USA TODAY asked Microsoft Copilot to project the records of all 32 NFL teams, and the experiment showed how AI-generated forecasts can include outdated information and internal inconsistencies.

WHAT’S HAPPENING

USA TODAY Sports asked Microsoft Copilot to predict the 2026 regular-season records of all 32 NFL teams and provide a brief explanation for each projection.

During the process, some problems emerged.

Copilot initially relied on outdated information involving players and coaching changes. After those issues were identified, the projections were revised.

The final predictions also produced a combined league record of 297 wins and 247 losses, which exceeds the number of wins possible during the NFL’s 272-game regular season.

Despite those issues, Copilot still produced projections for every division and identified teams it considered likely contenders.

WHY IT MATTERS

Sports predictions are uncertain by nature, so incorrect forecasts are expected from humans and AI alike.

The more important issue is whether the information supporting those predictions is current and internally consistent.

This experiment shows that AI can quickly analyze teams and generate detailed forecasts, but the results may still require verification before they are treated as reliable.

WHO BENEFITS

Sports fans and analysts can use AI as another tool for exploring possible outcomes, comparing teams and generating discussion before a season begins.

Media organizations can also use AI-generated projections alongside traditional analysis to offer audiences another perspective.

AI developers benefit from real-world examples that reveal where models may need stronger access to current information and better consistency checks.

WHO LOSES

Users who rely on AI-generated sports analysis without checking the underlying information could make decisions based on outdated or inconsistent data.

Sports bettors may face additional risk if they treat conversational AI projections as equivalent to validated statistical models or professional betting analysis.

AI companies can also face credibility challenges when noticeable factual or mathematical inconsistencies appear in otherwise polished responses.

WHAT HAPPENS NEXT

AI-generated sports analysis is likely to become more common as models gain access to fresher information and improved reasoning capabilities.

The technology may become increasingly useful for comparing teams, identifying trends and generating possible scenarios.

For now, the USA TODAY experiment offers a useful reminder: AI can add another layer to sports analysis, but its predictions still benefit from human review and verification.