Researchers in Germany have created a machine learning tool that has predicted the winner of the ongoing FIFA World Cup. This tool says that Spain has a higher chance at the outset but if the Germans make it to the quarter-finals, the odds tilt in their favour.
The creation of this tool does not come as a surprise. Machine learning tools thrive at making predictions and this is just another example of a technology that has become adept at doing a task far better than humans.
However, impressive as a technology might be, there is always room to ask if it should be applied in a particular field. These were my thoughts as I read the article. It goes without saying that better predictions would be beneficial in a lot of fields, including ones like medicine and weather. But would they add value to a sport like football?
I must admit I pose this question from a philosophical bent of mind. Isn’t part of the thrill of a sport its inherent unpredictability, of the unexpected happening? Bookmakers and gamblers might beg to disagree but there is a reason why seeing an unfancied team win against all odds is deeply satisfying. Is it possible then that the invention of a machine learning tool that accurately predicts the result of every match might lessen the enjoyment of the game itself? Worse, can it contribute to a self-fulfilling prophecy where players contrive to fit the results of the prediction (of course, this means having a particularly low opinion of the free will of the human beings involved)? I do not have any concrete answers at present but this is a line of questioning worth pursuing, both for this particular application and for machine learning in general.