MIT's robotics luminary, Daniela Rus, co-founds Liquid AI, a startup aiming to develop general-purpose AI systems utilizing liquid neural networks. The company, emerging from stealth, announces a significant $37.5 million raised in a two-stage seed round from notable investors, valuing Liquid AI at $303 million post-money. Contributors include VCs like OSS Capital, PagsGroup, Automattic (WordPress parent company), Samsung Next, Bold Capital Partners, and angel investors such as GitHub's Tom Preston Werner, Shopify's Tobias Lütke, and Red Hat's Bob Young.

Liquid AI's core technology, liquid neural networks, was introduced in a research paper titled "Liquid Time-constant Networks" in 2020. The founding team comprises Ramin Hasani (CEO), Mathias Lechner (CTO), and Alexander Amini (Chief Scientific Officer), all with significant AI research backgrounds at MIT. Liquid neural networks, inspired by roundworm brains, offer smaller size and reduced computational requirements compared to traditional AI models.

Liquid neural networks exhibit uniqueness in their adaptability and interpretability. With fewer parameters and neurons, they require less compute power and offer greater interpretability. Additionally, these networks dynamically adjust their parameters over time, making them adept at handling shifts in surroundings, as demonstrated in various applications from predicting atmospheric chemistry to autonomous drone navigation.

Liquid AI's liquid neural network architecture shows promise in applications like drone search and rescue, wildlife monitoring, and analyzing phenomena that fluctuate over time, such as financial transactions or severe weather patterns. The startup aims for commercialization, positioning itself against foundation model companies like OpenAI. The recently secured seed funding will facilitate the development of new Liquid foundation models beyond existing generative models like GPT-4.

Looking ahead, Liquid AI plans to expand its team from 12 to 20 members by early next year. The company envisions continuing advancements in the liquid neural network architecture, providing on-premises and private AI infrastructure, and offering a platform for customers to build their own models. Liquid AI emphasizes the importance of accountability, safety, and efficiency in large AI models, positioning itself as a key player in the evolving landscape of AI technology.