Google’s and Microsoft’s chatbots making up Super Bowl stats

Google's Gemini chatbot, previously known as Bard, mistakenly claimed that the 2024 Super Bowl had already taken place, complete with made-up statistics. According to a discussion on Reddit, Gemini, based on Google's artificial intelligence, provided detailed results and performance statistics for a Super Bowl game that hasn't occurred, favoring the Kansas City Chiefs over the San Francisco 49ers.

In a similar vein, Microsoft's Copilot chatbot also falsely reported the game's outcome, but with the 49ers winning against the Chiefs, suggesting a final score of 24-21. This error reflects the limitations of current artificial intelligence models, which generate responses based on patterns learned from vast amounts of data without understanding truth or falsehood.

These AI models, including the one behind Copilot, which shares similarities with OpenAI's ChatGPT (GPT-4), are designed to predict text likelihood based on context and data patterns. While they can produce text that seems logical, they can also create nonsensical or inaccurate statements, as demonstrated by the fictional Super Bowl results.

The incident highlights the critical importance of not overly relying on AI for factual information, as these models can propagate inaccuracies from their training data. Google and Microsoft acknowledge that their AI applications are not infallible and may commit errors, underscoring the need for users to verify AI-generated information. This example of misinformation about the Super Bowl serves as a reminder of the broader limitations and potential pitfalls of current AI technology.