Tetsuwan Scientific, a San Francisco-based firm, is constructing artificial intelligence (AI) robotics that can do the activities of a scientist. The co-founders, CEO Cristian Ponce and the Chief Technology Officer (CTO) Théo Schäfer, launched the firm out of stealth in November following a successful seed round fundraising. The company's goal is to create intelligent software that can be combined with lab robots to automate the full scientific discovery and innovation process, from hypothesis generation to experimentation and conclusion drawing.
Developing AI-Powered Robotics Scientists
Founded in 2023, the firm has been working under stealth for the past year and a half to develop its first product, an AI scientist capable of doing tests. It has emerged from stealth and is now collaborating with La Jolla Labs on the development of RNA medicinal drugs. On its website, the firm describes its vision and the first product it is developing. Notably, none of its products are now in the public domain.
The firm emphasizes the problem statement it intends to answer, claiming that automation in research is focused on a large number of trials rather than a wide variety. This is because lab robots now require substantial programming to mimic certain processes. However, this has resulted in the development of a system that produces assembly lines rather than robots to help scientists, according to the business.
Tetsuwan Scientific remarked that the difficulty is that robots cannot grasp scientific purpose and hence cannot conduct experiments on their own. However, using generative AI models, the business claims it is now able to bridge this communication gap and educate robots how to behave like scientists. It is a two-pronged challenge that necessitates clever software and adaptable robotics gear.
In an interview with TechCrunch, Ponce stated that large language models (LLMs) can bridge the software divide by enabling developers to transmit scientific purpose to a robot without having to write thousands of lines of code. The CEO emphasized that the Retrieval-Augmented Generation (RAG) methodology can also aid in reducing AI delusion.
According to the report, Tetsuwan Scientific is developing nonhumanoid robots. These robots, which are also shown on the website, are big square-shaped glass-like structures that are believed to review data and modify scientific research without requiring human participation. These robots are believed to use AI software and sensors to learn about technical standards including calibration, liquid class characterization, and other features.
Notably, the business is now in the early phases of developing autonomous robotic AI scientists capable of automating the entire scientific process and inventing things.