The former Openai and DeepMind researchers collect a huge seed of $ 300 million to automate science

Periodic laboratories It came out of the ghost on Tuesday with the war chest of $ 300 million as a seed tour, with the support of the technology industry from: Andrien Horwitz, Dest, Nafidia, Aseel, Jill, Jeff Dean, Eric Schmidt, and Jeff Bezos.

The period laboratories were established by Ekin Dogus Cubuk and Liam Fedus. CUBUK led the Google Brain and DeepMind material team, for example, for example, Amnesty International called GNome. That tool Discover more than 2 million new crystals in 2023The researchers say the materials that can one day be used to operate new generations of technology.

FEDUS is a former vice president for research in Openai, and one of the researchers who helped create Chatgpt. He also led the team that was established The first trillion parameter nervous network.

Its small team is filled with researchers who worked on other projects of artificial intelligence and materials, from Building Openai to work on Microsoft’s Mountgen, A Llm Mateials Science Ai.

The company says that the goal of periodic laboratories is no less than the automation of scientific discovery, and the creation of artificial intelligence scientists. This means building laboratories where robots perform material experiments, collect data, repetition, try again, learn and improve with their progress.

The first goal of the laboratory is to create new superior connectors that it hopes to perform better and may require less energy than highly connected materials. But well -funded emerging also hopes to find other new materials.

Another goal is to collect all the data of the material world produced by artificial intelligence scientists while mixing and heating them and dealing with various forces and raw materials in their search for something new.

“Until now, AI’s scientific progress came from trained models on the Internet” and LLMS “has exhausted” the Internet as a source of consumption, as the company says in a preliminary blog post. “In the patrol, we build artificial intelligence scientists and independent laboratories for them to work.”

Hope is that the laboratories will not only invent the following generation materials, but will also produce invaluable new data that can be consumed by artificial intelligence models to continue their development.

Although this may be one of the most impressive researchers groups to assemble a startup for this purpose, it is not the only one that works on artificial intelligence scientists. AI as a tool to automate chemistry discovers The subject of academic research since 2023. It is the pursuit of small startups like Tetsuwan ScientificAs well as not profits like Future home And the University of Toronto The acceleration union.