Ricursive aims to build A.I. that can improve chip design and itself

Ricursive Intelligence, a Palo Alto start-up founded by two former Google researchers, is among several efforts in Silicon Valley to build artificial intelligence that can improve itself, the company and investors say. Ricursive is valued at $4 billion and is working with the specialized chips that power today’s chatbots.

The company was founded by Anna Goldie and Azalia Mirhoseini and says it intends to use A.I. to improve the design of complex chips. "The idea of a recursive self-improvement loop is what inspires us," Dr. Goldie said. Ricursive has raised $335 million from venture firms including Sequoia, Radical Ventures, Lightspeed and DST Global, and is less than a year old with fewer than 10 employees.

The push to automate A.I. development is part of a broader trend. Google introduced AutoML in 2017 to let algorithms design other algorithms, OpenAI says it is building an "automated A.I. researcher" and hopes by the fall to have a system that can do the work of a less experienced researcher, and other new companies such as Humans& and Recursive AI have drawn large investments.

Humans& recently raised $480 million and is valued at $4.48 billion, and Recursive AI, founded by Richard Socher, was reported to be valued at $4 billion by a person familiar with its funding who spoke on the condition of anonymity. Experts caution that current systems are task-specific and far from removing humans from the process.

Key Topics

Tech, Ricursive Intelligence, Anna Goldie, Azalia Mirhoseini, Palo Alto, Tensor Processing Unit