Key points from article :
The program amazed the world in 2017 when it mastered chess from scratch in 4 hours.
This is despite not being programmed how to win.
After a year of testing and analysis, programme is now improvising like a human.
Moves based on 'nebulous sense that it is all going to work out in the long run'.
Won 155 times, lost 6, the rest draw in 1000 games vs Stockfish (world’s best chess machine).
AlphaZero readily sacrificed its soldiers for a better position in the skirmish.
Saying it has subtle sense of intuition, helps balance out all different factors.
Intuition, creativity, accessible to machine intelligence as well, not only humans.
Reinforcement learning process like the way human brain learns.
AlphaZero could be a powerful teaching tool for the whole community.
Targets real world problems like protein folding leading to drug discovery.
Research by DeepMind, published in Science.