Three years later, Bajaj cofounded and became CEO of Foresite Labs, an effort to incubate new AI and data science startups. He recently left that job to focus on Project Prometheus, according to the three people who spoke on condition of anonymity.
Project Prometheus is among a wave of companies focused on applying AI to physical tasks, including robotics, drug design and scientific discovery. This year several prominent researchers left Meta, OpenAI, Google DeepMind and other big AI projects to found a company called Periodic Labs, which is focused on building AItechnology that can accelerate new discoveries in areas like physics and chemistry.
Last year, Bezos invested in Physical Intelligence, a startup that is applying AI to robots.
But the $US6.2 billion in funding behind Project Prometheus potentially gives it an advantage in the expensive race to build AI technologies. This year, Thinking Machines Lab, founded by a group of former OpenAI employees, raised $US2 billion in funding.
Project Prometheus has already hired nearly 100 employees, including researchers poached from top AI companies such as OpenAI, DeepMind and Meta, the three people said.
A number of well-known AI companies — including OpenAI, Google and Meta — are already working on technologies meant to accelerate work in the physical sciences. Two researchers at Google DeepMind, the company’s primary AI lab, recently won a Nobel Prize in chemistry for their work on a project called AlphaFold, which can help accelerate drug discovery in small but important ways.
Executives at these companies and others in the field often say that large language models — the technologies that power chatbots like OpenAI’s ChatGPT — will soon achieve significant scientific breakthroughs. OpenAI and Meta say their technologies are already approaching this goal in areas like math and theoretical physics.
But companies like Periodic Labs and now Project Prometheus aim to build AI models that learn in more complex ways than chatbots do.
Large language models learn their skills by analysing massive amounts of digital text. By pinpointing patterns in Wikipedia articles, news stories and other information culled from across the internet, these systems learn to mimic the way people put words together. They can even learn to write computer programs and solve math problems.
Loading
The new companies are focusing on systems that can also learn from the physical world. Periodic Labs, which has $US300 million in backing, plans to build its own lab in Northern California where robots will run scientific experiments on an enormous scale. By analysing this physical trial and error, AI systems can learn to perform experiments largely on their own — at least in theory.
Project Prometheus will explore similar work, according to the people familiar with the company’s plans.
The New York Times
The Business Briefing newsletter delivers major stories, exclusive coverage and expert opinion. Sign up to get it every weekday morning.
