I can understand why some of you will think I’m stubborn traditionalist–a luddite when it comes to technology and the use of advanced statistics applied to sports. I think it’s more accurate to say that I’m a pragmatist–my position on both depends on whether I think either is effective or not. To give you an example, I want to talk about a scenario where I can see AI being an important and valuable tool. My understanding is that one of the things AI is good at is recognizing patterns and maybe even using these patterns to extrapolate to recommend specific decisions or actions. If this is accurate, I think AI could be really useful–perhaps better than human beings–in terms of evaluating players. To give you one example, I think the best way to evaluate a QB’s ball security is to determine if they have any patterns that increase the likelihood of an interception (or turnover). I assume one could feed the AI with thousands of footage of risky throws. Assuming the AI could account for all the variables that would go into making a throw risky, eventually the AI would have concept of what constitutes high-risk passes. If this is true, then it should be able to, at the very least, to tally the number of risky throws by a QB. Hopefully, it could categorize different types of risky throws and maybe weight the level of risk. It’s conceivable to me that an AI could assess ball security better and faster than a human being. And this would be an instance where the AI’s objectivity could be one of the main reasons it would be superior to a human. The same sort of thing could happen with tendencies of an offense or defense; how a team handles things in certain crucial game situations.
Now, all this assumes that the ultimate judgments made by the AI is actually reliable. It’s possible that the AI can’t handle all the variables that go into what makes a pass risky or maybe there’s some other reason the AI makes unreliable conclusions.