A recent investigation has exposed major accuracy issues with Apple's newly ChatGPT-enhanced Siri, particularly when handling basic sports trivia questions.
Tech blogger Paul Kafasis conducted a comprehensive test, asking Siri to name the winner of each Super Bowl from I through LX. The results were startling - Apple's AI assistant correctly identified only 20 winners out of 58 Super Bowls played, achieving a mere 34% accuracy rate.
In one notable blunder, Siri claimed the Philadelphia Eagles had won an impossible 33 Super Bowl championships. The team has actually won just one title in franchise history. The AI assistant also failed to correctly identify winners for 15 consecutive Super Bowls between Super Bowl XVII and XXXII.
The performance gap becomes even more apparent when compared to other AI platforms. Testing by tech journalist John Gruber revealed that competing services from Google, Anthropic, and DuckDuckGo handled similar Super Bowl questions with significantly higher accuracy.
The issue extends beyond just Super Bowl history. When asked about the 2004 North Dakota high school boys' state basketball championship, standalone ChatGPT provided the correct answer with supporting details, while Siri failed the query completely.
What makes these errors particularly concerning is that the previous version of Siri would typically acknowledge its limitations and direct users to web resources. The new ChatGPT-integrated version instead provides incorrect information with high confidence - a characteristic problem of unrefined AI systems.
Industry observers note that this represents a step backward for Apple's digital assistant. As Gruber puts it, the current ChatGPT-enhanced Siri shows "a massive regression" in handling straightforward factual queries compared to its predecessor.