Zhou Hongyi: China's AI Can Catch Up to Sora within One or Two Years

In anticipation of the upcoming 2024 National People's Congress, Zhou Hongyi, a prominent member of the Chinese People's Political Consultative Conference and founder of 360, once again shared his unique insights on the hot topic of artificial intelligence (AI). Despite widespread fears and concerns about AI leading to mass unemployment, Zhou Hongyi believes that AI is humanity's most valuable friend.

Recently, during an interview at the 360 headquarters in Beijing, Zhou Hongyi emphasized that AI is not a destroyer, but an innovator. He stressed that AI will inject new vitality into industries such as short videos, film and television, and advertising, becoming a significant force driving industry transformation. Companies and individuals who refuse to embrace AI may face the risk of being left behind by the times.

In regards to the controversial "AI online celebrity" Li Yizhou, Zhou Hongyi criticized the high fees charged for his AI courses and the lack of public education spirit. He announced his plan to launch free AI courses, using simple and understandable language to popularize AI knowledge among the public and benefit more people from AI technology.

Zhou Hongyi highly praised OpenAI's recent release of the text-to-video model Sora. He believes that Sora's emergence has provided valuable inspiration to the Chinese AI community, making us more aware of the gap between China and the United States in AI technology. However, he firmly believes that this gap mainly lies in the direction of development. As long as domestic companies can find the right direction, they have the ability to catch up quickly.

Regarding the cooling phenomenon in the large-scale model industry, Zhou Hongyi believes that large-scale models have never experienced a downturn. He stated that although public opinion may fluctuate, entrepreneurs' enthusiasm for AI technology remains high. He believes that China's speed in catching up with AI technology is astonishing, especially in the field of general large-scale models. He predicts that 2024 will be the year of large-scale model applications, and large-scale models in vertical fields will become a new development direction.

Addressing the issue of AI security, Zhou Hongyi emphasized the importance of trustworthiness, controllability, and reliability of large-scale models. He pointed out that large-scale models may create "illusions," which are a manifestation of their creativity but may also lead to inaccurate results. In addition, large-scale models may face the risk of injection attacks and fraud caused by "deepfakes." He believes that to address these challenges, large-scale models need to have anti-PUA capabilities and strengthen content supervision to prevent the generation of false information.

Regarding the future prospects of AI, Zhou Hongyi remains optimistic. Although the entire large-scale model track has not yet become profitable, he believes that the dividend period will last for at least another ten years. He encourages entrepreneurs to continue investing in research and development, believing in the enormous development potential of China in the field of AI.

In addition, Zhou Hongyi expressed his views on the relationship between AI and new quality productivity. He believes that new quality productivity is a hot topic in today's society, and AI technology is the key driving force behind it. He emphasized that companies should establish a firm belief in AI and develop towards the direction of "All in AI."

So, how should companies proceed? Zhou Hongyi proposed three key steps.

First, companies should deepen their understanding of AI technology before rushing to integrate it with their own business. He revealed that 360 is assisting companies in the internal deployment of general large-scale models to solve basic office problems, allowing employees to become more familiar with and accustomed to applying AI technology.

Second, before customizing AI, companies should upgrade their big data platforms to knowledge platforms. This is because training AI requires rich knowledge resources. However, much of the critical knowledge belongs to "tacit knowledge," which is difficult to find on the Internet but crucial for companies. This knowledge is scattered in employees' computers, emails, and chat records. Therefore, companies need to effectively consolidate this "tacit knowledge."

Finally, companies should empower themselves with small models trained for specific scenarios. Zhou Hongyi pointed out that the core of general large-scale models lies in algorithms, models, and computing power. Nowadays, algorithms are open, models can be obtained for free, and enterprise-level models have relatively low computing power requirements. Therefore, companies should pay more attention to the integration of scene knowledge and business, using large-scale models to drive the transformation and upgrading of various industries. This is also the key to creating new quality productivity.

When discussing the current AI hype, Zhou Hongyi compared it to the previous metaverse craze. He believes that the rise of the metaverse relied mainly on concept speculation and lacked solid technological support. However, now three technologies may have a significant impact on the metaverse: the rise of headsets, AI technology producing a large amount of 3D content within the metaverse, and AI conducting 3D modeling and rendering within the metaverse. Zhou Hongyi predicts that the metaverse concept may resurge in 2025, but its foundation will still be the development of AI technology.

In addition, Zhou Hongyi discussed the phenomenon of internet celebrity economy. He pointed out that currently, 800 million people in China no longer read text but prefer to watch short videos. He used the well-known anchor Dong Yuhui as an example, believing that he successfully converted traffic into income by combining a business model. He emphasized that companies should not limit themselves to the narrow field of live-streaming sales but should grasp the rules of dialogue in the era of short videos and endorse every product. By fully utilizing AI technology, companies can create more attractive content and experiences, attracting more users and enhancing brand influence.