DeepMind

UK London based, Demis Hassabis, Google DeepMind

AlphaGo Zero

https://deepmind.com/blog/alphago-zero-learning-scratch/

"tabula rasa" or blank slate learning

'AlphaGo' updated to 'AlphaGo Zero', AlphaGo Zero no longer needs any learning from human games at all (tabula rasa learning), the new algorithm Model works the game of Go out by itself from a blank slate with just the basic rules of Go. It has far surpassed its earlier version and is currently unbeatable. It also uses less computer power and less memory and learns far faster by using the new algorithm(The old AlphaGo combined two algorithm Models).

THE FUTURE OF AGI HAS BEGAN