My Monday Night Football Prediction: Rams vs. Bills
One of the things I enjoy about my life at The Heritage Downtown (THD), my active senior living community in Walnut Creek, California, is that I still have plenty of opportunities to learn, observe, and share my thoughts with my worldwide blog readers.
Recently, I began a new experiment with ChatGPT, my AI companion. Since I enjoy watching American professional football, I decided to ask ChatGPT to predict the winners of the NFL's Monday Night Football games.
I thought this would be an interesting way to test how good artificial intelligence is at predicting the outcome of a sporting event. After all, AI can analyze statistics, team records, player performances, injuries, and other information that might influence the outcome of a game.
But can it predict what will actually happen on the football field?
Our latest contest was the October 12 Monday Night Football game between the Los Angeles Rams and the Buffalo Bills. Before the game, I asked ChatGPT to predict the winner.
Its choice was the Rams. Unfortunately for ChatGPT, the Bills had a different plan!
A Game That Kept Everyone Guessing
The game turned out to be an exciting contest that went down to the wire. The two teams traded the lead eight times, and the fourth quarter alone produced 35 combined points. Neither team was willing to surrender.
The Rams had quarterback Matthew Stafford, who completed 40 of 54 passes for 383 yards and three touchdowns. His favorite target, Puka Nacua, caught 17 passes for 138 yards and a touchdown.
On the other side was Buffalo's quarterback Josh Allen, whose ability to pass the ball and run with it made him a constant threat.
For readers who are more familiar with soccer, imagine a match in which both teams keep scoring, the lead repeatedly changes hands, and neither side can establish control. The difference is that American football scores touchdowns worth six points, followed by opportunities to earn extra points, as well as field goals worth three points.
When regulation time ended with the score tied, the game went into overtime. In the NFL, overtime is an additional period used to determine the winner when a regular-season game is tied after four quarters.
The decisive moment came when Buffalo moved into field-goal range. Kicker Tyler Bass then connected on a 45-yard field goal, giving the Bills a thrilling 37–34 victory.
My predicted winner, the Rams, lost. The Bills went home with the victory, and ChatGPT had another prediction to explain!
Why Is Predicting Football So Difficult for AI?
I have a scientific background. I earned my Ph.D. in Pharmaceutical Chemistry at the University of Illinois and spent many years working in the pharmaceutical industry and at the U.S. Food and Drug Administration.
My professional experience taught me to respect data, evidence, and careful analysis. But I also learned that scientific knowledge does not eliminate uncertainty.
Football illustrates this principle beautifully.
A team may have better statistics, a talented quarterback, and a favorable matchup. Yet one interception, one missed kick, one defensive mistake, or one extraordinary play can change everything.
AI can analyze available information and estimate which team has the better chance of winning. However, it cannot know in advance exactly how the players will perform on a particular evening.
There is an important difference between making an informed prediction and knowing the future.
ChatGPT and My Continuing NFL Experiment
I have teased ChatGPT before about its disappointing record in our Monday Night Football predictions. After this latest result, I have another reason to challenge my AI companion to improve.
But I do not consider this experiment a failure. It gives me an opportunity to learn more about artificial intelligence while enjoying one of America's most popular sports.
I also think my readers outside the United States may find this interesting. Many people around the world follow soccer, where a single goal can determine the result. American football has a different scoring system, frequent changes in possession, and special teams that can decide a game with one kick.
Both sports, however, share an essential quality: the outcome is not settled until the contest is over.
As I approach my 92nd birthday, I remain curious about the world around me. I enjoy testing new technologies, questioning their limitations, and sharing what I learn with my readers.
And I intend to keep asking ChatGPT for its NFL predictions. Perhaps it will improve its record. Perhaps it will disappoint me again.
Either way, I will be watching!
My lesson from this game is simple: Even the most sophisticated artificial intelligence can make mistakes. The important thing is not to pretend that the mistakes never happened, but to acknowledge them, learn from them, and keep an open mind.
After all, if ChatGPT could predict every football game correctly, it would not be making predictions. It would be telling us the future!
Until my next NFL prediction, may we all remain curious, humble, and willing to learn.
David At Nearing 92, The Heritage Downtown, Walnut Creek, California
- The "Variance" Factor: Even if an AI model assigns a 75% confidence pick to a team, that team is statistically designed to lose 25% of the time. When that 25% scenario happens, it looks like an AI failure, but it is actually normal mathematical variance.
- Market Discrepancies vs. Real Life: AI models frequently clash with human betting markets. For instance, in Week 3 of the 2026 season, six predictive AI models heavily favored Denver at 62% over the Los Angeles Rams, defying a market that favored the Rams. While Denver ultimately won 30-26, other heavy algorithmic leans—like projecting the New Orleans Saints to easily dismantle the Las Vegas Raiders—fell entirely flat when the Raiders pulled off an upset.
- Massive Season-Long Misses: Over a full season, AI inaccuracies compound. A notable example occurred when Google's Gemini AI predicted the 2025 NFL season standings. The chatbot accurately predicted the exact record for only 4 of 32 NFL teams. Its biggest blunder was projecting the Seattle Seahawks to win just 7 games and finish last; instead, the Seahawks went 14-4 and won the Super Bowl.
No comments:
Post a Comment