Kai-Fu Lee denies Zero2Everything's sale of assets to Aliyun, calls rumors “vicious vilification”

Kai-Fu Lee, the founder of Chinese startup Zero2One, recently strongly denied rumors that the company sold its pre-training team and infrastructure to Aliyun, calling the claims “vicious vilification.” In a Jan. 6 social media post, Kai-Fu Li said that Zero2One has recognized revenue of more than 100 million yuan in 2024 and expects to see multi-fold growth in 2025, and that the company's relationship with Aliyun is a “vicious smear campaign”. He emphasized that the company's cooperation with Aliyun is based on deep technology sharing and business collaboration, not an acquisition relationship. The two companies jointly announced on January 2 the establishment of the “Industrial Big Model Joint Laboratory”, which aims to promote the application of big model technology in real industries.


Zuckerberg Follows Musk's Lead, Ditches Fact-Checking, Introduces Community Notes


Meta CEO Mark Zuckerberg recently announced that the social media platform will no longer use third-party fact-checking organizations, but will instead introduce a “community notes” system similar to Musk's X platform. The decision is aimed at increasing user trust in content, with Zuckerberg saying that traditional fact-checking mechanisms are politically biased and erode user trust. The new system will allow users to collectively assess and provide contextual information to better identify potentially misleading content, a change Meta plans to implement first in the U.S. and gradually ease restrictions on sensitive topics such as immigration and gender in order to promote free expression.


Google Forms New Team to Develop 'World Modeling' AI for Gaming and Robotics Training


Google has announced that it will create an artificial intelligence team focused on “world modeling,” with the goal of developing AI models that can simulate the physical world. The team is led by Tim Brooks, former co-leader of OpenAI's video generation project Sora, who joined Google DeepMind last October.The new team will work with Google's existing Gemini, Veo, and Genie teams to solve key problems and increase the computational power of the models.Brooks said on social media that the models will be used for real-time interaction generation, supporting a variety of domains including visual reasoning, simulation and entertainment.

Comments

Popular posts from this blog

WordPress.org has launched a high-performance translation plug-in to solve performance problems caused by translation

The iPhone suddenly dropped in price, and the 14 Pro dropped by up to 800 yuan! Sales in Greater China have fallen sharply, is Apple in a hurry?

120x zoom! Mi 13 Ultra revealed to be launched in April