AI model Grok has outperformed human participants in prediction markets, achieving a 75% success rate compared to humans' 66.7% and Gemini's 52.4%. The study, conducted by Nan Zhi, tested AI's ability to predict market outcomes against human predictions, excluding the crypto sector. Grok's success highlights the potential of AI in leveraging vast knowledge bases to make accurate predictions. Despite Grok's superior performance, the study identified areas for improvement in AI predictive logic. Issues such as time misjudgment and reliance on common-sense assumptions rather than evidence were noted, particularly in Gemini's predictions. These findings suggest that while AI can surpass human accuracy, refining AI's reasoning processes could further enhance its predictive capabilities.