Sou Hu Cai Jing·2025-06-09 12:15Group 1 - The current stage of AI development is referred to as "AJI" (Artificial Jagged Intelligence), indicating a non-linear progression characterized by both significant advancements and fundamental errors [3][5][7] - AI models are capable of solving complex problems but often fail at basic tasks, highlighting the unpredictable nature of AI's growth compared to human development [5][7] - Sundar Pichai predicts that by 2030, there will be significant advancements in AI across multiple dimensions, necessitating the establishment of an AI content identification system to differentiate reality [7][8] Group 2 - Pichai emphasizes that the evaluation of AI models should not solely focus on their ability to tackle complex challenges but also on their performance in basic logical checks and common-sense judgments [7][8] - The ability to avoid basic errors is considered a safety baseline for AI, as frequent common-sense mistakes can undermine user trust and decision-making [8]