I was reading this thread here about how AI works, and I thought, as i do, about applying the question to sports generally, and to baseball in particular.
Seems to me AI has many uses in baseball which we can expect to be exploited shortly. One arbitrary example would be using AI to predict pitcher injuries. As the other thread explains, AI is able to diagnose way in advance when people will come down with life-threatening problems by scrutinizing thousands of data bits about their lives, lifestyles, family history, genetics, environmental exposure and literally hundreds of other factors we never thought to apply to the problem, so many that AI cannot explain how it reaches their diagnoses, only that it reaches them and that they’re correct.
So it seems to me that AI can use, say, pitchers’ records and hundreds of other factors in their lives to solve some of the seemingly unsolvable mysteries, like why some pitchers develop career-ending injuries. Ai should be able to review the career of every pitcher up to the age of 27, about when they start to enter free-agency, and tell which ones will develop mysterious arm issues and which ones won’t, which will change the way teams evaluate their young pitchers (and players in general of course–this is just one small example).
It’s exciting, in a way, to have knowledge that we have never had before, but it’s also depressing. The first team to use AI successfully in this way will have tremendous advantages over its competitors, but eventually this knowledge will spread throughout MLB and the game will change, and not for the better, to my mind.
This is likely true, especially considering the amount of biometric data that is being collected on players these days. I was listening to a podcast that was discussing pitching mechanics, and how neck pivoting range of all things was being used to evaluate the potential of pitchers to take the next step.
The next collective bargaining agreement battle is likely going to be over some major issues regarding the collection of this information, who owns it, who has access to it, and what sort of compensation is going to be needed to collect/save/analyze it.
Speaking of AI in sports, I’ve seen ads during football for AWS talking about how long it took to create an NFL schedule each season and now they can do a bazillion permutations in no time to get the perfect schedule.
While I’m sure AI is a useful tool for creating the NFL schedule, was it really that difficult before? I mean, each team in a division plays each other twice, home and away, and then one game against each team in another division that changes every year, and that’s about 90% of each team’s schedule every year, right? Throw in a couple more games against random teams and you’re done. Seems like something I could do in a weekend if I didn’t have anything better to do.
It’s more like trying to make a schedule for a fast food restaurant where all your employees have dates and times they need off. It’s not hard to make a schedule you have to work it around planning for TV, making sure teams that are expected to be good and those that draw ratings get the late afternoon, Sunday night or maybe the Monday night game. You’d also have to keep in mind the concerts, college football games and other special events that each stadium fills their days with. You’ve also got an increasingly large number of overseas and foreign games to keep in mind.
It’s not random teams. Back in the 16 game schedule, the final two teams were those teams in the same position (1 through 4) last year in the same conference but not your division nor the over division you play all of.
The 17th game now is played against a team in the opposite conference in one of the 3 divisions you’re not playing and based somehow on divisional rankings from the previous year.
The AWS schedule, they say, takes into account predictions about injuries and such. Something probabilistic like that is going to be significantly more intensive than the old way.