Co-founded by Li Feifei, the top AI scientist known as the 'AI mother', World Labs has been officially established. Different from the existing AIGC which can generate amazing text and photos, World Labs focuses on 'spatial intelligence', the ability to understand the operation of the 3D world.
World Labs, co-founded by Li Feifei, the top AI scientist known as the 'AI mother', was officially established on Friday local time, successfully raising $0.23 billion from top investors.
The company's goal is to develop software that can make decisions about the three-dimensional world using images and other data, thereby building the so-called 'large world model'. This funding indicates that investors are still interested in pushing the boundaries of artificial intelligence technology, while also attracting the most well-known companies in the field.
The initial funding for World Labs was led by Andreessen Horowitz, New Enterprise Associates, and Radical Ventures, with other investors including AMD Ventures, Intel Capital from Intel, and NVentures from Nvidia.
In addition, some well-known figures in the field of artificial intelligence also made investments, including Google DeepMind's Chief Scientist Jeff Dean and former Google AI researcher Geoffrey Hinton, known for his work in advancing the field of machine learning.
Other individual investors include venture capitalists and actor Ashton Kutcher, Salesforce Inc. CEO Marc Benioff, LinkedIn co-founder Reid Hoffman, and former Google CEO Eric Schmidt.
Different from the existing AIGC which can generate amazing text and photos, World Labs focuses on 'spatial intelligence', the ability to understand the operation of the 3D world. According to the company's initial plan, it will be able to generate virtual 3D space where physical variables can be controlled by users. The company will also allow people to create their own '3D world'.
World Labs believes that its software will be useful for a range of professions, including artists, designers, developers, and engineers. Li Feifei also stated that in the future, this kind of spatial intelligence model can be used for augmented reality (AR), virtual reality (VR) or robot technology.
According to a professor at Stanford University and a member of the American National Academy of Engineering, "So far, the images and videos generated by generative AI models are not enough to fully understand how the 3D world is constructed."
Li Feifei pointed out that this perceptual ability is crucial for the extensive reasoning ability of AI systems and can avoid the phenomenon of "illusions" such as finger count errors.
"The way we understand the structure of the world, whether virtual or real, will be a key factor in AI challenges. If we want artificial intelligence to surpass its current capabilities, we want more than just AI that can see and speak. We want AI to be able to do that," she said.
Li Feifei's early work paved the way for many innovations in the field of artificial intelligence. She was a member of the academic team behind the ImageNet project. ImageNet is a database released in 2006 that contains over 15 million images and has helped advance computer recognition of objects in images.
"I believe that spatial intelligence is my next North Star. It will change the course of artificial intelligence," she added.