One of the researchers behind AlphaGo is seeking tens of millions of dollars from a small initial group of backers for a new startup called Metis Reasoning, Bloomberg reported Thursday (Sept. 24). AlphaGo is the Google DeepMind program that beat a world champion at the board game Go in 2016 by learning from millions of games against itself and planning moves ahead. Thore Graepel left Alphabet’s research lab this summer, and the talks are preliminary, people familiar with the matter told Bloomberg.

Metis Reasoning is working on artificial intelligence that can respond to unfamiliar problems and choose an action, with uses in robotics, science and engineering, according to its website as cited by Bloomberg. Graepel’s personal site github describes the project as bringing AlphaGo-style reasoning to frontier AI so machines can plan and act under uncertainty.

A large language model predicts the next word from human-written text. The systems Metis Reasoning and its peers are chasing learn from their own actions and weigh what to do before doing it, according to the report.

David Silver, who led DeepMind’s reinforcement learning team, raised $1.1 billion at a $5.1 billion valuation in April for Ineffable Intelligence, TechCrunch reported. Reinforcement learning is a method in which AI improves through trial and error rather than by studying human examples. Sequoia and Lightspeed co-led the round, with Nvidia and Google among the participants, CNBC noted.

5 billion pre-money valuation in March to build world models, according to TechCrunch. A world model predicts how an environment will respond to an action, letting a robot compare options before it moves, according to an Aug.

Source link