People will gamble on anything. Which horse can run around a dirt track fastest? What number will a little white ball fall on in a big spinning wheel in a Vegas casino? Who can eat the most hot dogs in 10 minutes?
Some of the biggest bets happen on Super Bowl Sunday, and no matter what the sports bookies say, two things are for certain: There will be elated winners and more than a few sore losers. While it’s impossible to predict the exact outcome of the game, AI systems might have a unique ability to parse information and accessible statistics, giving the average Joe a better grasp of what’s possible on game day.
CNET
Here’s how I used four AI chatbots — ChatGPT, Microsoft Copilot, Google Gemini and Claude AI — to try to predict the 2026 Super Bowl.
Guessing and gambling can be fun as an adjunct to sports, but if you’re doing it for money, please be responsible. Also, be very aware that generative AI makes mistakes and, like the outcomes of sporting events, can be unpredictable in gambling and other topics.
Read more: I Almost Won My March Madness Pool Thanks to ChatGPT’s Bracket Suggestions
The teams: Seattle Seahawks vs. New England Patriots
We’re deep into the latter end of the NFL season, with only four teams having made it to the conference championships this weekend, the final contest before the Super Bowl. Of the four AI systems I asked to generate predictions, all predicted the Seattle Seahawks would face the New England Patriots on Feb. 8.
All four chatbots also agreed that the only way this wouldn’t happen was if a key player on either team were suddenly badly injured.
The chatbots all got incredibly detailed in their predictions of when and what the pivotal moments in the game would be, who would have dominance in the first half, which players would have standout performances and what the memorable highlights of the game would be.
Don’t miss any of our unbiased tech content and lab-based reviews. Add CNET as a preferred Google source.
The Super Bowl LX winner predictions by Claude (left) and Copilot (right).
Claude/Copilot/Screenshot by CNET
The Super Bowl LX winner predictions by ChatGPT (left) and Gemini (right).
ChatGPT/Gemini/Screenshot by CNET
We’ll be able to check the AI systems’ accuracy after this weekend, when the New England Patriots play the Denver Broncos, and the Seattle Seahawks face off against the Los Angeles Rams, both on Sunday.
The coin toss: Seattle Seahawks
Three out of four of the AI systems I used predicted that the Seattle Seahawks would win the coin toss in the 2026 Super Bowl, choose tails over heads and give the ball to the New England Patriots to kick off the game. Gemini was the outlier, saying the Patriots would win the coin toss, choose tails and give the ball to the Seahawks.
Notably, ChatGPT and Microsoft Copilot both made guesses, while Google Gemini and Claude AI relied on historical Super Bowl data to predict the outcome of the coin toss.
The Super Bowl LX coin toss winner predictions by Copilot (top left), ChatGPT (top right), Gemini (bottom left) and Claude (bottom right).
Claude/Copilot/Gemini/ChatGPT/Screenshot by CNETThe outcomes: Seattle Seahawks win by 27 points
This is where the predictions started to get interesting: All four AI systems made the same predictions for who would win the 2026 Super Bowl and, what’s more, almost exactly the same predictions for the end score.
According to ChatGPT, Microsoft Copilot, Google Gemini and Claude AI, the Seattle Seahawks will win against the New England Patriots, either 27-23 or 27-24.
When asked to generate the reasoning behind these startlingly similar predictions, all the AI systems gave different answers.
The Super Bowl LX winner predictions by Copilot (top left), ChatGPT (top right), Gemini (bottom left) and Claude (bottom right).
Claude/Copilot/Gemini/ChatGPT/Screenshot by CNETThe halftime show: Cameos by Cardi B and J Balvin
Don’t bet on any unexpected rainfall or wardrobe malfunctions during the 2026 Super Bowl halftime show. All the AI systems queried about the ins and outs of the performance pretty much agreed with each other: Bad Bunny will bring his special brand of multilingual pop, hip-hop, Latin beats and lyrics to the show, in spite of political discourse around his history-making appearance.
Cameos by special guests like Cardi B and J Balvin were also commonly predicted, as well as a generally tame execution, missing the more socially and politically charged undertones of previous performances by Kendrick Lamar and Beyoncé.
The extras: National anthem, viewership, most expensive Super Bowl ad
It seems that great artificial minds think alike again. Every AI system we tested generated very similar predictions for the length of the national anthem performance (around 2 minutes), total viewership for the game (between 118 million and 130 million), and halftime show viewership (around 130 million).
The AI systems differed slightly in their predictions about which company would plonk down the most cash for a Super Bowl commercial in 2026, and were split between beer, gadgets, and AI itself.
Claude predicted that an AI company would spend the most, at $8.5 million per 30 seconds of screen time. Gemini predicted that Anheuser-Busch would be the biggest spender overall, with a commitment of 2.5 minutes of total airtime, but also predicted that OpenAI would spend the most money for screen time. ChatGPT predicted Apple would go all in on a Super Bowl ad and drop $14 million to $16 million on a 60-second spot, with Copilot in agreement.
Super Bowl betting: Total spend
Finally, the figures on Super Bowl betting were just as broad and open as the space between goal posts. Claude predicted $9.2 billion would be spent on legal betting around the 2026 Super Bowl, while Gemini predicted a whopping $25 billion in bets.
ChatGPT and Copilot had to be prompted to speculate on the mounds of money in play, but both gave much lower estimates in the range of $2.5 billion to $6 billion. That can buy a lot of beer and wings.
(Disclosure: Ziff Davis, CNET’s parent company, in April filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.)