Alibaba leads $290m investment for Shengshu Vidu AI world model

BEIJING — Alibaba’s The cloud is investing in a new type of AI designed to better replicate the real world, using a different approach than chatbots like OpenAI’s ChatGPT.

Shift recognizes: Limits of “big language models” First of all, I received training on the text. Instead, developers are starting to focus more on: “world models” is built on videos and real-life physical scenarios.

To keep up with this trend, Alibaba invested 2 billion yuan ($290 million) in ShengShu, the startup behind AI video creation tool Vidu, the company announced Friday. TAL Education and Baidu Ventures also participated in the series B financing round.

The investment comes about two months after ShengShu raised investment 600 million yuan From Qiming Venture Partners and other backers. The startup declined to disclose its valuation.

ShengShu said the latest funding will support the development of a “generic world model” that uses AI to bridge two currently separate domains: the digital world of games and AI-generated videos, and the physical world of autonomous driving and robots.

“ShengShu believes that a general model of the world built on multimodal data such as sight, sound, and touch more naturally captures how the physical world works than large language models,” the three-year-old startup said in a statement. he said.

“We aim to combine perception and action,” ShengShu founder Zhu Jun said in a statement, allowing artificial intelligence systems to better model and predict real-world behavior in a consistent manner.

ShengShu’s latest Vidu Q3 Pro model, released in January, is among the top 10 AI models for creating video from text and images. Artificial Analysis.

The company launched Vidu globally, months before OpenAI made its now-shuttered Sora tool for AI video generation widely available. Chinese short video companies kuaishou and ByteDance have also released similar competing AI tools for creating video.

Chinese tech giant and Baidu Ventures became leaders last month 50 million dollar investment On Tripo AI, a platform that uses artificial intelligence to quickly create digital 3D models from photos. Tripo also said that he is moving from the techniques used by language models to artificial intelligence tools based on physical space and developing his own world model.

In September, Alibaba also led a $60 million investment in PixVerse, an A.I. world model A feature released earlier this year that allows users to manage how a video opens while it’s being created.

E-commerce startup Alibaba has also released free, open-source AI models for video production and in February launched a model to power robots.

Shengshu said Friday that it has strategic partnerships with companies developing embodied artificial intelligence (systems such as humanoid robots that interact with the physical world) for use in industrial, commercial and home environments.

Kevin Kelly, co-founder of US technology magazine Wired, wrote that world models are critical to robotics because more than a master’s degree is needed to make the technology work. last month In your substack.

Ultimately, to replicate human intelligence, AI will need three things: reasoning, understanding the physical world, and continuous learning, Kelly said. While artificial intelligence has not yet been developed for the learning category, Masters-backed chatbots create the element of knowledge and make world models an important area requiring breakthrough, he said.