At the beginning of 2024, Sora's debut threw a heavy bomb into the AI world. This cultural video model, released by the American artificial intelligence company OpenAI, only requires a prompt text to generate high-definition videos with multiple characters and specific action types, and the theme and background are basically accurate. Compared to AI video generation applications such as Runway Gen 2 and Pika, Sora can generate continuous, stable, and high-quality videos for up to 60 seconds with a few seconds of coherence. The more comprehensive and precise the prompt text and details, the more realistic the generated videos will be. However, due to concerns about potential abuse, OpenAI stated that it has not yet publicly released Sora's plans. The limited access rights to the model are only granted to a small group of researchers and creatives, in order for OpenAI to obtain their usage feedback. Currently, 48 demonstration videos generated by Sora have been updated on the official website. The clear and realistic details and ultra-high accuracy of these videos inevitably prompt people to think: does this mean the arrival of General Artificial Intelligence (AGI) with human intelligence equivalent or surpassing human intelligence? After the release of Sora, the founder of 360 Group, Zhou Hongyi expressed his opinion that the emergence of Sora has advanced the arrival time of AGI. Originally estimated to take about ten years, now it may only take two to three years. He believes that although Sora may seem like just a cultural video tool, it is actually a milestone for AI to understand and interact with the world, which will bring tremendous progress to the entire industry. "There are various technical routes to realize AGI, involving different research methods and application directions." Wang Jinqiao, Deputy Chief Engineer of the Institute of Automation of the Chinese Academy of Sciences and Executive Deputy Director of the Zidong Taichu Large Model Center, introduced to the reporter of Science and Technology Daily that at present, there are three AGI technical routes widely discussed by the academic and industrial circles. One is information intelligence, which means "big data+self supervised learning+big computing power". This method relies on a large amount of data, trains the model through self supervised learning algorithms, and requires enormous computational power to handle complex tasks. The second is game intelligence. This technological approach emphasizes training intelligent agents through reinforcement learning in human-computer interaction, enabling them to engage in autonomous learning and decision-making. The third is neuromorphic intelligence. This method attempts to achieve AGI by imitating the operation of the human brain. In Wang Jinqiao's view, according to the demonstration video on the official website, Sora has achieved breakthroughs in image quality, long video generation, multi shot consistency, learning world rules, and multimodal fusion. "Sora has caused such a sensation not only because it generates longer and clearer videos, but also because it can simulate the motion and interaction of objects in the physical world to a certain extent." Wang Jinqiao said, "This ability is of great significance for AGI research because it involves deep understanding and high simulation of the real world by machines, which are the core challenges of achieving AGI." The reporter learned that in order to accurately simulate the physical world, Sora was fed with a large amount of training data and used advanced algorithms such as diffusion models. ". “