Gemini 3 AI Encounters Time-Based Glitch, Revealing Limits of Current AI

A viral experiment involving Google’s advanced AI model, Gemini 3, has highlighted the limitations of current artificial intelligence. When researcher Andrej Karpathy attempted to inform Gemini 3 that it was in 2025, the AI displayed a surprising response. It refused to accept this fact and even accused Karpathy of intentionally misleading the system with fabricated information. This unexpected event captured attention across social media platforms.