Siri's AI Upgrade Falters: Massive Errors in Super Bowl Trivia Test

Siri's Dismal Performance in Super Bowl Trivia

Apple's virtual assistant Siri, despite recent AI enhancements, has come under fire for its poor performance in a simple Super Bowl trivia test. Paul Kafasis of One Foot Tsunami conducted a comprehensive experiment, asking Siri to identify the winners of Super Bowls I through LX. The results were strikingly poor, with Siri correctly identifying winners only 34% of the time - just 20 correct answers out of 58 played Super Bowls 1

1

Embarrassing Errors and Inconsistencies

Perhaps the most glaring error was Siri's repeated and incorrect attribution of 33 Super Bowl victories to the Philadelphia Eagles, despite the team having won only one championship in their history. The virtual assistant's responses ranged from providing information about wrong Super Bowls to offering completely unrelated football facts 2

2

In one particularly damning streak, Siri missed 15 consecutive Super Bowl winners from Super Bowl XVII through XXXII. This level of inaccuracy raises serious questions about the reliability of Siri's knowledge base, especially concerning popular and easily verifiable information 3

3

Comparison with Other AI Assistants

John Gruber of Daring Fireball conducted a comparative analysis, testing Siri against other AI assistants and search engines such as ChatGPT, Kagi, DuckDuckGo, and Google. All of these alternatives fared significantly better than Siri when asked similar Super Bowl trivia questions. They even demonstrated the ability to handle questions about future Super Bowls that haven't occurred yet, providing appropriate responses 1

1

Related Stories

Apple's Siri Struggles with Basic Queries, Raising Concerns About AI Progress

Apple's AI Struggles Continue: Concerns Grow Over New Siri's Performance in iOS 26.4

Apple's Siri Overhaul: Internal Struggles and Leadership Changes Revealed

Apple launches dedicated Siri app, marking the assistant's biggest transformation yet

Regression in Functionality

Perhaps most concerning is the observation that the new AI-enhanced Siri performs worse than its predecessor in some aspects. The old version of Siri would acknowledge its limitations on certain queries and provide relevant web links. In contrast, the new Siri, powered by Apple Intelligence with ChatGPT integration, often provides confidently incorrect answers - a characteristic of unrefined AI systems 2

2

Implications for Apple's AI Strategy

This poor performance comes at a crucial time for Apple, as the company is reportedly developing a much smarter version of Siri utilizing advanced large language models. The goal is to better compete with chatbots like ChatGPT or Claude. However, the current integration issues raise concerns about the effectiveness of future implementations 1

1

Apple is expected to announce an LLM-powered Siri as soon as 2025 at WWDC, with a planned launch in spring 2026 as part of iOS 19. However, the current state of Siri's performance suggests that significant improvements are needed before such a launch can be successful 1

1