AI & Sports Predictions: Limitations, Benefits & Reasons to be Wary
In 2015, long before anyone had heard terms like OpenAI, ChatGPT, or Google Gemini, IBM launched its first foray into sports predictions backed by its AI supercomputer, Watson. Over the last decade, the program has expanded, partnering with fantasy sports providers and widening the AI capabilities so that it is no longer the supercomputer, Watson, but IBM’s AI and data platform, Watson (note how the lowercase spelling as branding is fashionable in the 2020s).
You can be sure that most fantasy sports players have heard of Watsonx. It has had mixed results over the last decade, with some fantasy sports players praising its analytical insight. Yet, the truth is that if it were highly effective at predicting the outcome in sports, a lot more people would be using it. Again, that’s not to dismiss what IMB has achieved, nor is it to ignore the likelihood that Watsonx, and AI in general, will improve in sports predictions. Yet, it remains the case that we just aren’t there yet.
AI & Sports Predictions Have a Future
Of course, that isn’t to say that AI can’t benefit sports predictions. You can benefit from feeding ChatGPT some data before prompting the chatbot to make picks for NFL player props or MLB run lines. But what we are getting at is the perception of AI as some vast repository of knowledge with a dash of clairvoyance that will suddenly spew out accurate sports predictions is wide of the mark.
Indeed, right now, as we approach the start of the 2025 NFL season, you will undoubtedly see lots of articles titled “AI predicts the Super Bowl winners” or “AI predicts the win-loss totals for every team.” There’s nothing wrong with these articles, although they do feel a bit like “filler,” but, notably, they rarely deviate from the betting markets, power rankings, and ‘human experts’ analysis on the season ahead.
And that, in a sense, is the point. The allure of AI and sports predictions is that we believe it will find something unseen, a “eureka” moment not visible to the sportsbooks and the experts. Yet, in reality, what AI is doing is summarizing the information that is already available to you. As we said, it can be helpful, but it is limited.
None of this means you should not be using AI for fantasy sports and betting. But it is perhaps essential to think about it differently. For instance, it’s well understood that AI is only as good as the data it can access. Something like ChatGPT does not have access to all the advanced statistics that sportsbooks use to set odds, for example. ChatGPT can search the web, but there are limits to what it can access. And you can be sure that advanced data providers like Betradar guard their data jealously. It is their livelihood after all.
Be Wary of Those Selling Tips and Systems
Yet, it does follow that if you train AI on valuable data sets, perhaps those you have acquired from a subscription to a professional sports data provider, you may be able to work with the AI to catch anomalies, spot value in the odds and fantasy markets, and so on. There are still limits to what AI can do – sports will never be 100% predictable – but it could be helpful to all the same.
We will end with a warning, though. Social media is rife with adverts and accounts purporting to offer AI tips and betting strategies. You should be highly skeptical, especially of those claiming to have foolproof systems to win. Think about it: if you somehow learned how to use AI to beat the bookies, would you be flogging subscriptions for ten bucks a month on Facebook?
