Former Openai and Deepmind researchers increase up to USD 300 million to automate learning

Periodic laboratories He left Stealth on Tuesday with a warrior of $ 300 million as a seed round, supported by the technology industry who’s who: Andreessen Horowitz, DST, Nvidia, Accel, Elad Gil, Jeff Dean, Eric Schmidt and Jeff Bezos.

The Period Laboratory was founded by Ekin Dogus Cubuk and Liam Fedus. Cubuk led a team of materials and chemistry at Google Brain and Deepmind, where one of his projects was, for example, the AI ​​tool called Gnome. This is a tool discovered over 2 million recent crystals in 2023Scientists say that one day materials that might be used to power recent generations of technology one day.

Fedus is a former vp for research at Openai and one of the researchers who helped create chatgpt. He also led the team he created The first trillion of the parameter neural network.

His small team is also stuffed with researchers who worked on other primary projects AI and Materials Science, from building the operator Openai to working on Mattergen, LLM Material Science Discovery AI.

The company claims that the goal of periodic laboratories is to automate scientific discoveries, creating scientists from AI. This means the construction of laboratories in which robots perform physical experiments, collect data, eaten and try again, learning and improving as they work.

The first goal of the laboratory is to come up with recent supercarries that hopes that they achieve higher results and may require less energy than existing superconducting materials. But a well -financed startup also hopes to find other recent materials.

Another goal is to collect all physical world data that AI scientists produce during mixing and heating, and in a different way manipulate various power and raw materials in search of something recent.

“Until now, AI scientific progress came from models trained on the Internet,” and LLM “exhausted” the web as a source that will be consumed, says the company in the blog introducing post on the blog. “At the period, we build AI scientists and autonomous laboratories so that they can work.”

We hope that laboratories won’t only come up with recent generation materials, but will produce invaluable recent data that AI models can devour to proceed evolution.

Although this may occasionally be one of the most impressive groups of researchers to meet the startup for this purpose, this is not the only work on AI scientists. AI as a tool for automation discovering chemistry is The topic of educational research from at least 2023. It is a pursuit of small startups, similar to Tetsuwan Scientific, and also no profit Future house and University of Toronto’s Acceleration consortium.