“The next industrial revolution has begun”. Those were the words that set in motion one of the most incredible phenomena in the history of the stock market. Nvidia is the toast of Wall Street and artificial intelligence is the code word for the next major epoch in human ingenuity. But this is a baseball blog, and therefore, my question after watching Jensen Huang throw out the first pitch at a Weichuan Dragons game Saturday was, how does AI apply to baseball?
In the first half of 2024, we already hear rumblings about how AI can change baseball forever. Intuitive applications could be in defining game strategy and more deeply analyzing individual player habits and projected trends. AI could well manage a game soon enough, and probably better than any analytics department. Maybe AI could replace a team's whole analytics department altogether. Given today's usage of Rapsodo, Trackman and so on to compile data, visualize it, and analyze it, the logical next step is likely to have AI do the analysis that humans do today. For example, in today’s world, humans write SQL queries and create Python scripts. They write or record macros. They make calculations and create graphs, dashboards, and other visuals. So imagine a world where AI just does this on a GM or coach’s command, giving him in mere instants all the data he could want or need in an easily digestible format that can easily be communicated to players in pre-game meetings.
However, the most interesting prospective application I've seen so far is something totally different. Mud is lathered on MLB baseballs in preparation for their in game usage. But that is not just any mud. The preparation process is secretive and labor intensive. It is also highly manual and error prone, which means different balls could be muddied with varying consistencies leading to subtle differences in how the ball behaves when pitched. This creates unfairness naturally so ideally, every ball must be caked in mud that is perfectly consistent and coated in a perfectly even fashion all over the ball. Imagine a world in which machines could have the right data analysis learned such that they can perfectly perform his task each and every time, and also perform the needful QA to ensure that the process was done correctly, therefore ensuring no pitcher will ever be unduly disadvantaged by a badly muddied ball again.
Another thought is the question of how we project a player’s future performance. Up to now, even with advanced analytics, we are looking towards the future by looking into the past. We are much better at making educated guesses at which players simply got lucky, which were genuinely great, and which are better than what their basic stats suggest them to be, but we could be better still. That is where AI comes in . We’d still need to look into the past of course. The AI will need data from the past to learn from. The law that says we must understand the past to see the future holds and will continue to hold. But what we aim to achieve is refining predictive analytics to the point where they become far more accurate than anything in existence today.
Fantasy baseball players would also be pleased to hear that AI could lead to important advances in gauging the likelihood of player injuries. This has always been a tricky topic and many teams have seen promising seasons completely derailed by the injury bug. My own team is a mess due to so many injuries. The key driver is the biometric data that can already be collected real-time during games and that is already leading to some modeling on injury likelihood. Particularly with the current epidemic of pitcher arm injuries. Data on body movements can help identify patterns in a pitcher’s delivery that could exacerbate injury risk and also give insight into other risk factors, using that data to create more advanced risk scoring systems to gauge how injury prone a player is. On the prevention front, fitness and conditioning for athletes in general is becoming increasingly data driven, from tracking sleep quality, (ask Shohei Ohtani how important sleep is) to tracking things like heart rate and gauging body fatigue. AI is even being looked at to personalize a player’s rest and recovery regimen and even a player’s rehabilitation from injury to ensure a quicker and more complete recovery.
The last key usage I will discuss here in the field of scouting and player development where everything from biometric data to virtual reality training is being used for training a batter’s eye for 100 mph fastball, 3000 rpm sliders, and so forth, all from the comfort of either home or some team facility. Early data on these methods is already revealing improvements in reaction time and decision making ability. Of course, coaching will still need to be a thing so as discussed earlier, AI will help give coaches the info they need to make them more effective at their jobs.
Baseball has always been a numbers game, a scientific game. Logically, AI would make its way to this sport above all the others. That’s not to say AI will not drive innovation in other sports, but the data driven nature of baseball makes it the most fertile ground for bringing AI into the world of sports. Be prepared to stretch your imagination as our game makes its way into its third century of existence. Baseball, and humanity as a whole, are in for an era of fast paced innovation the likes of which history has not yet seen.
Comments