of dating sites such as match.com and eHarmony, a new market could have been created for animal dating. Although it is quite hard to swipe right with a paw.