September is fast approaching, and with it comes the 2025/26 NFL season kick-off. It’s an exciting time for the fans, with renewed optimism and hope guiding them into the opening weekend.
Of course, it’s also boom time for bettors and fantasy sports players, who are working out their game plans for the season ahead. Between September and January, they will have 272 games to navigate, building strategies for daily fantasy sports picks and NFL player props bets. For those truly serious about succeeding, data will undoubtedly shape their picks across the season.
However, we might ask, how many of them will be using AI to make their football predictions? Indeed, it will remain niche in that respect, but the number will grow from season to season as AI adoption strengthens. But is it viable to use, for example, ChatGPT to predict NFL games? Is there any discernible advantage? Let’s look at some of the arguments for and against.
AI Should Be Approached Skeptically
Let’s begin with a disclaimer: AI is not some prescient tool that can provide “eureka” insights into football. The jury is still out on whether it can – or ever will – surpass human insight into sports analysis. It can be helpful, even if only as a sounding board. Moreover, it will likely get better. But for now, remain healthily skeptical about using it.
It’s important to note that AI is only as beneficial as the data it accesses. So, for instance, if you ask ChatGPT to predict all over/under totals from the Week 1 games, you will receive a fairly basic answer if you have not provided specific data to work with. ChatGPT will search the web for news and updates to shape its answer, but it will only have a cursory look at some predictions and sports analysis that you can easily find.
If you look at, for instance, Reddit threads on AI and gambling, they settle on two main areas. First, they say you need to train AI with bespoke data (some players talk of scraping data from professional sports and betting data providers). That will require some advanced technical know-how that the average sports fan might lack. Or at the very least, it might require a paid subscription to an advanced sports data platform, which must be factored into your profit margins.
The second point of discussion that comes up on Reddit is the idea of prompting correctly. Knowing how to prompt is a common theme in AI use, but in this case, it’s all about getting the AI tool to stay on point. You might have to constantly repeat yourself and resubmit the data that you want to look at.
It Can Get You Ahead, But Limits Remain
The overarching question is whether it can help you get ahead. As mentioned, it’s up for debate. Even with masses of data, if you ask AI to predict the Super Bowl contenders, there won’t be a massive departure from what you see on sportsbooks and football websites. That is logical because sportsbooks use the same data and advanced algorithms you ask AI to analyze.
So, what then is the point? Well, what you are trying to find are anomalies. By that, we mean instances where the data says something different than the betting odds. It’s not always going to be completely obvious. Still, it may be able to pinpoint that a player was unlucky in not surpassing a specific number of passing yards the previous season or that a team might be slightly worse than their 14-3 winning record suggests. Sometimes humans can spot those anomalies, but AI might be better at it with the correct data and prompting.
We will finish by pointing out that nobody has mastered AI and sports predictions. We know this because it would cause mass panic across the sports industry. Moreover, human fantasy sports players are still beating the AI analysts at fantasy sports. This includes a 67-year-old British woman who topped the Fantasy Premier League table (11-4 million players) with a “pen-and-paper” approach. AI can and will be beneficial, but nothing out there has supplanted human intuition yet.
